Trusting Nudges

Many “nudges” aim to make life simpler, safer, or easier for people to navigate,
but what do members of the public really think about these policies? Drawing on
surveys from numerous nations around the world, Sunstein and Reisch explore
whether citizens approve of nudge policies. Their most important finding is sim-
ple and striking. In diverse countries, both democratic and nondemocratic, strong
majorities approve of nudges designed to promote health, safety, and environmen-
tal protection—and their approval cuts across political divisions.
In recent years, many governments have implemented behaviorally informed
policies, focusing on nudges—understood as interventions that preserve freedom of
choice, but that also steer people in certain directions. In some circles, nudges have
become controversial, with questions raised about whether they amount to forms of
manipulation. This fascinating book carefully considers these criticisms and answers
important questions. What do citizens actually think about behaviorally informed
policies? Do citizens have identifiable principles in mind when they approve or dis-
approve of the policies? Do citizens of different nations agree with each other?
From the answers to these questions, the authors identify six principles of
­legitimacy—a “bill of rights” for nudging that build on strong public support for
nudging policies around the world, while also recognizing what citizens ­disapprove
of. Their bill of rights is designed to capture citizens’ central concerns, reflecting
­widespread commitments to freedom and welfare that transcend national b ­ oundaries.

Cass R. Sunstein is the Robert Walmsley University Professor at Harvard, USA. From
2009 to 2012, he was Administrator of the White House Office of Information and
Regulatory Affairs. He is the founder and director of the Program on Behavioral
Economics and Public Policy at Harvard Law School.

Lucia A. Reisch is a behavioral economist and Professor at Copenhagen Business

School, Denmark. She also holds a permanent Guest Professorship at the Zeppelin
University of Friedrichshafen, Germany, and an appointment as honorary Leibniz
Chair, awarded by the German Leibniz Association and the Leibniz Institute of
Prevention Research and Epidemiology.
Routledge Advances in Behavioural
Economics and Finance
Edited by Roger Frantz

Traditionally, economists have based their analysis of financial markets and

corporate finance on the assumption that agents are fully rational, emotion-
less, self-interested maximizers of expected utility. However, behavioural
economists are increasingly recognizing that financial decision makers may
be subject to psychological biases, and the effects of emotions. Examples
of this include the effects on investors’ and managers’ decision-making
of such biases as excessive optimism, overconfidence, confirmation bias,
and illusion of control. At a practical level, the current state of the financial
­markets suggests that trust between investors and managers is of ­paramount
importance.

Routledge Advances in Behavioural Economics and Finance presents innovative and

cutting-edge research in this fast-paced and rapidly growing area, and will
be of great interest to academics, practitioners, and policy-makers alike.

All proposals for new books in the series can be sent to the series editor,
Roger Frantz, at [email protected].

1. Behavioural Economics and Business Ethics

Interrelations and Applications
Alexander Rajko
2. Bounded Rationality and Behavioural Economics
Graham Mallard
3. Behavioural Approaches to Corporate Governance
Cameron Elliott Gordon
4. Trusting Nudges
Toward a Bill of Rights for Nudging
Cass R. Sunstein and Lucia A. Reisch
For more information about this series, please visit: www.routledge.com/
Routledge-Advances-in-Behavioural-Economics-and-Finance/book-series/
RABEF
 Trusting Nudges

Toward a Bill of Rights for Nudging

Cass R. Sunstein and Lucia A. Reisch

First published 2019
by Routledge
2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN
and by Routledge
52 Vanderbilt Avenue, New York, NY 10017
Routledge is an imprint of the Taylor & Francis Group, an informa business
© 2019 Cass R. Sunstein and Lucia A. Reisch
The right of Cass R. Sunstein and Lucia A. Reisch to be identified as authors of
this work has been asserted by them in accordance with sections 77 and 78 of
the Copyright, Designs and Patents Act 1988.
All rights reserved. No part of this book may be reprinted or reproduced or
utilised in any form or by any electronic, mechanical, or other means, now
known or hereafter invented, including photocopying and recording, or in any
information storage or retrieval system, without permission in writing from the
publishers.
Trademark notice: Product or corporate names may be trademarks or registered
trademarks, and are used only for identification and explanation without intent
to infringe.
British Library Cataloguing-in-Publication Data
A catalogue record for this book is available from the British Library
Library of Congress Cataloging-in-Publication Data
Names: Sunstein, Cass R., author. | Reisch, Lucia A., author.
Title: Trusting nudges: toward a bill of rights for nudging / Cass R. Sunstein and
Lucia A. Reisch.
Description: Abingdon, Oxon; New York, NY: Routledge, 2019. |
Series: Routledge advances in behavioural economics and finance |
Includes bibliographical references and index.
Subjects: LCSH: Policy sciences—Psychological aspects. | Common
good—Psychological aspects. | Decision making—Psychological aspects. |
Economics—Psychological aspects. | Public opinion.
Classification: LCC H97 (ebook) | LCC H97 .S9535 2019 (print) |
DDC 320.6—dc23
LC record available at https://lccn.loc.gov/2018043517

ISBN: 978-1-138-32278-3 (hbk)

ISBN: 978-0-429-45164-5 (ebk)

Typeset in Joanna
by codeMantra
To the memory of our fathers
 Contents

List of figuresviii
List of tablesix
Prefacexi

1 Why public opinion matters1

2 The United States, 1: Evidence8

3 The United States, 2: Principles19

4 Europe29

5 A global consensus? Not quite53

6 Trusting nudges71

7 Educative nudges and noneducative nudges95

8 Misconceptions119

9 A Bill of Rights for Nudging128

Acknowledgments 139
Index141
 Figures

4.1 Bar charts for information nudges: Government

campaigns, total support in % (unweighted)37
4.2 Bar charts for information nudges, governmentally
mandated; total support in % (unweighted)38
4.3 Bar charts for default rules, total support in % (unweighted)39
4.4 Bar chart for subliminal ads, total support in % (unweighted)40
4.5 Bar charts for other mandates41
5.1 General information campaigns. CI, confidence interval59
5.2 Mandatory information imposed by governments. CI,
confidence interval60
5.3 Mandatory default rules imposed by governments. CI,
confidence interval61
5.4 Mandatory subliminal advertising. CI, confidence interval62
5.5 Mandatory choice architecture. CI, confidence interval62
6.1 Correlation heatmap of relevant variables76
6.2 Overall nudge approval, conditional on trust in institutions77
6.3 Predicted marginal probabilities for approval,
conditional on institutional trust80
A6.1 Nudge approval by gender (all studies)84
A6.2 Nudge approval among different time periods for
South Korea (K), Denmark (DK), and Germany (G)86
 Tables

2.1 American attitudes toward prominent recent nudges10

2.2 American attitudes toward six educational campaigns11
2.3 American attitudes toward environmental and public
health nudges12
2.4 American attitudes toward some potentially
provocative nudges13
2.5 Unpopular defaults15
2.6 Unpopular education campaigns and disclosure17
4.1 The 15 items of the survey32
4.2 Overview on approval rates for the 15 nudges in the six
surveyed countries34
A4.1 Samples, sampling, and methodology48
A4.2 Overview of political parties in the surveyed countries49
A4.3 Clusters of the political parties in the surveyed countries50
A4.4 Estimates of demographics and political attitude on
nudge approval: Multilevel analysis51
5.1 Estimates of selected socio-demographics and
political attitude on nudge approval per nudge cluster:
Multilevel analysis64
A5.1 Observations RIM weighted/unweighted for all countries70
6.1 Weighted OLS regression for different nudge clusters
(2018 survey)78
x Tables

A6.1 Samples and sampling in the different countries (2018

study): Types of representativeness and methodology
(2018 survey)87
A6.2 Descriptive statistics—all variables (2018 survey)88
A6.3 Samples and sampling in the different countries: Types
of representativeness and methodology (16 countries,
all samples, all waves)90
A6.4 Weighted OLS regression for different nudge clusters93
7.1 Support for System 1 and System 2 nudges102
7.2 Preference when System 1 nudge “significantly more
effective”103
7.3 Preference for System 1 nudges with quantitative
information104
7.4 Preference for System 1 nudges when System 2 nudge
is “significantly more effective”105
7.5 Support for System 1 and System 2 nudges by partisan
affiliation106
7.6 Preference when System 1 nudge “significantly more
effective” by partisan affiliation 106
7.7 Preference with quantitative information by partisan
affiliation107
7.8 Preference when System 2 nudge is “significantly more
effective” by partisan affiliation107
7.9 Voter registration109
7.10 Childhood obesity110
7.11 Abortion110
7.12 Voter registration by partisan affiliation111
7.13 Childhood obesity by partisan affiliation111
7.14 Abortion by partisan affiliation111
7.15 Within-subjects results113
 Preface

All over the world, governments have been adopting behaviorally informed
policies—policies that grow out of new findings about what human beings
are actually like, and how they think and act. A growing literature exam-
ines those policies and how to make them work. But what do citizens of
the world actually think about such policies? Do they approve of them?
Do Russians disagree with Americans? How much, and exactly where?
What are the differences among citizens of China, Japan, and South Korea?
Australia and Brazil? France and Denmark? Ireland and United Kingdom?
Hungary and Germany?
Over the last several years, we have been exploring these questions. We
have conducted or have cooperated in nationally representative surveys in
seventeen nations:

Australia Germany Russia

Belgium Hungary South Africa
Brazil Ireland South Korea
China Italy United Kingdom
Denmark Japan United States
France Mexico

The list is far from complete, of course, but it captures a significant sub-
set of the nations of the world. As an initial step, we have narrowed and
xii Preface

focused our efforts by asking about a set of policies that have received
special attention from those interested in behavioral science in general and
behavioral economics more specifically. That focus allows for some sharp
comparisons and helps uncover (we think) a range of convictions about
freedom, welfare, trust, and paternalism.
Those convictions tell us something about where members of the human
species appear to agree with one another, and where national boundaries
create significant divergences. We are also willing to speculate, a bit, about
what accounts for differences within and across nations.
Ideally, of course, we would like to obtain answers to a very wide assort-
ment of questions, involving every policy under the sun (or something
close to that). We like to think that our narrower approach offers some
initial clues about what would emerge from that broader inquiry. For now,
we present what we have learned.
Our ultimate claim, offered by way of conclusion, involves a Bill of
Rights for Nudging, one that grows directly out of empirical findings about
the beliefs of citizens of numerous nations. By a Bill of Rights, we mean
to suggest not judicially enforceable rights, but a set of commitments, rea-
sonably taken to be rights, that political officials should respect. The Bill
of Rights includes a ban on manipulation, respect for people’s values and
interests, transparency, and a prohibition on the pursuit of illicit ends. In
the process of sketching that Bill of Rights, we will have something to say
about human autonomy and social welfare as well.
 1
Why Public Opinion Matters

The last several years have seen an outpouring of work on behavioral eco-
nomics, behaviorally informed policies, and “nudges,” understood as
interventions that steer people in particular directions but that also allow
them to go their own way.1 A reminder is a nudge (“You have a doctor’s

1 See generally, e.g., Richard H. Thaler and Cass R. Sunstein, Nudge: Improving Decisions About
Health, Wealth, and Happiness (2008) (arguing that the public’s choices are influenced by
small factors through the design of experiences, and that, with the knowledge of predict-
able psychology, nudges can help people make beneficial rational decisions); Richard H.
Thaler, Misbehaving: The Making of Behavioral Economics (2015) (discussing various behavioral
economic aspects that include endowment effects, mental accounting, consumption,
self-control issues, and explains the importance of finding new methods of economic
research); David Halpern, Inside The Nudge Unit: How Small Changes Can Make a Big Difference
(2015) (showing how a policy experiment, the Behavioural Insights Team, developed
small persuasive methods, nudges, that created solutions in tax, health care, crime re-
duction, and spurred economic growth); The World Bank, World Development Report 2015:
Mind, Society, and Behavior (2015) (showing how the use of human psychology will force
the redesigning of policies that target people’s choices and actions), www.worldbank.
org/content/dam/Worldbank/Publications/WDR/WDR%202015/WDR-2015-Full-
Report.pdf; Pete Lunn, Regulatory Policy and Behavioral Economics (2014) (­describing the
changes of behaviorally informed politics, how they are being regulated, and ­finding
new approaches to economic challenges); Rhys Jones et al., Changing ­Behaviours: On
The Rise of the Psychological State (2013) (exploring the evolution of using the human
2 W hy Public Opinion Matters

appointment tomorrow”). So is a warning (“Construction area”). A GPS

device nudges; a default rule, specifying what happens if people do nothing,
nudges. Disclosure of important information (about the risks of smoking
or the costs of borrowing) counts as a nudge. Save More Tomorrow plans,
allowing employees to sign up to give some portion of their future earn-
ings to pension programs, are nudges.2 So are Give More Tomorrow Plans,
allowing employees to sign up to give some portion of their future earn-
ings to charities.3 A recommendation is a nudge. A criminal penalty, a civil
fine, a tax, and a subsidy are not nudges, because they impose significant
material incentives on people’s choices. (To be sure, a very small fee or
subsidy might be purely nominal and yet prove effective for behavioral
reasons; if so, it might well be fair to characterize it as a nudge.)
In many nations, public officials have been drawn to nudges.4 In
2009, the United Kingdom created a Behavioural Insights Team, focused
largely on the uses of nudges, and choice architecture, to improve social
outcomes; its results have been impressive.5 Nudges play a large role in
American initiatives in multiple areas, including environmental protec-
tion, financial regulation, anti-obesity policies, and education.6 In 2014,

psyche to implement governing practices and challenges faced in the areas of health,
finance, and the environment); Nudge, Rev. Phil. Psych. 6, 341–529 (2015) (dedicating
the entirety of issue three to the topic of nudges); Riccardo Rebonato, Taking Liberties: A
Critical Examination of Libertarian Paternalism (2012); Mark D. White, The Manipulation of Choice:
Ethics and Libertarian Paternalism (2013). We are keenly aware that there are disputes about
definitional issues and that the various sources cited here do not use the same defini-
tion. For our purposes, use of the standard definition is sufficient, and the surrounding
debates need not detain us.
2 See Thaler, supra note 1, at 309–22. See generally Shlomo Benartzi, Save More Tomorrow: Practi-
cal Behavioral Finance Solutions to Improve 401(K) Plans (2012).
3 Anna Breman, Give More Tomorrow: Two Field Experiments on Altruism and Intertem-
poral Choice, J. Pub. Econ. 95, 1349 (2011).
4 See generally Halpern, supra note 1; Lunn, supra note 1; Jones et al., supra note 1; Cass R.
Sunstein, Simpler: The Future of Government (2013); House of Lords, Science and ­Technology
Select Committee, Behaviour Change (2011), www.publications.parliament.uk/pa/
ld201012/ldselect/ldsctech/179/179.pdf.
5 See generally Halpern, supra note 1.
6 See generally Sunstein, supra note 4. For recent evidence in one domain, showing that
calorie labels are having significant effects, see Partha Deb and Carmen Vargas, Who Benefits
from Calorie Labeling? An Analysis of its Effects on Body Mass, 1–3 (2016) (Nat’l Bureau of Econ.
Research, Working Paper No. 21992). On behaviorally informed approaches, nudges,
and credit cards, see Sumit Agarwal et al., Regulating Consumer Financial Products: Evidence from
Credit Cards, 16–22 (2013) (Nat’l Bureau of Econ. Research, Working Paper No. 19484).
 Why P ublic Opinion Matters 3

the United States created its own Social and Behavioral Sciences Team,
now called the Office of Evaluation.7
With an emphasis on poverty and development, the World Bank
devoted its entire 2015 report to behaviorally informed tools, with a par-
ticular focus on nudging.8 The World Bank has a Mind, Behavior, and
Development Unit, focused on poverty and development. Also in 2015,
President Barack Obama issued a historic Executive Order on uses of
behavioral sciences in federal agencies, calling for attention to the assort-
ment of tools standardly associated with nudging.9 The order continues
in effect. Behavioral science teams can be found in dozens of countries,
including Australia, the Netherlands, France, Canada, Ireland, Germany,
and Qatar. And even when formal teams are not in place, departments
and ­ministries—and the offices of presidents, prime ministers, and
­chancellors—are often using behavioral insights. In fact much of the most
important behaviorally informed work comes from departments and min-
istries, not from dedicated units of any kind.
The reason for the mounting interest should not be obscure. Nations
would like to make progress on pressing social problems with tools that
actually work and that do not cost a great deal. They would like to save
money and lives. They would like to improve education and to reduce
poverty. They would like to fuel economic growth. If governments can
achieve these goals with instruments that impose minimal burdens and that
preserve freedom of choice, they will take those tools extremely seriously.
In domains that include savings policy,10 climate change,11 poverty,12 and

7 Maya Shankar, Using Behavioral Science Insights to Make Government More Effective, Simpler, and More
People-Friendly, White House Blog (February 9, 2015, 12:19 PM), www.whitehouse.
gov/blog/2015/02/09/using-behavioral-science-insights-make-government-more-­
effective-simpler-and-more-us. William J. Congdon and Maya Shankar, The Role of Be-
havioral Economics in Evidence-Based Policymaking, The Annals of the American Academy of
Political and Social Sciences (AAAPS) 678, 81–92 (2018). Article first published online: June
18, 2018; https://doi.org/10.1177/0002716218766268.
8 See The World Bank, supra note 1.
9 See Exec. Order No. 13707, 80 Fed. Reg. 56, 365 (Sept. 15, 2015).
10 See Thaler, supra note 1, at 309–22.
11 See generally Frank Beckenbach and Walter Kahlenborn, eds., New Perspectives for Environ-
mental Policies through Behavioral Economics (2016); Lucia A. Reisch and John Thøgersen, eds.,
Handbook of Research on Sustainable Consumption (2015).
12 See generally Sendhil Mullainathan and Eldar Shafir, Scarcity: Why Having Too Little Means So
Much (2013).
4 W hy Public Opinion Matters

health care,13 among many others, behaviorally informed approaches have

attracted considerable attention, and often led to concrete reforms.
At the same time, some people have raised serious ethical concerns and
objections.14 An evident question is whether nudges should be counted
as unacceptably manipulative or as an interference with freedom, rightly
understood.15 To make progress on the ethical questions, it would be
possible to refer to defining commitments of various kinds—involving
autonomy, dignity, welfare, and self-government—and to ask whether
some, many, or all nudges run afoul of those commitments. It would also
be possible to imagine cases in which nudges might have illicit goals, in
which case the question would be how to identify the category of goals
that count as illicit.
This is a normative task, not an empirical one. But while the normative
discussions continue, it is worthwhile to ask some empirical questions.
What do people actually think about nudging and choice architecture? Do they have serious
ethical objections to official nudges, or to nudges that take the form of law? Or do they believe
that nudges are acceptable or desirable, even morally obligatory? Do they distinguish among
nudges? What kinds of distinctions do they make?
The answers cannot, of course, dispose of the ethical questions. The issue
is how to resolve those questions in principle, and empirical findings about
people’s answers are not decisive. Perhaps those answers are confused,
insufficiently considered, based on behavioral biases, or otherwise wrong.
There is a risk that if people are responding to survey questions, they will
not have time or opportunity to reflect, especially if those questions do not
offer relevant facts (for example, about the costs and the benefits of the
policies in question or the other policy options available). Quick answers
to survey questions are not exactly the best way to obtain policy guidance.

13 See generally Douglas E. Hough, Irrationality in Health Care: What Behavioral Economics Reveals
About What We Do and Why (2013).
14 The best discussion is Rebonato, supra note 1. See also the various contributions to
Nudge, Rev. Phil. Psych. 6, 341–529 (2015); White, supra note 1; Jeremy Waldron, It’s All
For Your Own Good, New York Review of Books (October 9, 2014), www.nybooks.com/
articles/­archives/2014/oct/09/cass-sunstein-its-all-your-own-good. Consider in par-
ticular this question: “Deeper even than this is a prickly concern about dignity. What
becomes of the self-respect we invest in our own willed actions, flawed and misguided
though they often are, when so many of our choices are manipulated to promote what
someone else sees (perhaps rightly) as our best interest?” Id. at 4. We shall have some-
thing to say about this question in Chapter 7.
15 T. M. Wilkinson, Nudging and Manipulation, Pol. Stud. 61, 341, 354 (2013).
 W hy P ublic Opinion Matters 5

Even if their answers are reflective, perhaps people do not value autonomy
or dignity highly enough, or perhaps they do not quite know what those
concepts mean. Perhaps people pay too little attention to social welfare,16 or
perhaps their judgments about social welfare are off the mark, at least if they
are not provided with a great deal of information. We will explore the pos-
sibility that different nations, and different groups within the same nation,
offer different answers, suggesting an absence of consensus.
Behavioral scientists would emphasize a related point: People’s answers to
ethical questions, or questions about moral approval or disapproval, might
well depend on how such questions are framed. Slight differences in framing
can yield dramatically different answers. Those differences are themselves a
nudge; they can have major effects, and they are not easy to avoid.17
Here is a small example of how ethical judgments can depend on fram-
ing.18 If people are asked whether they think that young people should
be valued more than old people, they will usually say, “certainly not!”
They will strenuously resist the idea that government should give a higher
value to young lives than to old ones. But suppose that people are asked
whether they want either (1) to save seventy people under the age of five
or (2) to save seventy-five people over the age of eighty. It is reasonable to
speculate (and evidence confirms) that most people will choose (1), thus
demonstrating that they are willing to value a young person more than
an old one.19 It would be child’s play to frame nudges so as to elicit one’s
preferred answer to ethical questions.
Notwithstanding these points, people’s answers to carefully designed
questions are interesting, because they elicit intuitions, potentially revealing
patterns of thinking among those who are not required to spend a great deal
of time on them. For three different reasons, they can also help to illuminate
political, legal, and ethical problems. The first, and the most important, is
that in democratic societies (and in nondemocratic societies as well), it is

16 See Louis Kaplow and Steven Shavell, Fairness Versus Welfare 7−8 (2006).
17 See generally Perspectives on Framing (Gideon Keren, ed., 2010).
18 See Shane Frederick, Measuring Intergenerational Time Preference: Are Future Lives
­Valued Less?, J. Risk and Uncertainty 256, 39, 40 (2003) (showing that people’s preferences
for life-saving programs depend on framing).
19 See Maureen L. Cropper et al., Preferences for Life Saving Programs: How the Public
Discounts Time and Age, J. Risk and Uncertainty 8, 243, 258–59 (1994) (explaining that,
while most survey respondents do not decide based only on a person’s life expectancy,
the fraction of people that decide to save younger people usually increases concurrently
as the ratio of young to old people saved grows).
6 W hy Public Opinion Matters

inevitable that public officials will attend to what citizens actually think. If
citizens have strong ethical objections, democratic governments will hesi-
tate before proceeding (if only because of electoral self-interest).
Such objections can operate as a kind of presumptive or de facto veto.
No public official will entirely disregard a strongly felt moral concern on
the part of significant segments of the public. And if people do not have
moral objections, and if they welcome nudges as helpful and desirable,
public officials will be attentive to their views. Widespread public approval
can operate as a license or a permission slip, or perhaps as a spur or a
prod.20 Similar points hold in nondemocratic nations, where public offi-
cials know that they can learn from what citizens think, and where they
are keenly aware that their power might depend on listening to them and
considering their concerns.
The second reason is epistemic: People’s judgments provide relevant
information about how to think about the ethical issues even if that infor-
mation is not conclusive. It is not necessary to make strong claims about
the “wisdom of crowds,” especially on contested ethical issues, in order to
believe that an ethical judgment on the part of those who might be subject
to nudges deserves respectful attention. Public officials should be humble
and attentive to the views of others, and if strong majorities favor or oppose
nudges, their views are entitled to consideration. We do not mean to sug-
gest that public approval or disapproval in a survey setting should dispose
of ethical (or other) issues. Reflection, deliberation, expertise, and informa-
tion greatly matter (see Chapter 9). But public reactions deserve attention.
The third reason involves the commitment to democratic self-­government.
If that commitment matters, officials should pay attention to what people
think, even if they disagree. To be sure, people’s considered judgments might
diverge from what emerges from brief surveys. Their considered judgments
deserve priority. And if public officials have a clear sense that an approach
or a nudge would reduce social welfare, there is a strong argument that they
should not adopt that approach or nudge even if people would like them to
do so—just as there is a strong argument that they should adopt an approach
that increases social welfare even if people oppose it. Individual rights and

20 We are bracketing here questions about interest-group dynamics and coalition forma-
tion, which can of course complicate official judgments. Politicians are interested in
many things that bear on reelection, not merely the views of the median voter. And of
course, there are important differences between the legislative and executive branches
on this count, with the latter frequently having more “space” for technocratic judgment.
 W hy P ublic Opinion Matters 7

private autonomy also have their claims, whatever majorities may think. We
shall explore these points in Chapter 9. But when public officials are uncer-
tain about whether an approach is desirable, it is reasonable, in the name of
self-government, for them to give consideration to the views of members of
the public.
As we shall see, current research in many nations supports a single con-
clusion: At least in general, the majority of citizens of most nations have no
views, either positive or negative, about nudging in general; their assessment
turns on whether they approve of the purposes and effects of particular nudges. As we shall
see, strong majorities in diverse nations tend to be supportive of nudges of
the kind that have been seriously proposed, or acted on, by actual institu-
tions in recent years.
With some qualifications, this enthusiasm extends across standard partisan
lines; perhaps surprisingly, it unifies people with diverse political convic-
tions. So long as people believe that the goal is both legitimate and important,
they are likely to favor nudges in its direction. When there is disagreement, it
is usually because of differences about the legitimacy and the importance of
the goal of the particular nudge. This is an important finding, because it sug-
gests that most people do not share the concern that nudges, as such, should
be taken as manipulative or as an objectionable interference with autonomy.
Some preliminary evidence suggests that people are far more negative about
mandates and bans, even when they are taken to have perfectly legitimate
ends; many people do care about freedom of choice as such, and they will
reject many well-motivated policies that do not allow for it.
To summarize the story that we shall tell here: People are most likely
to oppose those nudges that (1) promote what they see as illicit goals or
(2) are perceived as inconsistent with either the interests or values of most
choosers. A more particular finding, one that counts against some default
rules, is that people do not want policymakers to produce economic or other losses by
using people’s inertia or inattention against them. In addition, people tend to prefer
nudges that target deliberative processes to those that target unconscious
or subconscious processes, and may react against the latter—though they
do not by any means rule the latter out of bounds and will often approve
of them as well. When the political valence of nudging is clear, their evalu-
ation of nudges much turns on that valence, which reinforces the general
view that in most cases, it is people’s assessment of the ends of particular
nudges, rather than of nudging as such, that settles their judgments.
Now for the details.
 2
The United States, 1:
Evidence

To test public opinion in the United States, we devised a nationally repre-

sentative survey involving thirty-four nudges. The survey was administered
in 2016 by Survey Sampling International and included 563 Americans,
with a margin of error of plus or minus 4.1 percentage points. People were
asked simply whether they approved or disapproved of the relevant nudges.
From their responses, two dominant principles emerge. First, Americans
reject nudges that promote what they see as illicit ends (such as religious
or political favoritism). Second, Americans reject nudges that they view as
inconsistent with the interests or values of most choosers. By contrast, there
is widespread support for nudges that are taken to have legitimate ends and
to be consistent with the interests and the values of most choosers.
We shall turn to other nations in due course, but in the United States, at
least, it follows that numerous nudges—default rules, warnings, and pub-
lic education campaigns—are likely to attract bipartisan support, so long as
people approve of their ends and think that they are consistent with choos-
ers’ values and interests. Several of the policies tested here can be counted
as highly tendentious and arguably manipulative. Nonetheless, they
attracted majority support, with the single (and highly exotic) exception
 The U nited States, 1 : Evidence 9

of ­subliminal advertising (which, surprisingly, receives substantial minor-

ity support in the context of efforts to combat smoking and overeating).
It follows that Americans are reluctant to reject nudges as unacceptably
manipulative.1 Their evaluations are dominated by their assessment of the
legitimacy of the underlying ends.2
As we will see, political divisions sometimes affect the level of support;
Democrats are more favorably disposed toward certain health and safety
nudges than are Republicans. In cases that raise strong partisan differences,
such divisions will map onto nudges as well. It is easy to imagine nudges
that Republicans would support more strongly than would Democrats—for
example, nudges to discourage abortion or religious instruction. But across
a wide range, clear majorities of Democrats and Republicans (and also inde-
pendents) are in full agreement about what they support and what they reject.

Popular nudges
In recent years, the US government has adopted or promoted a large
number of nudges. Three of the most prominent include: (1) mandatory
calorie labels at chain restaurants; (2) mandatory graphic warnings on
cigarette packages3 (struck down by a federal court of appeals4); and (3)
automatic enrollment in savings plans, subject to opt out.5 The nation-

1 Notably, Janice Jung and Barbara Mellers, American Attitudes Towards Nudges, Judgement
and Decision Making 11, 62 (2016) find that people reject this nudge as manipulative:
“Use of increasingly narrower white lines on roadways that create the visual illusions of
speeding up to control vehicle speeding.” Id. at 66. This nudge might be taken to fall in
the same category as subliminal advertising because it is taken to fall right on the line
between manipulation and deception.
2 To be sure, provision of information about the consequences of nudges might unsettle
some of people’s responses, and perhaps move people in the direction of what follows
from an all-things-considered welfare assessment. If so, any such movements would
be consistent with the general claim here; they would merely reflect a more informed
judgment about what ends would, in fact, be promoted by nudges. For example, people
might be less enthusiastic about compulsory disclosure of uses of GMOs if they were
convinced that such disclosure did not provide useful information and might mislead
people. We shall have something to say about this question in later chapters.
3 See Cigarette Package and Advertising Warnings: Required Warnings, 21 C.F.R. § 1141.10
(2015).
4 On the FDA’s effort to require graphic warnings on packages, see R.J. Reynolds T­ obacco Co.
v. FDA, 823 F. Supp. 2d 36 (D.D.C. 2011), aff’d on other grounds, 696 F.3d 1205 (D.C. Cir.
2012).
5 For discussion of relevant laws and policies, see generally Automatic: Changing the Way America
Saves (William G. Gale et al., eds., 2009).
10 The U nited States , 1: Evidence

Table 2.1  American attitudes toward prominent recent nudges

Calorie Graphic warn- Federal Federal

labels ings (cigarettes) encouragement: mandate:
Auto-enrollment Auto-enrollment
Total support 87/13 74/26 80/20 71/29
(approval/
disapproval in
percentages)
Democrats 92/8 77/23 88/12 78/22
Independents 88/12 74/26 75/25 67/33
Republicans 77/23 68/32 73/27 62/38

ally representative sample found substantial majority support for all

three policies, including support for (3) regardless of whether it consists
of federal “encouragement” of such enrollment or a federal mandate for
automatic enrollment, imposed on large employers (see Table 2.1).
About 87 percent of Americans favored calorie labels and 74 percent
favored graphic warnings.6 Both policies had strong majority support from
Democrats, Republicans, and independents. Overall, 80 percent and 71 per-
cent respectively approved of encouraged and mandatory enrollment in
savings plans. Here as well, all three groups showed strong majority support.7
Three educational campaigns also attracted widespread approval (see
Table 2.2). Respondents were overwhelmingly supportive of a public
education campaign from the federal government to combat childhood
obesity (82 percent approval, again with strong support from Democrats,
Republicans, and independents). Similarly, they were highly supportive
of a public education campaign from the federal government designed to
combat distracted driving, with graphic stories and images (85 percent
approval). About 75 percent of people favored a federal education campaign
to encourage people not to discriminate on the basis of sexual orientation,
though here there was a noteworthy division across party lines—85 percent
of Democrats, 57 percent of Republicans, and 75 percent of independents.

6 Note that there were statistically significant differences with respect to calorie labels
between Republicans (77 percent approval) and both Democrats (92 percent approval)
and independents (88 percent approval).
7 Here as well, there were statistically significant differences between Democrats and Re-
publicans for both policies and between Democrats and independents with respect to
encouragement (Encouraged: 88 percent of Democrats, 73 percent of Republicans, and
75 percent of independents. Mandated: 78 percent of Democrats, 62 percent of Repub-
licans, and 67 percent of independents).
 The United States, 1 : Evidence 11

Table 2.2  American attitudes toward six educational campaigns8

Childhood Distracted Sexual Movie Animal Obesity

obesity driving orientation theaters Welfare (arguably
discrimination Society manipulative)
Total support 82/18 85/15 75/25 53/47 52/48 57/43
(approval/
disapproval in
percentages)
Democrats 90/11 88/12 85/15 61/39 59/41 61/40
Independents 81/19 84/16 75/25 51/49 55/45 60/40
Republicans 70/30 80/20 57/43 41/59 34/66 47/53

Three other educational campaigns attracted majority support,

but at significantly lower levels and with only minority approval
from Republicans. About 53 percent of Americans favored a federal
requirement that movie theaters run public education messages to
discourage people from smoking and overeating. Democrats showed
higher approval ratings than Republicans (61 percent as opposed to
41 percent, with independents at 51 percent). By a very small majority
(52 percent), Americans supported a public education campaign, by
the federal government itself, to encourage people to give money to
the Animal Welfare Society of America (a hypothetical organization)
(59 percent of Democrats, 34 percent of Republicans, and 55 percent
of independents; party was a statistically significant factor). This latter
finding seems surprising; it could not easily be predicted that respond-
ents would want their government to design a campaign to promote
donations to an animal welfare society.
About 57 percent of people supported an aggressive public education
campaign from the federal government to combat obesity, showing obese
children struggling to exercise and also showing interviews with obese
adults, who are saying such things as, “My biggest regret in life is that I
have not managed to control my weight,” and “To me, obesity is like a
terrible curse.” This question was designed to test people’s reactions to a
tendentious and arguably manipulative campaign, which might have been
expected to receive widespread disapproval, but it did not. Indeed, one
of the goals of the question was to establish such disapproval—but it was
not found here. Here, there was a significant disparity between Democrats

8 Percentages may not total 100 due to rounding.

12 The U nited States , 1: Evidence

(61 percent approval) and independents (60 percent approval) on the one
hand and Republicans on the other (47 percent approval); the difference
between the views of Democrats and those of Republicans was statistically
significant.
Most Americans were also supportive of multiple efforts to use choice
architecture to promote public health and environmental protection (see
Table 2.3). In recent years, there has been considerable international discus-
sion of “traffic lights” systems for food, which would use the familiar red,
yellow, and green to demarcate health rankings.9 In the United States, the
national government has shown no official interest in these initiatives, but
with respondents in the nationally representative survey, the idea attracted
strong support (64 percent). There was also majority approval of automatic
use of “green” energy providers, subject to opt out10—­perhaps surpris-
ingly, with support for automatic use of green energy whether it consisted
of federal “encouragement” (72 percent) or instead a federal mandate on
large electricity providers (67 percent).11 In these cases, there were sig-
nificant differences across partisan lines, but majorities of Democrats,
Republicans, and independents were all supportive.

Table 2.3  American attitudes toward environmental and public health nudges

GMO Salt Healthy food Traffic Organ Encouragement: Mandate:

labels labels placement lights donor Green energy Green energy
choice
Total sup- 86/14 73/27 56/44 64/36 70/30 72/28 67/33
port (in
percentages)
Democrats 89/11 79/21 63/37 71/29 75/25 82/18 79/21
Independents 87/13 72/28 57/43 61/39 69/31 66/34 63/37
Republicans 80/20 61/39 43/57 57/43 62/38 61/39 51/49

9 See Anne N. Thorndike et al., Traffic-Light Labels and Choice Architecture: Promoting
Healthy Food Choices, Am. J. Preventive Med. 46, 143, 143–44 (2014).
10 See Cass R. Sunstein and Lucia A. Reisch, Automatically Green: Behavioral Economics and
Environmental Protection, Harv. Envtl. L. Rev. 38, 127, 134–35 (2014).
11 On the difficulty of this question, see id. at 155–57.
 The U nited States, 1 : Evidence 13

Table 2.4  American attitudes toward some potentially provocative nudges12

Listing Automatic Husband’s last Mandatory Mandatory

incumbent voter name manufacturing manufacturing
politician registration label: Labor label: Aiding
first violations terrorists
Total sup- 53/47 53/47 58/42 60/40 54/46
port (in
percentages)
Democrats 58/42 63/37 61/40 67/33 56/44
Independents 51/49 50/50 56/44 57/43 49/51
Republicans 47/53 39/61 57/43 50/50 58/42

Most respondents were in favor of requiring companies to disclose

whether the food they sell contains genetically modified organisms (GMOs)
(86 percent approval). There was strong majority support (73 percent) for
a mandatory warning label on products that have unusually high levels of
salt, as in, “This product has been found to contain unusually high levels
of salt, which may be harmful to your health.” Perhaps surprisingly, most
respondents (but not most Republicans) approved of a state requirement
that grocery stores put their most healthy foods in prominent, visible loca-
tions (56 percent approval: 63 percent from Democrats, 43 percent from
Republicans, 57 percent from independents). Respondents also supported
a state requirement that people must say, when they obtain their driv-
er’s license, whether they want to be organ donors (70 percent approval:
75 percent from Democrats, 62 percent from Republicans, 69 percent
from independents).13 For all of these policies, the differences between
Democrats and Republicans were statistically significant.
Five other forms of choice architecture, which might be expected to
be far more controversial, nonetheless obtained majority support (see
Table 2.4). The first would list the name of the incumbent politician first,
on every ballot. It might be expected that this pro-incumbent nudge would
be widely rejected, because respondents might not want the voting process

2 Percentages may not total 100 due to rounding.

1
13 Another study, discussed below, finds that most Americans reject a default rule to the ef-
fect that people would be presumed to be organ donors, subject to opt out. See William
Hagman et al., Public Views on Policies Involving Nudges, Rev. Phil. and Psychol. 6, 439, 446
(2015).
14 The U nited States , 1: Evidence

to be skewed in favor of incumbents and because any effort to enlist order

effects might be seen as manipulative (as indeed it should be). But a bare
majority (53 percent) approved of this approach, perhaps because most
people believed that it would promote clarity, perhaps because they did not
see the risk of bias from order effects.
There was also majority approval (53 percent) for the approach, adopted
in several American states such as Oregon and in the majority of European
nations, of automatically registering eligible citizens as voters, subject to opt
out. Interestingly, most Republicans (61 percent) rejected this approach. One
reason might be that they believe that people who do not take the time to regis-
ter to vote ought not to be counted as voters. Another reason is that they might
believe that Oregon’s approach would favor Democrats. Yet another reason is
that they might believe that such an approach would increase the risk of fraud.
By a modest majority, most people (58 percent) also approved of an
approach by which women’s last names would automatically be changed
to that of their husband, subject to opt out. This approach obtained major-
ity support from Democrats, Republicans, and independents. This result
is especially noteworthy in view of the fact that an approach to this effect
would almost certainly be unconstitutional as a form of sex discrimination,
even if it tracked behavior and preferences.14 We might expect a difference
between men and women on this question, but notably, 58 percent of both
groups approved of this approach.
Finally, there was majority support for a federal labeling requirement
for products that come from companies that have repeatedly violated the
nation’s labor laws (such as laws requiring occupational safety or for-
bidding discrimination). About 60 percent of participants supported
that policy, with a significant difference between Democrats (67 percent
approval) and Republicans (50 percent approval). There was also majority
support for federally required labels on products that come from countries
that have recently harbored terrorists. This approach attracted 54 percent
approval—56 percent from Democrats, 58 percent from Republicans, and
49 percent from independents.

14 See Craig v. Boren, 429 US 190, 200–4 (1976) (finding Oklahoma’s statute permitting
­females over the age of eighteen to buy 3.2 percent beer but prohibiting males under the
age of twenty-one from buying the same beer “invidiously discriminates against males
18−20 years old”). For valuable discussion of the general topic, see Elizabeth F. Emens,
Changing Name Changing: Framing Rules and the Future of Marital Names, U. Chi. L. Rev.
74, 761, 772–74 (2007).
 T he United States , 1 : Evidence 15

Unpopular nudges
By contrast, twelve nudges were widely disapproved. Of these, seven
involved uses of default rules (see Table 2.5). Two of these defaults were
designed so as to be not merely provocative but also highly offensive, in
the sense of being violative of widely held principles of neutrality, and
strong majorities took them exactly as they were designed.
Under the first, a state would assume that people want to register as
Democrats, subject to opt out if people explicitly say that they want to
register as Republicans or independents. Of course, a default rule of this
kind should be taken as an effort to skew the political process (and it
would certainly be unconstitutional for that reason).15 The overwhelming
majority of people, including three-quarters of Democrats, rejected this
approach (26 percent total approval: 32 percent of Democrats, 16 percent
of Republicans, and 26 percent of independents, with statistically signifi-
cant differences between Democrats and Republicans). The second was a
state law assuming that people are Christian, for purposes of the census,

Table 2.5  Unpopular defaults

Democrat Christian Wife’s Red Cross Animal United Carbon

registration on census last name Welfare Way emissions
Society charge
Total support 26/74 21/79 24/76 27/73 26/74 24/76 36/64
(in percentages)
Democrats 32/68 22/78 28/72 30/70 30/70 26/74 43/57
Independents 26/74 17/83 23/77 28/72 25/75 25/75 34/66
Republicans 16/84 27/73 18/82 20/80 20/80 17/83 25/75

15 In principle, the problem would be most interesting in an area in which the default
rule tracked reality. If most people are, in fact, Democrats, is it clearly objectionable if
a city or state assumes that they are for purposes of registration? The answer is almost
certainly yes; political affiliations should be actively chosen, not assumed by govern-
ment. This principle almost certainly has constitutional foundations (though it has not
been tested): If a voting district consisted of 80 percent Democratic voters, it would
not be acceptable to assume that all voters intend to register as Democrats. But we are
aware that this brief comment does not give anything like an adequate answer to some
complex questions about the use of “mass” default rules that track majority preferences
and values. For discussion, see Cass R. Sunstein, Choosing Not To Choose: Understanding the Value
of Choice 77 (2015).
16 The U nited States , 1: Evidence

unless they specifically state otherwise. Such a default rule could also be
seen as an attempt to push religious affiliations in preferred directions
(and it would similarly be unconstitutional).16 Here too, there was wide-
spread disapproval (21 percent overall approval: 22 percent of Democrats,
27 ­percent of Republicans, and 17 percent of independents).
The third unpopular default rule (completing the set of unconstitu-
tional nudges) involved a state law assuming that upon marriage, husbands
would automatically change their last names to that of their wives, subject
to opt out (24 percent total approval: 28 percent of Democrats, 18 percent
of Republicans, and 23 percent of independents). Interestingly, there was
no gender disparity here (just as with the question that involved the oppo-
site defaults); 24 percent of both men and women approved. With the
fourth, the federal government would assume, on tax returns, that people
want to donate fifty dollars to the Red Cross, subject to opt out if people
explicitly say that they do not want to make that donation (27 percent
approval: 30 percent of Democrats, 20 percent of Republicans, 28 per-
cent of independents). The fifth was identical but substituted the Animal
Welfare Society for the Red Cross. Not surprisingly, that question also
produced widespread disapproval (26 percent approval: 30 percent of
Democrats, 20 percent of Republicans, and 25 percent of independents).
Somewhat surprisingly, and revealingly, the numbers were essentially the
same for the two charities, even though it might be expected that pre-
sumed donations for the Red Cross would be more popular.
With the sixth, state government assumed that state employees would
give twenty dollars per month to the United Way, a large US charity,
subject to opt out. It might be expected that because state government
and state employees were involved, approval rates might grow. But they
did not (24 percent approval: 26 percent of Democrats, 17 percent of
Republicans, and 25 percent of independents). With the seventh, a major-
ity (64 percent) disapproved of a federal requirement that airlines charge
people, with their airline tickets, a specific amount to offset their carbon
emissions (about ten dollars per ticket), subject to opt out if passengers

16 Here as well we could imagine interesting questions if the default rule tracked reality.
If most people in a city or state are Christians, is it so clearly illegitimate to presume,
for purposes of the census, that most people are Christians, subject to opt out? But with
respect to religion, as with respect to politics, there is a strong social and constitutional
norm in favor of official neutrality, which would be violated even if a particular default
reflected majority preferences and values.
 The United States, 1 : Evidence 17

said that they did not want to pay. Interestingly, a strong majority of
Democrats (57 percent) disapproved of this approach, although the num-
ber for Republicans was significantly higher (75 percent).
The five other unpopular nudges involved information and education
(see Table 2.6). With the first (and most extreme), a newly elected presi-
dent adopted a public education campaign designed to convince people
that criticism of his decisions is unpatriotic and potentially damaging to
national security. There was overwhelming disapproval of this campaign
(23 percent approval: 24 percent of Democrats, 21 percent of Republicans,
and 22 percent of independents). What is perhaps most noteworthy here
is not majority disapproval, but the fact that over one-fifth of Americans,
on essentially a nonpartisan basis, were in favor of this most unusual public
campaign.
With the second, the federal government adopted a public educa-
tion campaign designed to convince mothers to stay home to take care
of their young children. Over two-thirds of respondents rejected this
nudge (33 percent approval: 33 percent of Democrats, 31 percent of
Republicans, and 34 percent of independents). The third involved a gov-
ernment requirement that movie theaters run subliminal advertisements
to discourage smoking and overeating. Here too, there was majority dis-
approval (41 ­percent approval: 47 percent of Democrats, 42 percent of
Republicans, and 35 percent of independents). It is noteworthy and sur-
prising, however, that over two-fifths of people actually supported this
requirement.

Table 2.6  Unpopular education campaigns and disclosure17

Unpatriotic Stay-at-home Subliminal Mandatory Transgender

criticism mothers advertising manufacturing label:
Communism
Total 23/77 33/67 41/59 44/56 41/59
support (in
percentages)
Democrats 24/76 33/67 47/53 47/53 49/51
Independents 22/78 34/67 35/65 42/58 38/62
Republicans 21/79 31/69 42/58 43/57 29/71

17 Percentages may not total 100 due to rounding.

18 The U nited States , 1: Evidence

With the fourth, the federal government would require all products that
come from a Communist country (such as China or Cuba) to be sold with
the label “Made in whole or in part under Communism.” Slightly over
half of respondents disapproved of this requirement (44 percent approval:
47  percent of Democrats, 43 percent of Republicans, and 42 percent of
independents). With the fifth, a majority (59 percent) also rejected a pub-
lic education campaign from the federal government, informing people
that it is possible for people to change their gender from male to female or
from female to male and encouraging people to consider that possibility “if
that is really what they want to do.” There is yet another surprise here: this
somewhat adventurous campaign was endorsed by 41 percent of respond-
ents; note that approval rates differed between Democrats (49 percent),
Republicans (29 percent), and independents (38 percent).
 3
The United States, 2:
principles

What separates the approved nudges from the rejected ones? Two ­principles
seem to dominate the cases. First, Americans reject nudges that they take to have
illegitimate goals. In a self-governing society, for example, it is illegitimate
to attempt to convince people that criticism of a public official is unpatri-
otic. At least in the United States, nudges that favor a particular religion or
political party will meet with widespread disapproval, even among people
of that very religion or party.1 This simple principle justifies a prediction:
Whenever people think that the motivations of the choice architect are
illicit, they will disapprove of the nudge.

1 We could, of course, imagine a nation in which favoritism on the basis of religion or

party would attract widespread support and might be seen as analogous to a default
rule in which women’s last name changes to that of their husband (which is approved,
it will be recalled, by a majority of respondents here). In such a nation, a default rule
in favor of the most popular party, or the dominant religion, might be taken to track
people’s preferences and values, and not to be a violation of the governing conception
of ­neutrality at all.
20 The U nited States ,  2 : principles

To be sure, that prediction might not seem terribly surprising, but it

suggests an important point, which is that people will not oppose (for
example) default rules and warnings as such; everything will turn on what
they are nudging people toward. By contrast, mandates do run into some
opposition simply because they are mandates. When there are partisan
differences in judgments about nudges, it is often because of partisan disa-
greement about whether the relevant motivations are legitimate. Resolution
of such disagreements would of course depend on judgments having noth-
ing to do with nudging as such.
Second, Americans oppose nudges that they perceive as inconsistent with the
interests or values of most choosers. The most direct evidence is the finding that
while most Americans support automatic name change for women, they
reject automatic name change for men. The evident reason is that the for-
mer tracks people’s interests and values (at least in general), while the
latter countermands them.2 Any default rule, of course, is likely to harm
at least some people; some people will want, for good reason, to opt out,
and some people who want to opt out will not do so, perhaps because of
inertia and procrastination. This point is a potential objection to default
rules in general.
By itself, however, that fact is not enough to produce public opprobrium.
Recall that there is majority approval for automatic voter registration and
automatic enrollment in pension plans and green energy, apparently because
respondents think that those nudges are in most people’s interests.3 Recall too
that most respondents are favorably disposed toward public e­ ducation cam-
paigns designed to combat obesity and discrimination on the basis of sexual
orientation. By contrast, most people oppose public e­ ducation ­campaigns

2 Here as well, we could easily imagine a population that would reverse these results.
­Suppose that one believes that automatically assuming that wives take their husbands’
last names undermines sex equality, and automatically assuming that husbands take their
wives’ last names promotes sex equality. For those who have these beliefs, and are com-
mitted to sex equality, reversing the majority’s views might seem attractive.
3 Note, however, that savings defaults are importantly different from green defaults. The
former are adopted because they are in the interest of choosers; money that would go
to take-home pay goes into savings, and so choosers do not lose anything on net (while
also saving for retirement). The latter are adopted because they help to solve a collective
action problem. With respect to green defaults, the question did not specify whether
people would have to pay for green energy. Not surprisingly, people are more likely to
opt out if they would. See Simon Hedlin and Cass R. Sunstein, Does Active Choosing Promote
Green Energy Use? Experimental Evidence (Mossavar-Rahmani Ctr. for Bus. and Gov’t Reg. Pol’y
Program, Working Paper RPP-2015-13, 2015).
 The U nited S tates,  2 : principles 21

to encourage women to stay at home and to inform people that they can
change their gender, apparently on the ground that those campaigns are
inconsistent with what people regard as prevailing interests and values.4
When people are deciding whether to favor default rules, the size of the
group of disadvantaged people undoubtedly matters. If a default rule harms
a majority, it is unlikely to have much appeal. If the disadvantaged group
is large (but not a majority), people might reject a default rule and favor
active choosing instead. The precise nature of this principle remains to be
tested, but most respondents appear to accept an important third principle:
Before certain losses can occur, people must affirmatively express their wishes. The principle
forbids the state from taking certain goods by default.5
It is relevant here that most respondents favor a state requirement that
when obtaining their driver’s license, people indicate whether they want
to be organ donors (and thus favor active choosing), even though most
Americans reject a default rule in favor of being an organ donor. The appar-
ent idea involves the central importance of individual consent. Without
that consent, the government may not take things that people currently
have. The boundaries of this principle remain to be specified. People are
willing to approve of automatic enrollment if it will protect their future
selves (as in the case of pensions) and also if it will protect the environ-
ment. Most people do not oppose the tax system as such. But “takings”
seem to raise a red flag.

4 To be sure, there is an ambiguity in these findings. Do respondents reject nudges that

are (a) inconsistent with their own interests or values or (b) inconsistent with the inter-
ests or values of most choosers? On this question, the findings here do not provide a clear
test. When respondents reject nudges, they probably believe that the nudges that are
inconsistent with their own interests or values are also inconsistent with the interests
or values of most choosers. It would be interesting and possible to pose questions that
would enable us to choose between (a) and (b). Consider here the important finding
that when a nudge is said to be targeted at “you,” people are less likely to support it than
when it is said to be targeted at “people in general.” James F. M. Cornwell and David H.
Krantz, Public Policy for Thee, But Not for Me: Varying the Grammatical Person of Public
Policy Justifications Influences Their Support, Judgement and Decision Making 5, 433 (2014).
Our own study implicitly assumes the “people in general” frame.
5 Whether this principle is triggered will depend on a theory of entitlement, from which
any account of “losses” will flow. In the questions here, that issue is not especially
complicated. If a default rule will ensure that people give money to specified charities
(subject to opt out), it will impose a loss. But we could imagine harder cases—as, for
example, with adjustments in the social security program where losses and gains might
not be self-evident and might be subject to framing.
22 The U nited States ,  2 : principles

Note in this regard that strong majorities of people reject automatic char-
itable donations of diverse kinds. The apparent concern is that, as a result of
inertia, procrastination, or inattention, people might find themselves ­giving
money to a charity even though they do not wish to do so. We might
therefore complement the third principle with a fourth and narrower one,
which can be seen as a specification: Most people reject automatic enrollment in chari-
table giving programs, at least if they are operated by public institutions. Though it does
not involve money, the case of carbon offsets can be understood in similar
terms; while it does not involve a charitable donation and instead might be
seen as an effort to prevent a harmful act, Americans appear to want active
consent. As noted, we do not yet know the exact limits of apparent public
skepticism about default rules that would give away people’s money with-
out their active consent, but there is no doubt that such skepticism exists.
We have seen that people generally favor disclosures that, in their view,
bear on health, safety, or the environment (salt content, GMOs). At the
same time, the results leave open the question whether and when p ­ eople
will favor mandatory disclosures that involve political issues associated
with production of a product rather than the health and ­environmental
effects of the product itself. Americans seem closely divided on that
­question. With repeated violations of the nation’s labor laws, and nations
that ­harbor ­terrorism, such disclosure achieved majority support—but not
with ­products coming from Communist nations. Americans might well
demand a certain threshold of egregiousness, in terms of the behavior of
those who produce a good or service, before they will want to require
disclosure of that behavior. On this question, partisan differences are to be
expected because people will disagree about whether the relevant threshold
has been met, and about what it exactly is.
It is tempting, and not inconsistent with the data, to suggest that r­ eactions
to nudges also show the influence of a fifth principle: Americans reject
nudges that they regard as unacceptably manipulative. The subliminal advertising
finding can be taken as support for this principle. But what counts as unac-
ceptable manipulation?6 Most Americans are in favor of graphic ­warning
labels on cigarettes; they like default rules (if consistent with people’s
values and interests); a majority favors a mandatory cafeteria design to pro-
mote healthy eating; people approve of a graphic campaign to discourage

6 See Jung and Mellers, supra note 21, at 66−68 (finding public disapproval of visual ­illusion
designed to promote safety on the highways).
 The U nited S tates,  2 : principles 23

distracted driving; with respect to obesity, a majority favors a somewhat

tendentious public education, one that could plausibly be characterized as
manipulative. No one likes manipulation in the abstract, but there do not
appear to be many cases in which people are willing to reject nudges as
unacceptably manipulative, at least if they have legitimate ends and are taken
to be in the interest of most choosers.

Partisanship
What is the role of partisan differences? Democrats and Republicans will
sometimes disagree, of course, about whether the goals of a particular
nudge are illicit, and they will also disagree, on occasion, about whether
a nudge is consistent with the interests or values of choosers. For example,
those who disapprove of abortion will be especially likely to support nudges
that are designed to discourage abortion; those who do not disapprove of
abortion will be unlikely to support such nudges. Imagine an anti-abortion
nudge in the form of a law requiring pregnant women seeking abortions to
be presented with a fetal heartbeat or a sonogram. We can predict, with a
high degree of confidence, that Democrats would show lower approval rat-
ings than Republicans. Our own study on Amazon’s Mechanical Turk finds
exactly that: only about 28 percent of Democrats approve, while 70 percent
of Republicans do so.7 With respect to a public education campaign inform-
ing people that they can change genders, the significant difference between
Democrats and Republicans should not exactly come as a big surprise.
But there is another and more general division as well. Even when
majorities of Democrats, Republicans, and independents support a particu-
lar initiative, the level of support is sometimes higher within one group
than within another. Even if the underlying end is broadly shared—as it
is, for example, in the area of public health—some subset of Republicans
sometimes seems skeptical of government nudges, taken as such, and will
therefore disapprove of them even if they do accept the legitimacy of the end and do not think
that the nudge is inconsistent with choosers’ interests or values. Some Republicans, and
undoubtedly some Democrats and independents, appear to support another

7 The survey and the results are available from the authors on request. The precise question
asked people whether they approve or disapprove of a “state requirement that pregnant
women must see a sonogram of their fetus, and hear its heartbeat, before proceeding
to have an abortion.” Interestingly, only about one-third of independents approved, es-
sentially the same as Democrats.
24 The U nited States ,  2 : principles

principle: There should be a rebuttable presumption against nudging, at least if the govern-
ment can avoid it.
Most Americans reject this principle, and the survey does not provide
conclusive evidence that significant numbers embrace it, but it is highly
suggestive. Many people reject graphic health warnings on cigarette
packages (26 percent), an educational campaign for childhood obesity
(18 percent), an educational campaign for distracted driving (15 percent),
and a traffic lights system for food (36 percent). It is reasonable to infer
that those who oppose such nudges agree that they have legitimate ends
and are in the interest of most choosers—but nonetheless do not favor
government intervention.
It is important to see that the strength and domain of any anti-nudge
presumption will vary with the particular issue, with partisan affiliations,
and with competing views about the role of government. In some of the
cases, Republicans are more skeptical of nudges than are Democrats. With
calorie labels and childhood obesity campaigns, for example, there are sig-
nificant differences in the levels of support within the two groups, even
though majorities of both are supportive. But in some cases, Republicans
are undoubtedly more enthusiastic about nudges than are Democrats, as
in the case of the anti-abortion nudge. The fact that few such cases are
found here is an artifact of the particular questions. If the issue involved
automatic enrollment in programs by which high-income earners auto-
matically receive capital gains tax benefits, for example, we can predict,
with some confidence, that Republicans would be more supportive than
Democrats. Evidence supports that prediction.8

Nudges vs. mandates

We have suggested that many Americans are skeptical of certain ­mandates,
even if they have legitimate ends. To test that proposition, we used Amazon’s
Mechanical Turk (with 309 participants) to test American reactions to three
pairs of initiatives. The initiatives involved savings (with a 3 percent contri-
bution rate), safe sex education, and education about i­ntelligent design. In
all cases, the nudge was far more popular than the mandate (and received
majority support), and indeed, in all cases, the mandate ran into majority
disapproval. So long as people could opt out, the savings initiative received

8 See Id.
 The U nited S tates,  2 : principles 25

69 percent approval; safe sex education, 77 percent; and intelligent design,

56 percent. As mandates, the three fell to approval rates of 19 percent,
43 percent, and 24 percent respectively.
Consistent with other findings, it follows that most Americans do
oppose certain kinds of mandates as such, even when they are enthusiastic
about the underlying ends and are supportive of nudges that are designed
to promote those ends. We have seen that majorities of Americans have no
general view about nudges as such; their assessments turn on the principles
outlined here. With mandates, many Americans do have a general view,
and it is not favorable. Of course it is also true that Americans do sup-
port mandates of various kinds, especially when harm to others is involved
(as in the case of the criminal law and many regulatory requirements). In
that light, we do not mean to suggest that Americans oppose mandates as
such. That proposition would be too broad. The only point is that they
oppose mandates even when they approve of their goals—and that they do
approve interventions that preserve freedom of choice.

Partisan nudge bias

Do political judgments matter to people’s assessment of nudges? Our s­ urvey
of Americans finds some support for an affirmative answer, in the sense
that Republicans are less likely to approve of certain nudges than Democrats
are. As we have also noted, Republicans are more likely to approve of certain
nudges than Democrats are. Casual observation suggests a broader possi-
bility: When a particular administration uses behaviorally informed tools,
those who are inclined to oppose that administration are not likely to love
those tools. Consider this hypothesis: At least across a wide range, people have no
considered view on nudges as such. Their evaluations turn on whether they approve of the politics
of the particular nudge, or the particular nudges that come to mind.
More specific evidence supports this view. In a series of studies, David
Tannenbaum, Craig Fox, and Todd Rogers have found what they call
“partisan nudge bias.”9 Focusing on policies favoring automatic enroll-
ment in pension plans, they randomly assigned people to conditions in
which they learned that such policies had been implemented by the Bush

9 David Tannenbaum et al., On the Misplaced Politics of Behavioural Policy Interventions,

Nature Human Behaviour 1(0130) (2017), at 1–7 (suggesting that powerful behavioral
­insights can allow for a tendency to gravitate towards default options).
26 The U nited States ,  2 : principles

Administration, the Obama Administration, or an unnamed Administration.

After informing participants about the policy nudge, Tannenbaum et al.
specifically reminded them that defaults could be used “across a wide range
of policies beyond the illustration above” and asked how they felt, setting
the particular application aside, “about actively setting default options as a
general approach to public policy.”10
The basic finding was that on the general question, people were much
influenced by whether Bush or Obama was responsible for the particu-
lar nudge that they read about. When participants were informed that
the ­pension default had been implemented by Obama, liberals tended to
display relative support for the use of defaults as a general policy tool,
whereas conservatives tended to oppose them. But when told that the same
policy had been implemented by Bush, that pattern was eliminated; liberals
displayed relative opposition to the use of defaults, whereas conservatives
supported them.
Tannenbaum et al. also asked respondents about a series of nudges
that had an identifiable political valence, immediately triggering dispa-
rate reactions from liberals and conservatives. These included increasing
participation by low-income individuals in existing food stamp and supple-
mental nutrition assistance programs (liberal valence); increasing claims by
high-income individuals for existing capital gains tax breaks (conservative
valence); increasing participation in safe sex and effective contracep-
tion use educational programs for high-school children (liberal valence);
increasing participation in intelligent design educational programs for
high-school children (conservative valence); and a generic, context-free
policy illustration (no valence). There were five different types of policy
nudges: (1) automatic enrollment defaults, (2) implementation intentions,
(3) public commitments, (4) highlighting losses, and (5) descriptive social
norms. As in their first study, Tannenbaum et al. asked people about their
general views about nudges after seeing the relevant example. Participants
were specifically reminded that the approach was general and could be
used across a wide range of policies.
The result was unambiguous: People are significantly more likely to
approve of nudges in general when they favor the particular political objec-
tives used to illustrate them. When the nudges were applied to traditionally
liberal policies (food stamps, safe sex), liberals were relatively ­supportive

10 Id.
 The U nited S tates ,  2 : principles 27

of nudges as policy tools, while conservatives were relatively opposed to

their general use. This pattern reversed when those same nudges were
applied to traditionally conservative policy goals (capital gains programs,
intelligent design education programs).
Interestingly, and importantly, when nudges were attached to a generic
policy objective, there was no association between political orientation and
people’s evaluation of nudges; apparently, conservatives and liberals do
not disagree on the general question. A particularly striking finding is that
while libertarians were less likely to approve of nudges than those without
libertarian dispositions, attitudes about particular policies turned out to be a
far more significant predictor than attitudes about libertarianism in general.
Tannenbaum et al. used the same basic strategy to test the responses of
actual policymakers, consisting of US city mayors and high-level ­public
servants in state and local governments. They asked the participants to read
about two kinds of automatic enrollment defaults. Half read a scenario
in which low-income earners were automatically defaulted to receive
supplemental food assistance benefits, and half read a scenario in which
high-income earners were automatically defaulted to receive capital gains
tax benefits. Policymakers were explicitly reminded that the task was the
evaluation of nudges as general-purpose policy tools. The usual pattern
held: The overall assessments of policymakers were greatly affected by the
political valence of the examples.
In sum, “people find nudges more ethically problematic when they are
applied to policy objectives they oppose, or when applied by policymakers
they oppose, while they find the same nudges more acceptable when they
are applied to political objectives they support or by policymakers they
support.”11 It would not of course be surprising to find that people favor
nudges that support their own goals and reject nudges that undermine
those goals. What is more interesting is that many people seem not to have
strong or firm judgments about nudges, taken simply as such. Particular
examples drive their general views—perhaps because the examples cre-
ate some kind of affective reaction to the broad category, perhaps because
the examples are taken to convey information about how nudges would
­actually be used (which should of course bear on the overall evaluation).
In this respect, people use the examples as heuristics, or mental shortcuts,
in answering the broader and more difficult question.

11 Id. at 1.
28 The U nited States ,  2 : principles

There is a clear implication here for the political economy of n ­ udging:

Citizens’ judgments about the ethics of nudging, and even the general
enterprise, are likely to be, in significant part, an artifact of their ­substantive
judgments about the specific directions in which they think people are
likely to be nudged. It is noteworthy that in the United Kingdom, nudging
has been prominently associated with the Conservative Party (and Prime
Minister David Cameron), which has likely reduced concern from the right
(and perhaps heightened concern from the left). This point should not be
taken too far. As we have seen, even those who strongly support an incum-
bent president would be likely to object strenuously if he imposed a nudge
that entrenched himself (as, for example, through a system of default
­voting). In egregious cases of self-dealing, or of violations of widely held
social norms, citizens of a free society (or even an unfree one) might well
be outraged whatever they think of the underlying substance. But within
certain limits, political assessments are likely to reflect political judgments.
We conclude with two general points. First, most Americans are sup-
portive of nudges of the kind that democratic societies have adopted or
seriously considered in the recent past. Second, that support diminishes
when people distrust the motivations of the choice architects, or when
they fear that because of inertia and inattention, citizens might end up with
outcomes that are inconsistent with their values or their interests. In par-
ticular, Americans object to situations in which choice architects produce
outcomes by which people lose money, or other things of importance,
without their explicit consent.
 4
Europe

In many ways, the United States has a distinctive culture. Both scholars
and politicians have explored the idea of “American exceptionalism.”1
According to a conventional account, Americans have a strong concep-
tion of freedom—perhaps a result of the absence of any kind of feudal
past—which leads them to unusually high levels of distrust of government.
With respect to nudging, we have not found high levels of distrust of gov-
ernment. But it might well be expected that European nations, with their
different cultures and traditions, would show markedly different kinds of
reactions and perhaps greater receptivity.
We report here the results of surveys in six nations in Europe: Denmark,
France, Germany, Hungary, Italy, and the United Kingdom.2 Our choice of
those nations was of course selective; many European nations are missing
(and we will introduce more in later chapters). But there was a method to our

1 See, e.g., Seymour Martin Lipset, American Exceptionalism (1997).

2 An earlier version of this chapter has been published as: Lucia A. Reisch and Cass
R. ­Sunstein (2016), Do Europeans Like Nudges? JJDM 11(4), 310–25. See also Caezilia
Loibl, Cass R. Sunstein, Julius Rauber and Lucia A. Reisch (2018), Which Europeans like
nudges? Approval and Controversy in Four European Countries. Journal of Consumer Affairs,
52(3), 655–688 (for a deeper analysis of the same data set).
30 Europe

madness. We sought to represent different cultural and geographic regions of

Europe as well as different socio-economic regimes and political traditions:

•• a Nordic welfare state, which might be thought to be especially recep-

tive to nudging (Denmark);
•• a social market economy often thought to have a deep, historically
grounded distrust of paternalism, growing out of the Cold War and
Stasi experience (Germany);
•• a Central European post-socialist country (Hungary);
•• two Southern European countries with different political regimes,
problems, strengths, and experiences (France and Italy); and
•• the United Kingdom, which has helped to spearhead nudging as a pol-
icy tool worldwide since 2010, and hence had several years of debate
on that topic.

Our major findings are simple and (we think) surprising. As in the United
States, so too in the nations explored here: If people believe that a nudge
has legitimate goals, and think that it fits with the interests or values of
most people, they are overwhelmingly likely to favor it. With only modest
qualifications, there is broad support, throughout the six nations, for 12
of the 15 nudges that we tested—and broad opposition, throughout those
nations, to the remaining three nudges. In that respect, we find a substan-
tial consensus among disparate nations. (The Bill of Rights for Nudging,
discussed in Chapter 9, is meant to apply in all nations.)
Two of the three rejected nudges run afoul of a principle, noted in
Chapter 1, on which there is apparently a European consensus: The gov-
ernment should not take people’s money without their affirmative consent,
even for a good cause. With respect to both charitable donations and car-
bon offsets, majorities believe default rules to be unacceptable because they
offend that principle. We suspect that this finding reflects a broadly held
commitment to the idea that by default, people are entitled to keep their
own resources; without a clear statement of their own intentions, those
resources should remain theirs. There is an evident connection between
this finding and the well-known behavioral phenomenon of loss aversion.
Like Americans, people in the covered European countries also reject a
nudge that is unambiguously manipulative: a subliminal advertising cam-
paign in movie theaters, designed to convince people not to smoke and
overeat. Subliminal advertising can be seen as a defining example of manip-
ulation, because it appeals to people’s unconscious processing.
 Europe 31

Notwithstanding the general consensus, we find a noteworthy division

among nations. While majorities in both Denmark and Hungary are sup-
portive of many nudges, citizens of those nations show significantly3 lower
levels of receptivity to them than do citizens of France, Germany, Italy,
and the United Kingdom. Interestingly, however, we do not find, within
European nations, consistent and clear associations between party affilia-
tions and approval or disapproval of nudges.

The study

Sampling and survey

We used nationally representative online surveys in six European nations:
Denmark, France, Germany, Hungary, Italy, with about 1,000 respondents
each, and the United Kingdom, with about 2,000 respondents. Because the
respective nation’s online population nearly equals the full population in all six
countries, and because a stratified sample was used, we can assume almost full
representativeness of the surveys. At the same time, we are aware of the ­limitation
that online representativeness does not fully equal ad hoc ­representativeness.
The survey questionnaire built on that used in Chapter 1. To adjust to the
European setting (some of the US nudges are already imposed in Europe) and
also to be able to attain a representative sample in six countries, the number
of items was reduced to 15. We picked 13 from the US survey and added two
additional interventions that had been recently discussed in European poli-
tics: (1) requiring supermarket chains to keep cashiers free of sweets (Nudge
14) and (2) requiring canteens in public institutions to have one meat-free
day per week (Nudge 15, which, admittedly, is not quite a nudge because
it does not preserve freedom of choice; we call it Nudge 15 only for pur-
poses of exposition). The selection covered a wide range of types of nudges:
educative nudges, such as information campaigns, and defaults (i.e., differ-
ent levels of intrusion); noneducative nudges targeting automatic System 1
and educative nudges targeting deliberative System 2 (Chapter 7 focuses on
these); nudges covering different areas such as health/food, energy/climate,
sustainability, organ donation, and online contracts (see Table 4.1).
We emphasize that as in the US survey, these interventions were delib-
erately skeletal—for example, we did not identify them with any particular

3 We conducted chi-square tests showing significant differences between the two groups
of countries for 14 out of 15 nudges.
Table 4.1  The 15 items of the survey

1. The federal government requires calorie labels at chain restaurants (such as

McDonald’s and Burger King).
2. The federal government requires a “traffic lights” system for food, by which
healthy foods would be sold with a small green label, unhealthy foods with
a small red label, and foods that are neither especially healthy nor especially
unhealthy with a small yellow label.
3. The federal government encourages (without requiring) electricity providers
to adopt a system in which consumers would be automatically enrolled in
a “green” (environmentally friendly) energy supplier, but could opt out if
they wished.
4. A state law requiring people to say, when they obtain their driver’s license,
whether they want to be organ donors.
5. A state law requires all large grocery stores to place their most healthy foods
in a prominent, visible location.
6. To reduce deaths and injuries associated with distracted driving, the national
government adopts a public education campaign, consisting of vivid and
sometimes graphic stories and images, designed to discourage people from
texting, emailing, or talking on their cellphones while driving.
7. To reduce childhood obesity, the national government adopts a public
education campaign, consisting of information that parents can use to make
healthier choices for their children.
8. The federal government requires movie theaters to provide subliminal
advertisements (i.e., advertisements that go by so quickly that people are not
consciously aware of them) designed to discourage people from smoking
and overeating.
9. The federal government requires airlines to charge people, with their airline
tickets, a specific amount to offset their carbon emissions (about 10 Euro per
ticket); under the program, people can opt out of the payment if they explic-
itly say that they do not want to pay it.
10. The federal government requires labels on products that have unusually high
levels of salt, as in, “This product has been found to contain unusually high
levels of salt, which may be harmful to your health.”
11. The federal government assumes, on tax returns, that people want to
donate 50 Euro to the Red Cross (or to another good cause), subject
to opt out if people explicitly say that they do not want to make that
donation.
12. The federal government requires movie theaters to run public edu-
cation messages designed to discourage people from smoking and
overeating.
 Europe 33

13. The federal government requires large electricity providers to adopt a system
in which consumers would be automatically enrolled with a “green” (envi-
ronmentally friendly) energy supplier, but could opt out if they wished.
14. To halt the rising obesity problem, the federal government requires large
supermarket chains to keep cashier areas free of sweets.
15. For reasons of public health and climate protection, the federal government
requires canteens in public institutions (schools, public administrations, and
similar) to have one meat-free day per week.

source (e.g., a leader or a party), and we did not specify the process from
which the nudges emerged (e.g., with or without public support). While
it would be valuable to test whether and to what extent such characteristics
affect people’s judgments, our goal here was to examine those judgments
without any knowledge of them.
The questionnaire was fully structured and questions were rand-
omized. Respondents were required to follow the questions in the given
order and wording. Each item was shown on a single screen. Respondents
were asked: “Do you approve or disapprove of the following hypo-
thetical policy?” The two possible answers were displayed in a column
(“approve” first, “disapprove” second). The English version was taken as
a reference point for the translations and re-translations into the respec-
tive languages. In the Danish and Hungarian questionnaire as well as the
one for the UK, the currencies were adapted: Euros were replaced by
the equivalent amount in Danish kroners, Hungarian forints and British
pounds respectively. Details of both sampling and survey can be found in
the appendix of this chapter.
The surveys were copy-tested and run by the ISO-certified market
research organization GFK (Gesellschaft für Konsumforschung) during the
first two weeks in September 2015. This was just before European coun-
tries were struck by the so-called “refugee crisis” that has had (and still has
to date) an immense impact on the public’s views on politics and govern-
ment policies in Europe, shifting political attitudes markedly to populistic
parties on the political right.

The fifteen interventions

An overview of the assessment of all nudges in all countries is provided in
Table 4.2. With some exceptions (notably, subliminal advertising which
Table 4.2  Overview on approval rates for the 15 nudges in the six surveyed
countries

IT UK FR DE HU DK
  1 Requiring calorie labels in chain 86 85 85 84 74 63
restaurants
  2 Requiring traffic-light labels signaling 77 86 74 79 62 52
healthiness of food
  3 Encouraging defaulting customers into 76 65 61 69 72 63
green energy providers
  4 Law requiring active choice regard- 72 71 62 49 54 62
ing organ donation on obtaining the
driver’s license
  5 Law requiring supportive choice 78 74 85 63 59 48
architecture for healthy food in large
grocery stores
  6 Public education campaign with vivid 87 88 86 82 76 81
pictures against distracted driving
  7 Public education campaign for parents 89 88 89 90 82 82
promoting healthier food for their chil-
dren to fight childhood obesity
  8 Requiring subliminal advertising in 54 49 40 42 37 25
movie theatres against smoking and
overeating
  9 Requiring airlines to charge their 40 46 34 43 18 35
customers a carbon emission
­compensation fee
10 Requiring industry to put warning 83 88 90 73 69 69
labels on food with high salt content
11 Default citizens to donate 50 Euro for 48 25 29 23 37 14
the Red Cross on a tax return
12 Requiring movie theatres to run 77 67 66 63 40 35
­information campaigns against
­smoking and overeating
13 Requiring energy providers to default 74 65 57 67 65 55
customers into green energy
14 Requiring sweet-free cashier zones in 54 82 75 69 44 57
supermarkets
15 Requiring one meat-free day in pub- 72 52 62 55 46 30
lic canteens

Note: Total support in percentages; unweighted results.

 Europe 35

does not qualify as a “nudge,” mandated greenhouse gas compensation in

air travel, default donation on a tax return, and, to a lesser degree, man-
dated information campaigns against smoking and overeating as well as
meat-free days), we find majority approval in all countries.
For further analyses and exposition here, we categorize the interventions
in terms of increasing intrusiveness, resulting in five groups, calculated as
mean approval in percentages:

1) purely government campaigns to educate people about childhood obesity,

distracted driving, and smoking and overeating;
2) mandatory information nudges, imposed by government on the private sec-
tor, requiring disclosure of nutritional value and health risks of food
(calorie labels, high levels of salt, nutritional traffic lights);
3) mandatory default rules, imposed by government on the private sector, involv-
ing green energy provision, carbon emissions charges, and donations to
the Red Cross, along with mandatory choice architecture for retailers to
support healthy foods, and mandatory active choice on organ donation;
4) mandatory subliminal advertising, imposed by government on movie
­theaters, to discourage people from smoking and overeating;
5) mandatory choice architecture involving supermarkets, a nudge for c­ onsumers
(sweet-free cashier zones) and also choice editing that goes beyond mere
nudging (meat-free days in public cafeterias).

Socio-demographic variables and political preference

A number of socio-demographic variables were collected in the six coun-
tries. Due to the limited comparability across countries of most of those
variables, we report on only two that are robust: age and gender.
We also report on political preference, measured by asking for whom the
respondent voted in the last national elections (“When you think about the
last national election, which party did you vote for?”). On the basis of par-
liamentary groups represented in the European Parliament as well as expert
advice, we grouped the political parties into six clusters: conservative/
Christian democratic; left wing/socialist/communist; liberal; green; popu-
list; and “other.” While it was obvious that this instrument is rather rough
and quite difficult to apply for some countries and parties, we assumed that
if they exist, distinct partisan differences would be traceable. Details can be
found in Tables A4.2 and A4.3 in the appendix of this chapter.
36 Europe

Statistical analysis
In a first step, we focused on the main results of the analysis of the
­frequencies regarding approval/disapproval for individual nudges by
country. Approval rates are presented in Figures 4.1–4.5. In a next step, we
checked for significant differences in approval rates depending on socio-
demographic variables and political preferences within countries. As the
data has a nested structure, we ran a multilevel regression analysis with
the specification of a two-level random intercept model where the first
level is country and the second is the individual respondent. In s­ amples
such as ours, individual observations are generally not i­ndependent, as
individuals within one country tend to be more similar to each other than
across countries.
We estimated the multilevel regression for each level of intrusion (from
weak to excessive), with the approval rates being the dependent variables.
For this, we calculated the mean approval in percentages by the level of
intrusiveness. As outlined above, we categorize the 15 nudges in terms of
increasing intrusiveness, resulting in five groups. As independent variables
we use age, gender, and political attitude on the individual level, and coun-
try on the country level.

Results

Information: Government campaigns

We tested three nudges that seem minimally intrusive, in the sense that
they involve the mere provision of information by the government. The
nudges involved (1) public education campaigns to reduce childhood obe-
sity, (2) similar campaigns to reduce deaths and injuries from distracted
driving, and (3) similar campaigns, in movie theaters, to discourage peo-
ple from smoking and overeating.
Over all countries, the average approval rate for all three nudges is
76.9 percent. In all six nations, both (1) and (2) received overwhelm-
ing support (see Figure 4.1). We expected (3) to be more controversial,
and it was. It did receive majority support in Italy, the United Kingdom,
France, and Germany, but the levels were lower than for (1) and (2), and
in Denmark and Hungary, majorities disapproved (significant difference
between the two groups of countries confirmed).
 Nudge: Childhood obesity (1)
100.0
90.0
89.1 87.6 88.9 90.3
Approval in percentages 80.0
81.6 82.3
70.0
60.0
50.0
40.0
30.0
20.0
10.0
0.0
Italy UK France Germany Hungary Denmark
Nudge approval 95%−CI

Nudge: Distracted driving (2)

100.0
90.0
80.0 87.2 88.4 86.0
Approval in percentages

81.8 80.9
70.0 76.4
60.0
50.0
40.0
30.0
20.0
10.0
0.0
Italy UK France Germany Hungary Denmark

Nudge approval 95%−CI

Nudge: Smoking & overeating (3)

100.0
90.0
80.0
Approval in percentages

70.0 77.0
60.0 67.5 65.9
63.0
50.0
40.0
40.1
30.0 35.4
20.0
10.0
0.0
Italy UK France Germany Hungary Denmark

Nudge approval 95%−CI

Figure 4.1  Bar charts for information nudges: Government campaigns, total support
in % (unweighted)
 Nudge: Calorie labels (1)
100.0
90.0
Approval in percentages 80.0 86.1 85.1 85.4 84.0
70.0 73.8
60.0
63.4
50.0
40.0
30.0
20.0
10.0
0.0
Italy UK France Germany Hungary Denmark

Nudge approval 95%−CI

Nudge: High levels of salt (2)
100.0
90.0
87.5 90.1
80.0
Approval in percentages

82.9
70.0
73.0
60.0 69.2 69.1

50.0
40.0
30.0
20.0
10.0
0.0
Italy UK France Germany Hungary Denmark

Nudge approval 95%−CI

Nudge: Traffic lights (3)

100.0
90.0
80.0 85.6
Approval in percentages

70.0 77.4 78.9

73.8
60.0
61.7
50.0
51.9
40.0
30.0
20.0
10.0
0.0
Italy UK France Germany Hungary Denmark

Nudge approval 95%−CI

Figure 4.2  Bar charts for information nudges, governmentally mandated, total
support in % (unweighted)
 Europe 39

Nudge: Encouraging green energy (1) Nudge: Mandatory green energy (2)
100.0 100.0
90.0 90.0
80.0 80.0
Approval in percentages

Approval in percentages
70.0 76.1 70.0
72.0 73.6
60.0 69.0
64.9 62.9 60.0 64.8 67.1 64.9
61.4
50.0 50.0 57.3 55.3
40.0 40.0
30.0 30.0
20.0 20.0
10.0 10.0
0.0 0.0
Italy UK France Germany Hungary Denmark Italy UK France Germany Hungary Denmark
Nudge approval 95%−CI Nudge approval 95%−CI
Nudge: Carbon emission charge (3) Nudge: Red Cross (4)
100.0 100.0
90.0 90.0
80.0 80.0
Approval in percentages

70.0 Approval in percentages 70.0

60.0 60.0
50.0 50.0
40.0 45.5 40.0 48.0
43.0
40.1
30.0 34.6 30.0 37.4
34.1
20.0 20.0 28.8
24.6 23.2
10.0 18.3 10.0 13.9
0.0 0.0
Italy UK France Germany Hungary Denmark Italy UK France Germany Hungary Denmark

Nudge approval 95%−CI Nudge approval 95%−CI

Nudge: Healthy food placement (5) Nudge: Organ donor choice (6)
100.0 100.0
90.0 90.0
80.0 80.0
84.8
Approval in percentages

Approval in percentages

70.0 77.6 70.0

73.5 72.0 70.7
60.0 60.0
62.5 61.9 61.8
50.0 59.4 50.0
53.6
40.0 48.3 40.0 49.2

30.0 30.0
20.0 20.0
10.0 10.0
0.0 0.0
Italy UK France Germany Hungary Denmark Italy UK France Germany Hungary Denmark

Nudge approval 95%−CI Nudge approval 95%−CI

Figure 4.3  Bar charts for default rules, total support in % (unweighted)

Information: Governmentally mandated nudges

We tested three informational nudges that took the form of mandates
on the private sector, designed to promote healthy eating: (1) calorie
labels, (2) salt labels (for products with particularly high levels), and
(3) a “traffic lights” system for more or less healthy food. Because such
nudges require action by private institutions (companies), they might
seem more intrusive than educational campaigns by the government
itself. But all three obtained majority support, with an average approval
of 78.0 percent across all six nations (see Figure 4.2). The most note-
worthy division here is again between Denmark and Hungary on the
40 Europe

Nudge: Subliminal ads

100.0
90.0
80.0
Approval in percentages

70.0
60.0
50.0
54.0
40.0 49.1
40.4 42.2
30.0 37.3
20.0 25.3
10.0
0.0
Italy UK France Germany Hungary Denmark

Nudge approval 95%−CI

Figure 4.4  Bar chart for subliminal ads, total support in % (unweighted)

one hand and Italy, the United Kingdom, France, and Germany on the
other. The first two showed significantly lower levels of support, though
majorities approved.

Default rules
Default rules are often the most prominent and effective nudges. We asked
respondents about five potentially controversial kinds: (1) government
encouragement (without a mandate) of automatic enrollment in green
energy; (2) governmentally mandated green energy defaults; (3) defaulting
air travelers into the payment of carbon offsets; (4) defaulting taxpayers
into a 50 Euro (or equivalent) payment to the Red Cross; (5) requiring large
grocery stores to place healthy foods in a prominent, accessible location. We
also asked respondents about (6) requiring people to say, when they receive
a driver’s license, whether they wanted to be organ donors. Active choosing
is not a default rule, but because it is a form of choice architecture designed
to elicit people’s preferences, we group it with default rules here.
On average, 54.8 percent approved default rules across the six coun-
tries. In all nations, (1) and (2) received strong majority support (see
Figure 4.3). Majorities in all nations except Denmark favored (5).
 E urope 41

Nudge: Sweet-free cashier zone (1)

100.0
90.0
80.0
Approval in percentages

81.8
70.0 75.0
69.4
60.0
57.2
50.0 54.2
40.0 44.4

30.0
20.0
10.0
0.0
Italy UK France Germany Hungary Denmark

Nudge approval 95%−CI

Nudge: Meat-free day (2)

100.0
90.0
80.0
Approval in percentages

70.0 72.4
60.0 62.2
50.0 54.9
51.7
40.0 46.2

30.0
30.1
20.0
10.0
0.0
Italy UK France Germany Hungary Denmark

Nudge approval 95%−CI

Figure 4.5  Bar charts for other mandates

In all nations, both (3) and (4) were rejected by substantial majorities
(see Figure 4.3), which helps account for the relatively small margin of
majority support for all interventions in this category. Interestingly, the
nudge “encouragement of green energy” (1) is the only one without a
significant difference between the two groups of countries. There was
majority approval of (6) in all countries, with the interesting exception of
Germany.
42 Europe

Manipulation: Subliminal advertising

Finally, we asked respondents about an intervention that might be expected
to be widely rejected as a defining example of manipulation: compulsory
subliminal advertising in movie theaters, designed to discourage smoking
and overeating. And indeed, it was rejected, but not exactly widely, with an
average approval rate of 42.5 percent, and with the puzzling qualification
that in Italy and the United Kingdom we find majority or near-majority
support (see Figure 4.4).

Other mandates
Requiring (1) sweet-free cashier areas and (2) meat-free days in cafeterias
in public institutions are relatively strong government interventions. Both
ideas have produced controversy in European politics. To our knowledge,
while retailers and canteens in some countries increasingly experiment
with such choice architecture, they have not been tested in representa-
tive European surveys before. Sweet-free cashier areas can be regarded as a
nudge for consumers; meat-free days go far beyond a nudge.
The average approval rate across countries is 59.6 percent. Results in
Figure 4.5 show approval for sweet-free cashier zones in supermarkets by
majorities in all countries, except for Hungary. Somewhat surprisingly,
even a meat-free day in cafeterias in public institutions is approved by
majorities, except for Hungary and Denmark.

Approval rates: Sex, age, and politics

We found broad support for most of the 15 nudges that we tested,
­notwithstanding some striking differences across the six countries
(as shown in Figures 4.1–4.5). At the same time, we explored whether there
might be differences in approval across demographic categories and among
groups with different political preferences within countries or groups of
­countries. We ran a multilevel regression analysis which ­provided clear
results (­presented in Table A4.4 in the appendix).
The basic picture is that except for gender (females are slightly more pos-
itive), socio-economic characteristics do not significantly influence people’s
attitude towards the nudges in the six countries. We do see a tendency for
older respondents to be more in favor of information nudges and defaults,
 E urope 43

but the effect is not a strong one and not the same in all six countries. Yet
again, our results suggest that it is the aim that the government wants to
achieve with the nudge that determines approval. But as the different results
in Denmark and Hungary show, country differences can matter a great deal.

The overall pattern

The best explanation for the overall pattern of results is straightforward. As in
the United States, so too in Europe: When Europeans believe that a nudge has
legitimate purposes and is consistent with the interests or values of most people,
majorities are likely to support it. At least if nudges are presented in the simple
form used here, there is no opposition to nudging as such, even if it takes the
form of default rules or other arguably aggressive forms of choice architecture.
It would not be unreasonable to speculate that people might have some
kind of informal hierarchy in mind, corresponding to their intuitions
about intrusiveness—with, perhaps, government educational campaigns
being the weakest kind of nudge, and default rules the strongest, while
mandatory information disclosure from the private sector might be found
in the middle. But our results suggest that any informal hierarchy—even
if it exists—is not the principal driver of people’s judgments. What most
matters is what the nudge is trying to achieve. Most of the nudges we tested
were designed to promote health, safety, and clean energy, and people
generally approve of them, because they endorse those goals.
Importantly, our survey did not provide people with information about
benefits and costs, and their responses probably reflect intuitive (and
potentially inaccurate) judgments about likely consequences. Suppose,
for example, that people were informed that a certain educational cam-
paign was expensive to implement and would have little or no effect. If
so, people would be unlikely to support it. That is of course an extreme
case. It would be interesting to test whether the high levels of support
would increase with favorable benefit-cost ratios and if they would fall
with less favorable ones. In Chapter 7, we shall offer some evidence on
that question. Our claim about high levels of European support for nudg-
ing depends, of course, on how Europeans respond without being given
relevant information. In our view, it is relevant and important to find high
levels of receptivity to identifiable policy initiatives in the abstract, not
least because people’s judgments will inevitably be affected by their own
priors about effectiveness.
44 Europe

One of our most noteworthy findings is that most Europeans, like most
Americans, reject nudges that take people’s money without their affirma-
tive consent, even if the underlying cause is appealing. Apparently they do
not want choice architects to produce economic or other losses by using
people’s inertia or inattention against them. There appears to be a general
moral principle here, one that imposes a presumptive barrier to certain
nudges: If people are to give up some part of their existing holdings, it must be because they
have affirmatively indicated their willingness to do so. We have evidence that this is
a widely shared moral principle. It might anchor a range of ethical judg-
ments and may even lie at the root of contract law, which often calls for
explicit consent before certain losses can occur.
At the same time, this principle leaves many open questions; it is also
­subject to qualifications. Our own findings suggest that it applies to money
(and also bodily parts). Would it apply to any form of property (for ­example,
real property or copyright)? We suspect so. Would it also apply to time?
Again we suspect so. Recall that if the government is taking money from
people’s current selves for the benefit of their future selves, Americans do not
object (see Chapter 2); the same point holds for citizens of several European
nations as well. And if the point of the default rule is to compensate victims
of wrongdoing, the principle is unlikely to be violated at all; people would
not complain if thieves were required to return stolen money. As we have
noted, the principle is not meant as a general attack on the tax system.
But our evidence suggests that in any stylized case in which the government
is presuming something like a donation—as when a default rule requires such
a donation without explicit consent—most people will react unfavorably.
Perhaps they believe that donations, as such, require personal responsibility.
Far more work remains to be done on these questions, above all to identify the
boundary conditions of what we have described as a general moral principle.
In philosophical circles, there is an extensive literature on the subject
of manipulation. In ordinary language, the term is one of opprobrium,
which raises two distinct questions: What, exactly, is manipulation, and
what is wrong with it? We do not yet have anything like a “map” to peo-
ple’s answers to those questions. But subliminal advertising can be taken
as a defining example of unacceptable manipulation, because it influences
people without engaging their conscious or deliberative capacities. The
influence occurs surreptitiously. If the government engages in subliminal
advertising, people will not approve, because the use of such advertising
seems unambiguously manipulative.
 Europe 45

National characteristics
Of the six nations, Italy and the UK are most favorably disposed toward
the nudges that we tested. In Italy, only one nudge (Nudge 14: sweet-free
cashier zones in supermarkets) is less popular than in most of the other
countries. Similarly, the UK is in the top ranks of approval 11 out of 15
times. (France and Germany cannot be so clearly ranked.) It is reasonable
to ask why Italy and the UK are comparatively receptive. We do not have an
answer to that question, but it is worthwhile to note that Italy is not known
to have a tradition or recent history of antipathy to paternalistic interven-
tions, and perhaps the recent experience of the UK, involving many uses of
behavioral science, has influenced public opinion.4
Both Hungary and Denmark are consistently less favorably disposed
toward nudges in general. The case of Hungary is not so puzzling. In that
nation, there is widespread distrust of social institutions, which has been
below the Organisation for Economic Co-operation and Development
(OECD) average for a long time.5 The legacy of Communism may lead
Hungarians to disapprove of government in general. At the same time,
Hungary is the country (from our subset) with the highest corruption
index, sliding even deeper down in past years.6 Moreover, it is below the
OECD level in voting in national elections.7 It is safe to hypothesize that
this lack of confidence did not improve with the Orban government. The
Hungarian findings also cast light on differences, within nations, with
respect to nudges: Citizens who distrust their government, or government
in general, will be less likely to approve of nudges, even if they approve
of the particular ends that those nudges would promote.8 We shall explore
this issue in detail in Chapter 6.

4 See David Halpern, Inside the Nudge Unit: How Small Changes Can Make a Big Difference (2015).
5 See Organisation for Economic Co-operation and Development, Confidence in Social Institu-
tions. Society at a Glance 2011: OECD Social Indicators (2011). The OECD Confidence in National
Institutions Index is based on the Gallup World Poll; it is based on questions regarding
confidence in the military, the judiciary and the national government. See also the Trans-
parency International Corruption Index at www.transparency.org/country/HUN.
6 See Id.
7 See Organisation for Economic Co-operation and Development (OECD) (2011), Voting.
In Society at a Glance 2011: OECD Social Indicators. http://dx.doi.org/10.1787/soc_glance-
2011-29-en. However, voting in the UK and France is even lower.
8 For a related finding, see David Tannenbaum et al., On the Misplaced Politics of Behav-
ioural Policy Interventions, Nature Human Behaviour 1(0130) (2017).
46 Europe

With respect to Denmark, our findings are far more difficult to explain.
That nation is not exactly known for its distrust of government, or for its
firm opposition to anything that smacks of paternalism. Denmark has had
one of the highest levels of trust in government from all OECD countries.9
However, while trust in politicians on a communal and regional level has
remained high, there has been a decline in trust in national politicians and
government over the last years. The results of a national survey in 2015
show that the trust in Danish politicians fell from 70 percent in 2007 to
28 percent in June 2015, an all-time low.10 In particular, our results might
be related to distrust in the new conservative policy landscape after the
federal elections in 2015. The new government had just started its term a
few months before the survey was executed. But overall, Denmark regu-
larly ranks first or second in international corruption rankings and trust in
government is still comparatively high. Some controversial health-related
interventions in Denmark (including a tax on foods with high levels of
saturated fats) might have contributed to our findings.

Politics and demographics

We do not find clear differences across party lines within Europe. One of our
main findings, and among the most surprising, is that party affiliations are
not correlated in any systematic way with support for the nudges we tested.
Within countries, however, there are some weak correlations and two over-
all patterns. (1) In France, Green Party and left-wing supporters are more
favorably disposed toward the tested nudges. (2) In the United Kingdom,
people who have voted for populist parties are particularly skeptical toward
information nudges. (3) Over all countries, European liberals are somewhat
less inclined to favor health nudges. (4) Over all countries, Green Party
voters are somewhat more inclined to favor environmental nudges (not sur-
prisingly). We suggest, however, that these findings should be taken with
some caution, in light of our rough measurement of political preferences
(most recent vote) and the clustering of political parties in Europe.

9 See EU European Commission, Public Opinion in the European Union. Standard Eurobarometer
No. 83. Brussels: EC (2014). See also Organisation for Economic Co-operation and Develop-
ment, Confidence in Social Institutions. In Society at a Glance 2011: OECD Social Indicators (2011).
10 www.altinget.dk/artikel/historisk-faa-danskere-stoler-paa-politikerne?ref=newslett
er&refid=17813&SNSubscribed=true&utm_source=Nyhedsbrev&utm_medium=e-
mail&utm_campaign=altingetdk.
 E urope 47

With respect to demographic differences, only one characteristic seems

to be correlated with people’s attitudes toward the nudges we tested:
gender. Women favor such nudges more than men do, with a less pro-
nounced (but still significant) gender divide in France and Denmark. In
general, we did not otherwise find statistically significant differences.
With respect to other demographic characteristics, we identified no rel-
evant correlations.

Larger themes
In Europe, there is strong majority support for nudges of the sort that have
been adopted, or under serious consideration, in democratic nations. If
respondents believe that a nudge has legitimate goals, and that it fits with
the interests or values of most people, they are likely to favor it. At the same
time, the citizens of six nations reject nudges that offend two principles
that command a consensus: First, government should not take people’s
money without their explicit consent; and second, government should not
manipulate people (at least in the defining case of subliminal advertising).
Despite the general European consensus, we find markedly lower levels
of support for nudges in two nations: Hungary and Denmark. In Hungary,
this finding is best explained by reference to reduced levels of trust for
government—a point that confirms the intuition that when distrust of
the competence or the motivation of public officials is high, even choice-
preserving interventions will be unwelcome. Lower levels of support in
Denmark are more challenging to explain.
In Europe, we have generally been unable to link political affiliations or
demographic variables to support for (or opposition to) nudges. Among the
few exceptions are: somewhat stronger female approval for the tested nudges;
a tendency (as expected) for Green Party voters to support environmental
nudges; and lower levels of support among European liberals for health nudges.
We do not doubt that people with certain political convictions are a
bit like the citizens of Hungary and Denmark, and therefore suspicious of
any government action, whether it consists of nudges, taxes, subsidies, or
mandates. But notably, we have been unable to find clear and consistent
evidence to this effect for any political party within Europe. It is also true
that some nudges seem to split Europeans along political lines. But when
this is so, it is because of the particular direction in which people are being
nudged—not because they are being nudged as such.
 Appendix to Chapter 4
Table A4.1  Samples, sampling, and methodology

Country Sample size Representativeness Survey method Weighting Sample Recruiting for the Census/Population Frame of the survey
method panel
Italy N=1,011 Online ­representative CAWI No Quota Offline and 35 mio internet No frames
for gender, age, region Omnibus weighting sampling online users, 18–64 years
UK N=2,033 F2f representative for CAWI RIM Quota Online 50.9 mio internet About saving
gender, age, region Omnibus sampling users, 18+ years and spending
habits
France N=1,022 F2f representative for CAWI Target Quota Online 41.05 mio internet About views on
gender, age, region Omnibus sampling users, 16–64 years the Ukraine
Germany N=1,012 Online ­representative CAWI RIM Quota Offline and 55.06 mio internet About views on
for gender, age, region Omnibus sampling online users, 14+ years the economy
Hungary N=1,001 F2f representative for CAWI ad RIM Quota Offline 7.35 mio internet Ad hoc, no
gender, age, region hoc sampling users, 15–69 years other frames
Denmark N=1,000 F2f representative for CAWI Target Quota Offline 4.54 mio internet About con-
gender, age, region Omnibus sampling users, 18+ years sumer goods
and the Great
Belt Bridge

Notes: “f2f (face to face) representative” means representative for the resident population. “Online representative” means representative for private internet ­users.
“CAWI” means Computer Assisted Web Interview. “Ad hoc” is used in Hungary where no omnibus survey was available. “RIM (Random Iterative Method) w ­ eighting”
and “target weighting” are statistical weighting methods applied to ensure validity of the multidimensional model when not every group of the ­population is
equally represented in a sample. “Quota sampling” means that data collection was done following quotas for specific socio-demographic c­ haracteristics, and
then the observations were weighted according to their frequency in the population. (The exception is Italy, where no weighting was needed.) “Frames” might
­unintendedly influence answers in omnibus surveys with multiple unrelated questionnaires. While frames cannot be avoided in practice, it is important to know
which questions had been asked before the question of interest to estimate unintended influence. “Mio” means million.
Table A4.2  Overview of political parties in the surveyed countries

Italy UK France Germany Hungary Denmark

Partito Democratico (PD) Conservative Socialiste, républicain et CDU/CSU Fidesz—KDNP Socialdemokra-terne (A)
citoyen
Movimento 5 Stelle Labour Les Républicains SPD Jobbik Dansk Folkeparti (O)
Il Popolo della Libertà (PdL) SNP (Scotland) Union des démocrates et Grüne MSZP Venstre (V)
indépendants
Scelta Civica con Monti per Liberal Radical, républicain, Die Linke Demokratikus Enhedslisten (Ø)
l’Italia Democrats démocrate et progressiste Koalíció (DK)
Sinistra Ecologia Libertà Plaid Cymru Écologiste FDP Lehet Más a Liberal Alliance (I)
(SEL) (Wales) Politika (LMP)
Lega Nord UK Gauche démocrate et Piraten Együtt 2014 Alternativet (Å)
Independence républicaine
Party
Fratelli d’Italia Green Party Front AfD Párbeszéd Det Radikale Venstre (B)
National Magyarországért
(PM)
Unione di Centro Freie Wähler Socialistisk Folkeparti (F)
Others Others Others Others Others Others
I didn’t vote. I didn’t vote. I didn’t vote. I didn’t vote. I didn’t vote. I didn’t vote.
Don’t know / no answer Don’t know / no Don’t know / no answer Don’t know / no Don’t know / no Don’t know / no answer
answer answer answer
Table A4.3  Clusters of the political parties in the surveyed countries

Political attitude cluster

Country Conservative Left-wing Liberal Green Populists & others
Italy Il Popolo delle Partito Democratico (PD) Scelta Civica con Sinistra Ecologia Liberta Movimento 5 Stelle
Liberta (PdL) Monti per l’Italia (SEL) Lega Nord
Unione di Centro Fratelli d’Italia
Others
United Conservative Labour Liberal Democrats Green SNP (Scotland)
Kingdom Plaid Cymru (Wales)
UK Independence
Others
France Les Républicains Socialiste, républicain et citoyen Écologiste La Front National
Union des Radical, républicain, démocrate et Others
démocrates et progressiste
indépendants Gauche démocrate et républicaine
Germany Christian Democrats Sozialdemokraten (SPD) Freie Demokraten Die Grünen Piraten
(CDU/CSU) Die Linke (FDP) AfD
Freie Wähler
Others
Hungary Fidesz—KDNP MSZP Lehet Más a Politika (LMP) Jobbik
Demokratikus Koalíció (DK) Párbeszéd Magyarországért Others
Együtt 2014 (PM)
Denmark Socialdemokraterne Venstre Alternativet Dansk Folkeparti
Enhedslisten Liberal Alliance Others
Socialistisk Folkeparti Det Radikale Venstre
(social-liberal)

Note: Reflects the political spectrum in 2015 for national elections.

Table A4.4  Estimates of demographics and political attitude on nudge approval: Multilevel analysis

(1) Information: (2) Information: (3) Default rules (4) Manipulation (5) Other mandates
Government campaigns Governmentally
mandated nudges
Male −2.105** −3.160*** −4.509*** −5.217*** −7.661***
(.671) (.723) (.661) (1.166) (.879)
[−3.420,−.790] [−4.577,−1.742] [−5.805,−3.213] [−7.502,−2.932] [−9.383,−5.939]
Age (categories) .407*** .127 .705*** .566** .037
(.111) (.120) (.109) (.193) (.146)
[.189,.625] [−.108,.362] [−.920,−.491] [−.944,−.187] [−.249,.322]

Political attitude
Conservative ref. ref. ref. ref. ref.
Left-wing −1.724 −.593 1.153 −7.165*** 1.46
(.987) (1.064) (.973) (1.715) (1.293)
[−3.658,.210] [−2.678,1.492] [−0.754,3.059] [−10.526,−3.804] [−1.074,3.993]
Liberal −2.88 −7.912*** −3.750* −13.760*** −6.314**
(1.618) (1.745) (1.595) (2.809) (2.120)
[−6.052,.292] [−11.332,−4.492] [−6.876,−.625] [−19.266,−8.255] [−10.468,−2.159]
Green −.920 −1.774 5.131*** −19.736*** 6.168**
(1.526) (1.645) (1.504) (2.651) (1.999)
[−3.910,2.071] [−4.999,1.450] [2.183,8.079] [−24.931,−14.540] [2.250,10.085]
Populist & others −5.370*** −5.679*** −3.170** −7.436*** −4.804**
(1.128) (1.217) (1.112) (1.960) (1.478)
[−7.582,−3.159] [−8.064,−3.295] [−5.350,−0.990] [−11.277,−3.595] [−7.701,−1.907]
(Continued)
 (1) Information: (2) Information: (3) Default rules (4) Manipulation (5) Other mandates
Government campaigns Governmentally
mandated nudges
Don’t know / did not −5.749*** −6.554*** −2.724** −8.024*** −4.197**
vote (1.027) (1.108) (1.012) (1.784) (1.346)
[−7.763,−3.736] [−8.724,−4.383] [−4.708,−0.739] [−11.521,−4.528] [−6.834,−1.560]
Obs. 7,079 7,079 7,079 7,079 7,079
Wald Chi^2 69.34 84.81 141.19 89.23 132.63
p-value (.000) (.000) (.000) (.000) (.000)
ICC (country) .069 .076 .043 .027 .070
(.037) (.041) (.024) (.016) (.038)

Note: * p≤.05; ** p≤.01; *** p≤.001. Estimates of a 2-level random intercept model. Standard errors (Confidence intervals) in parentheses. Dependent vari-
ables are the average nudge groups by intrusiveness (Min: 0; Max: 100). The intraclass correlation coefficient (ICC) is the proportion of total variance that is
attributed to the cluster “country.” “Ref.” means “reference variable”, i.e., a variable that serves as a baseline to measure the deviation of other related variables.
 5
A Global Consensus?
Not Quite

The discussion can be taken to offer five general lessons. First, citizens in
diverse nations generally approve of nudges, at least of the kind that have
been adopted or are under serious consideration in recent years. Second,
­citizens do not approve of nudges that they perceive to be inconsistent with
the interests or values of most choosers, such as a default rule by which
men’s last name would automatically change to that of their wives. Third,
citizens do not approve of nudges that are perceived as having an illicit
goal, such as religious or political favoritism. Fourth, citizens object to
manipulation, but they define it quite narrowly, as in the case of subliminal
advertising. Fifth, and quite surprisingly, political affiliation is generally a
weak predictor of citizens’ reactions to the tested nudges.
In this chapter, we offer results from eight countries: Australia, Brazil,
Canada, China, Japan, Russia, South Africa, and South Korea. These nations
were chosen in order to obtain a broad sample of countries with diversity
along identifiable lines. We include countries widely distributed on a scale
from liberal democracies with freedom of speech to authoritarian one-
party regimes. We also include four of the five BRICS countries (BRICS is
54 A Global Consensus ? Not  Quite

the acronym for an association of five major emerging national economies:

Brazil, Russia, India, China, and South Africa).1 The relevant countries have
markedly different levels of GDP and welfare. They include countries repre-
senting the “cultural clusters” explored in cultural studies literature.2 Least
interestingly, the countries have sufficient internet penetration rates3 to
conduct meaningful online representative surveys.4
We conducted such surveys (representative for age, gender, region, and
education), providing data from about 1,000 respondents per country,
who were asked whether they approve or disapprove of the 15 nudges
described in Chapter 4. Again we asked simply for a statement of approval
or disapproval, without measuring the intensity of approval or disapproval
on any kind of scale. In order to be able to compare and enlarge the overall
data set, we used the same survey instrument and largely the same meth-
odology applied in Chapter 4.
In view of the highly preliminary state of existing research, and the
inevitability of surprises, our goal was to learn about national similarities
and differences, and we did not begin with firm hypotheses. Tentatively,
however, we expected to support two hypotheses:

Hypothesis 1: The apparent cross-national consensus with respect to

nudges, reflected in the five lessons from earlier studies, would be found,
with modest variations, in all of the nations in our survey, with the excep-
tions of China and Russia.
Hypothesis 2: As nondemocratic nations, China and Russia would
show overwhelmingly high levels of support for nudges of all kinds, either

1 Unfortunately, we could not cover India with our online survey design due to many
different languages, a high sample size needed to capture the different regions and
­minorities, and a surprisingly low internet penetration rate.
2 See Vipin Gupta et al., Cultural Clusters: Methodology and Findings, J. World Bus. 37(1)
(2002), 11–5; Robert J. House et al., Culture, Leadership, and Organization: The GLOBE Study of
62 Societies (2004); Robert J. House et al., Strategic Leadership Across Cultures: The GLOBE Study of
CEO Leadership Behaviour and Effectiveness in 24 Countries (2014). The ten “culture clusters” used
typically in cultural studies (e.g., in the GLOBE study, House et al., 2014) are South Asia,
Anglo, Arab/Middle Eastern, Germanic Europe, Latin Europe, Eastern Europe, Confucian
Asia, Latin America, (Sub-Sahara) Africa, and Nordic Europe. Together with European
data collected in 2015 (see Chapter 3), we cover all clusters except for two (“Southern
Asia” and “Middle Eastern”).
3 www.internetlivestats.com/internet-users-by-country/.
4 An earlier version of this chapter has been published as: Cass R. Sunstein, Lucia A. Reisch
and Julius Rauber, A Worldwide Consensus on Nudging? Not Quite, But Almost, ­Regulation
and Governance 12(1), 3–22, doi:10.1111/rego.12161.
 A G lobal C onsensus? N ot Q uite 55

because of a belief that disapproval might be punished (even though we

guaranteed anonymity), or because disapproval of government policies
would be distinctly rare among citizens who are accustomed to auto-
cratic rule.

As we shall see, the first hypothesis was generally supported, but with an
important qualification: The first three lessons, and the fifth, can indeed be
found in essentially all of the nations that we studied, but the fourth (for-
bidding manipulation via subliminal advertising) cannot. As we shall also
see, the second hypothesis was supported with respect to China but not
with respect to Russia (whose citizens look more like those of the United
States and Europe). In addition, we found a number of surprises, above all
involving Japan (which showed unusually low approval rates) and South
Korea (which showed unusually high approval rates).
The most general lesson is that majority support for nudges cuts across
many nations with diverse cultures, political inclinations, and histories.
At the same time, our largest finding, which we did not anticipate, is that
the nations of the world can be provisionally grouped into three catego-
ries. The first, consistent with the first hypothesis and the existing US and
European data, reflects all of the five lessons sketched above. Of the nations
for which data are available, this is the largest group of the nations tested
thus far. Call these principled pro-nudge nations.
The second category, consistent with data from Denmark and Hungary,
generally shows majority approval but significantly lower approval rates;
Japan now joins this category. Call these, cautiously pro-nudge nations. The
third category, identified for the first time here, consists of nations with
massively high approval ratings. China and South Korea are the current
examples. (Recall that Russia is not included.) Call these, overwhelmingly
pro-nudge nations.
We suspect that many and probably most other nations would fall into
one of these three categories. We cannot, of course, exclude the p ­ ossibility
that some nations would show an altogether different pattern—with,
for example, far lower approval ratings than what we find in Denmark,
Hungary, and Japan (anti-nudge nations), or with divergent approval rates
across various categories of nudges (selectively pro-nudge nations).
Our main goal in this chapter is to report on the various national find-
ings, without offering a full account of why nations fall into one of the
three categories. By themselves, our findings do not provide any such
56 A Global Consensus ? Not  Quite

account, and in view of limits in existing knowledge, any speculations will

inevitably have an ad hoc character. Nonetheless, we shall not refrain from
offering some such speculations here; we shall turn in Chapter 6 to the
question of trust.

The study

Sampling and survey

Sampling and survey were performed with the support of Qualtrics, an
ISO-certified international market research company. To ensure the nec-
essary level of rigor, we monitored and commented on each step of the
sampling and survey implementation. As with the surveys reported in
previous chapters, we intended the eight country samples to be online rep-
resentatives with respect to age, gender, educational level, and region. To
reach this high level of representativeness, several steps were undertaken,
including oversampling and RIM weighting. (An overview of the weighted
and unweighted samples can be found in Table A5.1 in the appendix to
this chapter.)
As before, the core instrument is a simple questionnaire with 15 different
interventions. Also as before, respondents were asked to indicate for each
item whether they “approve” or “do not approve” of this specific “hypo-
thetical policy.” The potentially confusing word “nudge” (or the respective
translation) was deliberately not used in the survey; rather, the policy
instrument was described as simply and intelligibly as possible. We did not
intend to frame the policies in a way that would skew people’s answers. The
15 items of the questionnaire can be found in Chapter 4.
The items of our questionnaire were first entered in the Qualtrics web
interface in October 2016. Qualtrics checked the items in order to ensure
that they were understandable and consistent to an English-speaking audi-
ence. The questionnaire was translated from English (the blueprint for
all country studies) into the respective languages (Brazilian Portuguese,
Canadian French, Mandarin, Japanese, Russian, and Korean) and was back-
translated by native speakers and corrected accordingly. This additional step
was designed to ensure that people would have the same understanding of
the items in the different countries and that infrequently used words or con-
cepts would be fully understood and interpreted in the same way. Monetary
amounts used in some items were adapted to the ­specific ­countries based
 A Global C onsensus? N ot Q uite 57

on the exchange rate of the currency and its average income. As before, the
questions were presented in a randomized order.5
To ensure high quality samples, we included a range of validity and
robustness checks. Apparently inattentive or careless respondents were
excluded by employing a time filter (sorting out respondents who used
less than half of the median time needed to answer the survey) as well as by
adding two attention filters in the survey.6 Responses were allowed to enter
the final sample only when they met these attention standards and were
provided by adults (18 years and older) who lived in the respective country
and used its official language. The latter was ensured by a language default
using the language of the browser of a participant. In Canada, participants
could choose between French and English. Respondents also were forced
to answer all questions (i.e., no skipping and “cherry picking”). Only fully
completed questionnaires were accepted.
Field work started with a soft launch of ten percent of the data in all
countries concurrently in November 2016. Results were checked for con-
sistency, validity, and robustness. Minor adaptations were made for the
remaining 90 percent of the sampling. Field time ended after about five
weeks in December 2016.

Socio-demographics and political attitudes

We collected information on socio-demographic variables and politi-
cal attitudes. Comparability of socio-demographic variables among the
eight countries was given for gender (male/female), age (years), city size
(number of inhabitants), relationship status (married/civil partnership;
long-term relationship; single; divorced; widowed; others), and number
of children.
Comparability is less clear-cut for region, education, and income.
“Region” is country-specific, and so we used the categories provided
by national statistics; these data are more relevant for discussion of the
results within the respective countries than for comparison of countries.

5 The full questionnaire as well as complete information on socio-demographic variables

and political attitudes can be found in Sunstein et al., supra note 62, 17–22.
6 Attention filter 1 (after Nudge 7): “This is an attention filter. Please click on ‘3’ to go on
with the survey”). Attention filter 2 (after Nudge 15): “This is an attention filter. Please
click ‘approve’ to go on with the survey”—with the order of the two answer categories
being switched.
58 A Global Consensus ? Not  Quite

“Education” was measured in two ways: (1) the usual brackets of the
countries’ statistics (“highest degree reached”), allowing limited compa-
rability; and (2) “number of years of formal education,” which can more
easily be compared against the backdrop of the respective country specif-
ics such as average education level. “Income” must be understood in light
of the country’s income distribution and level to be useful; we therefore
developed and applied an algorithm based on the gross household median
income in each country.
Political attitude was measured in two ways: first, by choosing “political
party voted for in the latest election” (except for China, which has a one-
party system) from the full set of available political parties in the respective
country that received at least five percent of the votes in the last country-
wide election; and second, by a self-assessment political preference item
(“Where would you place yourself on this scale?”) presented as a Likert
scale ranging from (1) denoting “extremely liberal (left)” to (7) denoting
“extremely conservative (right).” The second approach was introduced as
a robustness check, and it also provided quantitative input in the multilevel
analysis described in the following paragraph. Admittedly, both measures
are rough, and hence our results should be interpreted cautiously.

Results

Approval rates
The country data sets were merged into one data set. Approval rates were
calculated per nudge and per country. Due to its nested character, the data
were suited for a multilevel analysis. For the latter, five independent vari-
ables were constructed, mirroring the analysis of the European country
data. As before, the 15 nudges were clustered with respect to their level
of intrusiveness in five clusters; average approval rates for each dependent
variable were calculated. Coding and analysis was done with SPSS. Analyses
were conducted for unweighted and weighted samples.7

7 There were only marginal differences between the two approaches. Regarding the
­approval rates, the largest difference between weighted and unweighted samples exists
for the nudge “healthy food placement” in Japan, with five percentage points (weighted
sample: 47 percent; unweighted sample: 42 percent). For all other nudges, maximum
deviation is three percentage points between the weighted and unweighted samples in
all countries. With respect to significant coefficients in the multilevel analysis (Table 5.1),
 A G lobal C onsensus? N ot Q uite 59

As noted, we categorized the 15 nudges in five levels of depth of interven-

tion: governmental information campaigns (Nudges 6 and 7) (Figure 5.1);
mandatory information disclosure requirements imposed by g­ overnments
(Nudges 1, 2, 10, and 12) (Figure 5.2)8; mandatory default rules imposed
by governments (Nudges 3, 4, 5, 9, 11, and 13) (Figure 5.3); ­mandatory
subliminal advertising (Nudge 8) (Figure 5.4); and mandatory choice archi-
tecture in supermarkets and public cafeterias (Nudge 14 and Nudge 15)
(Figure 5.5).
As noted, our findings are generally consistent with Hypothesis 1. With
respect to information campaigns, we observe majority support in all eight
nations, and consistent with that hypothesis, the similarities are far more
noteworthy than the differences. For nudges tackling childhood obesity
and distracted driving, the level of support is overwhelming.
Consistent with Hypothesis 1, mandatory information disclosure also
receives very high levels of support, and here too, the similarities dwarf the

Figure 5.1  General information campaigns. CI, confidence interval

there were only two differences between weighted and unweighted samples: the age
coefficient in Column 1 becomes insignificant when using the unweighted sample; the
coefficient of “years in school” becomes significant when using the unweighted sample.
8 Note that Nudge 12 (“The federal government requires movie theaters to run public
education messages designed to discourage people from smoking and overeating”) was,
in the present study, moved from Nudge Cluster 1 (“pure governmental information
campaigns”) to Nudge Cluster 2 (“mandatory disclosure requirements”). This change
was a result of what we found to be convincing arguments with respect to our European
study (Chapter 3). Statistics show that the results of the European study are still largely—
though not fully—comparable with the present one.
60 A Global Consensus ? Not  Quite

Figure 5.2  Mandatory information imposed by governments. CI, confidence

interval

differences. Japan is the obvious and only outlier. Majorities do approve,

but in all cases, Japan shows the highest levels of disapproval, in particular
for traffic lights and educational messages against smoking and overeating.
With respect to default rules, approval levels diminish, and in some cases
majorities disapprove. Here again, we find general support for Hypothesis 1.
Consistent with findings in the United States and the previously tested
nations in Europe, strong majorities favor not only encouragement of auto-
matic enrollment in green energy but also a mandate to that effect. (To be
sure, approval rates should be expected to decrease if people were told
that the cost of green energy was higher than that of other sources.) Also
consistent with earlier findings, majorities disapprove of a default carbon
charge and also a default charitable donation. Again, the basic principle
here seems to have something to do with loss aversion: In general, people
do not favor default rules that would take people’s money without their
explicit consent.
Because of their high approval rates, China, South Korea, and (to a lesser
extent) Brazil are outliers here. It is worth underlining the fact that in both
 A G lobal C onsensus? N ot Q uite 61

Figure 5.3  Mandatory default rules imposed by governments. CI, confidence

interval

China and South Korea, strong majorities favor default rules that are widely
disapproved in most other nations. Here is evidence against Hypothesis 2:
the unusually high approval ratings come from China and South Korea, not
China and Russia.
With respect to healthy food placement and active choosing for organ
donation, the picture is broadly consistent across nations. For healthy food
62 A Global Consensus ? Not  Quite

Figure 5.4  Mandatory subliminal advertising. CI, confidence interval

Figure 5.5  Mandatory choice architecture. CI, confidence interval

placement, Japan is yet again the evident outlier. For organ donation, China
and South Korea show what is, for those nations, anomalously low approval
ratings. Majorities disapprove in Russia as well as Japan. The Russian case
is interesting and not simple to explain, though it is likely connected with
identifiable (though to our knowledge not yet identified) aspects of the
political and cultural backdrop.
As we have noted, previous work shows high levels of disapproval of
­subliminal advertising (along with significant support), even for what
might seem to be a good cause, and the standard pattern is roughly observed
in Canada, Japan, and Russia. One possible way to think about such data is
 A Global C onsensus? N ot Q uite 63

that the minority approval rates (often in excess of 40 percent) represent

the upper bound on the degree to which people will approve of violations
of autonomy when the goals of the policy are deemed laudable.
But puzzlingly, and in a partial but important rejection of Hypothesis 1,
we find overwhelmingly high approval rates in China and South Korea, and
majority support as well in Australia, Brazil, and South Africa. We did not
anticipate this result and are unsure how to explain it. One p
­ ossibility is that
the very idea of subliminal advertising is not perceived as ­especially trou-
bling in those nations. Another possibility is that the public policy goals are
taken to be sufficiently compelling as to justify the use of a ­presumptively
unacceptable tool.
Sweet-free cashier zones and meat-free days produced strikingly similar
patterns of results. The former—a kind of choice architecture designed to
promote health—did obtain majority support in all nations except Japan,
but with a significant spread between the highest rate (in Australia) and
the lowest (in Russia). Meat-free days also obtained majority support in all
nations with the exception of Japan, but here China was the most support-
ive and Australia the least.
Consistent with Hypothesis 1, the patterns that we observe here are simi-
lar to those found in Europe and the United States. With respect to sweet-free
cashier zones, for example, Australia, China, and South Africa look a lot like
Italy, France, and the United Kingdom. With respect to meat-free days, Italy
and France are quite similar to Brazil, China, and South Africa.

Demographics and political attitude

We estimated the multilevel regression for each of the five levels of depth
of intervention, with the approval rates of the 15 nudges being dependent
variables. We calculated mean approval in percentages by level of interven-
tion. Gender, age, educational level (in years of schooling), and political
attitude (self-assessed) were used as independent variables on an individual
level, and country on the country level. Results are presented in Table 5.1
and put into perspective below.
Notably, Table 5.1 shows that gender has a systematic influence on par-
ticipants’ approval of nudges: Women approve four out of five nudge types
(2, 3, 4, 5) significantly more than men do. We would take this finding
with some caution. Undoubtedly there are some nudges that would show
the opposite pattern (nudges that encourage boxing, gambling, drinking,
Table 5.1  Estimates of selected socio-demographics and political attitude on nudge approval per nudge cluster: Multilevel analysis

Nudge clusters (1) Information: Government (2) Information: Governmentally (3) Default (4) Subliminal (5) Other
campaigns ­mandated nudges rules advertising mandates

Male .281 −1.429** −1.707** −5.767*** −4.024***

(.459) (.531) (.574) (1.065) (.862)
.034* .068*** −.060*** −.069* .191***
Age (in years)
(.014) (.017) (.018) (.034) (.028)
.005 .179** −.091 −.235* −.037
School (in years)
(.049) (.057) (.061) (.114) (.092)
Political attitude −.008 −.448* −.533* 1.544*** −.760*
(1= liberal to 7=conserv.) (.186) (.215) (.233) (.432) (.350)
Obs. 7,594 7,594 7,594 7,594 7,594
ICC (intercept=country) .024 .098 .118 .101 .112
P value intercept variance (.056) (.048) (.047) (.048) (.047)

Legend: * p≤.05; ** p≤.01; *** p≤.001.

Note: Estimates of a 2-level random intercept model. Standard errors in parentheses. Dependent variables are the average nudge groups by intrusive-
ness (Min: 0; Max: 100). The intraclass correlation coefficient (ICC) is the proportion of total variance that is attributed to the cluster “country.”
 A G lobal C onsensus? N ot Q uite 65

and hunting?). If women are more supportive of the nudges tested here, it
is probably because of the particular goals of those nudges.
The influence of age is strong but operates differently for different nudges.
Older people tend to favor less intrusive interventions—such as information
campaigns and information nudges that are mandated by the government—
significantly more than younger people. Similarly, a meat-free day and
sweet-free cashier zones are favored more strongly by older survey partici-
pants. At the same time, younger people are more likely than older people
to approve of more intrusive interventions (such as manipulative messages
and default rules). We would not make much of these findings.
Education (measured by school attendance in years) has a weaker influ-
ence, and it cuts in intriguingly different directions: The higher the number
of years in school, the higher the approval level for governmentally man-
dated information nudges and the lower the approval level for subliminal
advertising. It is plausible to speculate that more formally educated people
are highly receptive to information as a regulatory tool; and they might
be more skeptical of any use of government power to manipulate people.
A mixed picture emerges with respect to the influence of participants’
self-assessed political attitude. As expected, acceptance for nudges rises with
the grade of “liberalism” (meaning left-of-center) for three out of the five
nudge types (2, 3, 5). Interestingly, the opposite is true for subliminal
advertising, which is more likely to be supported by conservatives than by
liberals. In general, and importantly, political attitudes have only a modest
effect on approval rates, consistent with Chapters 2 and 4.

Three categories of nations

Hypothesis 1 is strongly supported (with a few qualifications). Overall,
the level of approval of the presented nudges in our survey countries is
generally high. The majority of respondents approve of most of the nudges
in nearly all of the countries. In general, we find more similarities than
differences among the surveyed countries. The same holds true when we
compare the results with the US and European studies discussed in previous
chapters.
There appears to be a large category of nations where majorities are
likely to approve of nudges so long as they have legitimate ends and are
consistent with the interests and values of most people. This, then, is the
first of the three categories of nations that we are now in a position to
66 A Global Consensus ? Not  Quite

describe: principled pro-nudge nations, where identifiable principles sepa-

rate majority approval from majority disapproval.
The category of principled pro-nudge nations includes the industrialized
Western democracies of our sample (Australia, Canada, France, Germany,
Italy, the UK, the US), where we find exceptionally similar approval rates.
Apparently such nations have similar norms and values, at least with respect
to nudges. It may also be relevant that in much of the Anglo-Saxon world
and in some of our sample countries in particular, nudges have been used
and publicly debated for many years.
Russia, Brazil, and South Africa show broadly similar patterns, and
with appropriate qualifications they can be placed in the same category as
Western democracies. Of the three, Russia is the most surprising (and in
partial rejection of both of our hypotheses). More research on the three
nations would be necessary to explain the basic findings here.9
By contrast, the three Asian countries look very different. Contrary
to our expectations, Japan is—like Denmark and Hungary in Europe—a
clear outlier, with systematically and significantly lower approval rates
than all other countries in 13 out of 15 cases. (The two exceptions are
that Russians show higher disapproval rates of the carbon emission charge
and that Canadians are more likely to disapprove of the Red Cross default
­donation.) In this light, it seems safe to say that a cluster of nations shows
distinctly lower enthusiasm for nudges—and that in the fullness of time,
we will have a clearer sense of which nations will join the category now
containing Denmark, Hungary, and Japan. These are cautiously pro-nudge
nations; there may turn out to be anti-nudge nations as well, though we
have not yet found any.
It should also be possible to obtain a much clearer understanding of
exactly why approval rates are lower in those nations. It is plausible to spec-
ulate that there is, in those nations, relatively less enthusiasm for the ends
that the relevant nudges are designed to promote. If, for example, reducing
smoking does not seem so important, then there will be less support for
nudges that are designed to reduce smoking. We suspect that lower levels
of enthusiasm for the relevant end do explain some of our findings. But
with respect to Denmark, Hungary, and Japan, the more natural (though

9 There are some examples for interventions based on behavioral insights in these coun-
tries: According to the Moscow Times, Russia had planned the introduction of “traffic light”
food labeling in 2017. Brazil adopted a law in 1997 regarding organ donor choice when
getting the driver’s license; however, the law was repealed in 1998 (http://news.bbc.
co.uk/1/hi/health/7733190.stm).
 A Global C onsensus? N ot Q uite 67

also speculative) explanation points to reduced levels of trust in govern-

ment. Many people might follow a kind of heuristic: If the government plans
to do it, it is probably a bad idea. More systematic analysis would, of course, be
necessary to test this explanation, and to understand why trust would be
reduced in the relevant periods.
At the current time, there is only limited recent data on levels of
­confidence in government, and it does suggest that reduced levels do help
to explain the data for Hungary in particular.10 Also consistent with our
­findings, Japan’s confidence in government level is slightly below the
OECD average (yet rising since 2007),11 but puzzlingly, Denmark sticks
out—as do all Scandinavian countries—with high trust levels.12 We attempt
to make more progress on these issues in Chapter 6.
South Korea and China are also outliers, but in the other direction,
generally showing overwhelmingly high approval rates for all nudges.
It  therefore seems safe to say that there is a third category of nations
­showing especially high enthusiasm for nudges, and therefore warranting
the name of overwhelmingly pro-nudge nations.
We do not yet know how many nations fall into this category, nor do
we know what accounts for their high levels of enthusiasm. A tempting
though again speculative explanation, paralleling that just given, is that there
is a consensus, in those nations, that particular nudges have compelling
­justifications—say, because of a widespread belief that distracted driving is
a serious problem. Another explanation is that in those nations, trust in gov-
ernment is particularly high, so that strong majorities are inclined to support
any policy, even if it is hypothetical. They follow a kind of heuristic: If the
­government plans to do it, it is probably a good idea. There is survey evidence that in
China, ­levels of trust in government are indeed high and rising.13 The Chinese

10 See OECD, “Confidence in National Government in 2014 and its Change since 2007” in
Government at a Glance 2015 (2015); OECD, Trust and Public Policy: How Better Governance Can Help
Rebuild Public Trust (2017).
11 www.oecd.org/gov/GAAG2013_CFS_JPN.pdf.
12 See OECD, “Confidence in National Government in 2014 and its Change Since 2007” in
Government at a Glance 2015 (2015); OECD, Trust and Public Policy: How Better Governance Can Help
Rebuild Public Trust (2017).
13 “The Chinese trust in their government has been rising steadily as Chinese perceive that
their government is acting for their best interests—rather than for a privileged few”
www.quora.com/Do-Chinese-citizens-trust-their-government (accessed February 20,
2017). See also Pew Research Center, Global Attitudes and Trends (www.pewglobal.
org/2013/05/23/chapter-1-national-and-economic-conditions/). According to the
Edelman Trust Barometer, trust percentage in government rose again in 2018; as a result,
68 A Global Consensus ? Not  Quite

people have high expectations that the state will raise living standards through
direct control over investments and markets, and on average, living standards
have improved markedly; the government has delivered in this respect.
We suspect, though we cannot prove, that these various explanations
­capture a large part of the picture—but not all of it. Begin with China, where
approval rates tend to be highest of all. One reason could be that environ-
mental issues in most of China are severe, and the adverse effects can be felt
directly by the citizens, leading to general enthusiasm for nudges that involve
the environment, health, or safety. Air pollution has received sustained
attention in China, and during our field time, Premier Li ­publicly called for
cleaner energy sources to adhere to the Paris Agreement; 23 Chinese cities
had issued “red alerts” in December 2016 due to alarmingly unhealthy air
pollution levels. It is also possible that Chinese people do indeed trust their
government strongly and they genuinely approve most of its policies.
But it may also be relevant that in China, people are used to an authori-
tarian regime, run by the Chinese Communist Party, which intrudes on
people’s private decisions through mandates and bans (as, for example,
through the one child policy and a recent plan to introduce a national
smoking ban in public places). If mandates and bans are background facts,
nudges might seem entirely unobjectionable.
Yet another possibility is that even though they were guaranteed a­ nonymity,
our respondents felt some pressure to declare support for the r­ elevant ­policies.
Consider the “Citizens Score,” used by the Chinese g­ overnment to classify its
citizens into “good” or “bad” citizens; the existence of the score might act
as a strong incentive to approve (online) everything the government plans.14
In short, our results, showing stunningly high approval rates, might reflect a
form of “preference falsification.”15
What about South Korea? We did not expect to find the patterns there, and a
full account would require a detailed investigation. The markedly high approval
rates for most nudges might again be a product of enthusiasm for the policy

China ranks among the leading countries worldwide (www.forbes.com/sites/niallmc-

carthy/2018/01/22/the-countries-that-trust-their-government-most-and-least-
infographic/#1ab14f79777a). We cannot, of course, exclude the possibility that these
results are unreliable in a nation that is neither free nor democratic. For potential reasons
and an academic discussion of this phenomenon—however from a decade ago—see
Zhengxu Wang, Before the Emergence of Critical Citizens: Economic D ­ evelopment and
Political Trust in China, Intl. Rev. of Soc. 15, 155 (2005).
14 www.aclu.org/blog/free-future/chinas-nightmarish-citizen-scores-are-warning-americans.
15 See Timur Kuran, Private Truths, Public Lies:The Social Consequences of Preference Falsification (1995).
 A Global C onsensus? N ot Q uite 69

goals and of general trust in government. In addition, there has been consider-
able discussion of nudging in the South Korean press. Though we did not use
the term “nudge,” and though we would not claim that press discussions are
causal here, the idea of choice-preserving interventions, designed to promote
health and safety goals, may well be familiar in the South Korean culture.
On the other hand, the high approval rates can be seen as surprising
against the background of the two-month-long mass protests during our
field time against the former President Park Geun-hye, who was accused of
corruption. There was a threat of declaration of martial law to quell public
protests. In the recent past, people in South Korea showed low levels of con-
fidence in their government, and their suspicion with respect to corruption
equaled that of Hungary.16 Apparently public concerns about corruption,
and about the current government, were insufficient to produce significant
levels of disapproval of the kinds of policies tested here—a fact that may
well attest to the deep cultural receptivity, in South Korea, to those policies.

Lessons
Studying diverse nations, we find strong majority support for nudges, with
the important exception of Japan, and with spectacularly high approval
rates in China and South Korea. The largest conclusion is that the nations of
the world appear to fall into three groups: (1) a sizeable group of nations,
mostly liberal democracies, where strong majorities approve of health and
safety nudges; (2) a small group of nations where overwhelming majori-
ties approve of nearly all nudges; and (3) a small group of nations where
majorities generally support nudges, but where the level of support is
markedly lower than in nations that fall in category (1).
With respect to cross-national differences, much remains to be learned
and our explanations have been tentative and speculative; this is an important
domain for further work and for the development of testable hypotheses. For
example, we do not know whether the very high levels of support in China
reflect trust in government, enthusiasm about the policy goals, adaptation to
the extensive use of government power, or some form of “preference falsi-
fication,” producing misleadingly high levels of support in surveys.
Nor do we know, as yet, whether many countries fall within the cat-
egory of overwhelmingly pro-nudge nations, now containing only China

16 See OECD, “Confidence in National Government in 2014 and its Change since 2007” in
Government at a Glance 2015 (2015).
70 A Global Consensus ? Not  Quite

and South Korea, or whether the category of more cautiously pro-nudge

nations is small and greatly dominated, in terms of sheer numbers, by
the principled pro-nudge consensus among democratic nations (as now
appears). It also remains possible that some nations would show only
minority support for the nudges tested here.
We have speculated that for Hungary and Japan, a lack of trust in govern-
ment is a significant part of the picture. It would be parsimonious to show
that trust, across nations and across time, provides the principal explana-
tion of cross-national differences. We now turn to that question.

Appendix to Chapter 5

Table A5.1  Observations RIM weighted/unweighted for all countries

Weighted sample Unweighted sample

Country Frequency Percent Frequency Percent

Australia 1,000 12.5 1,001 12.6

Brazil 1,000 12.5 1,000 12.6
Canada 1,000 12.5 1,137 14.3
China 1,000 12.5 985 12.4
Japan 1,000 12.5 1,005 12.7
Russia 1,000 12.5 918 11.6
South 1,000 12.5 949 12.0
Africa
South 1,000 12.5 932 11.8
Korea
Total 8,000 100.0 7,927 100.0

Note: For each country, we predefined country-specific quotas for socio-demographic

­variables on the basis of the respective national census data to be reached in the sampling.
In Australia, Brazil, Canada, and Japan, quotas for age, gender, and region could be reached.
In China, Russia, South Africa, and South Korea, it turned out to be impossible to recruit
the needed numbers of low-educated respondents—which is not so surprising in light of
the fact that we used a web-based instrument. After several extensions of field time, we had
to loosen the quotas for education in all countries. To make up for this s­ hortcoming, we
used oversampling and weighting. To ensure representativeness with respect to ­gender, age,
­region, and education for China, Russia, South Africa, and South K­ orea, RIM ­weighting was
conducted. The same procedure was conducted for the samples of Australia, Brazil, Canada,
and Japan (which were already representative with respect to age, gender, and region) in
order to ensure representative results regarding education levels.
 6
Trusting nudges

In this chapter, we dig deeper. We seek to understand the differences

among nations in more detail. We focus above all on the question of trust.
To do that digging, we collected additional data in four of our study
countries (Germany, Denmark, South Korea, and the US) in 2018.1 We
chose one nation from each of the three categories of nudge endorsement
and one from each of three different cultural clusters. We also added com-
parable survey data from Belgium (Flanders).2 In addition to the 15 nudges
and the socio-demographic variables, we asked participants to answer a
large questionnaire including anthropometrics (to calculate Body Mass
Index), lifestyle factors, consumption of specific products (alcohol, smok-
ing, and meat), employment status and type, subjective health status and
health satisfaction, social trust and trust in institutions, concerns about the
environment, world-views and thinking styles (i.e., future outlook, belief

1 An earlier version of this chapter was published as Cass R. Sunstein, Lucia A. Reisch and
Micha Kaiser (forthcoming), Trusting Nudges? Lessons from An International Survey, Journal
of European Public Policy, www.tandfonline.com/doi/full/10.1080.13501763.2018.1531912.
2 This data has been provided by the Flemish government that followed our survey
­procedure. For Flemish results see: Veerle Beyst and Kristof Rubens (2018), Wordt ­“nudging”
in het beleid aanvaard in Vlaanderen?, VTOM 2018/4, pp. 53–65.
72 Trusting nudges

in free markets, political attitudes, risk aversion), and several additional

variables. We speculated that these variables could help explain differences
between social groups as well as across nations.
Above all, we were interested in the psychological concepts of social and
institutional trust. These concepts have since long been depicted as impor-
tant indicators of the strength and quality of societies, communities, and
governments across the world. Validated measurement items as well as prev-
alence estimates are available for most countries worldwide—for instance,
from the World Values Survey data set.3 In our study, we hypothesized that
people who have a high trust in public institutions would be more willing
to accept government nudging in our tested areas. We also speculated that
strong believers in the free market might be less inclined to do so.
We also tested other variables. The influence of environmental concern
on attitudes and behavior has been studied in depth and in international
contexts.4 It seems intuitive that people who have a marked concern
regarding the environment5 will endorse environmental policies in general
and “green nudges” specifically.6 As expected, this is what we found. For
similar reasons, we speculated that a fragile individual health status and
high health concerns for oneself and others might be positively correlated
with approval of health nudges. A recent study7 reported that a higher
Body Mass Index (BMI) was positively correlated with support for menu
labelling policies (which is Nudge 1 in our list of 15 nudges). We also
explored the influence of consumption habits (i.e., meat, tobacco, alcohol,
and mobility) on the approval of the respective nudges.

3 See Ronald Inglehart et al. (eds.), World Values Survey: Round Six—Country-Pooled (2014). Data-
file Version: www.worldvaluessurvey.org/WVSDocumentationWV6.jsp. Social trust was
measured by the questions from the World Values Survey (WVS): “Would you say that
most people can be trusted?” and “How much do you trust people from the following
various groups”.
4 E.g., A. Franzen and D.Vogl, 2013,Two Decades of Measuring Environmental Attitudes: A Com-
parative Analysis of 33 Countries, Global Environmental Change 23(5), 1001–8; Wouter ­Poortinga,
Linda Steg, and Charles Vlek (2004), Values, Environmental Concern, and ­Environmental
­Behavior: A Study into Household Energy Use, Environment and Behavior 36(1), 70–93.
5 Measured by the question: “How much are you concerned about the environment?”.
6 Note that the aim of this research was not to compare the approval of different policy
tools such as legislation, taxes, or behavioral nudges, nor the different ways nudges are
framed, e.g., as win or loss. Other studies have done that.
7 J. Bhawra, J. L. Reid, and C. M. White et al. (2018), Are Young Canadians Supportive of
Proposed Nutrition Policies and Regulations? An Overview of Policy Support and the
Impact of Socio-demographic Factors on Public Opinion, Canadian Journal of Public Health.
 Trusting nudges 73

As before, we were also interested whether approval rates of nudging

depend on political attitudes of people. As we saw in Chapter 2, Republicans
in the United States show somewhat lower approval rates for some nudges
than do Democrats. However, this could well be due to the choice of pol-
icy domains; recall “partisan nudge bias.” In earlier chapters, we found
no systematic correlation along approval and party affiliations. Finally, we
speculated that risk aversion, job satisfaction, and subjective well-being
might have an impact on approval.
In brief: With this study, we aimed to understand why people in selected
countries approve or disapprove of a set of 15 nudges, mainly in the field of
environmental protection and health. Regarding explanatory variables, our
principal focus is on trust in governmental institutions. Further, by repli-
cating the surveys in earlier chapters, we test the robustness of our earlier
results regarding approval rates and socio-demographics, in particular the
influence of gender. And by compiling all available data on nudge approval
rates from the three waves in 16 countries (as far as methodologically pos-
sible), we hope to shed broader light on public acceptance of nudges.
The remainder of this chapter is organized as follows. We first present the
methodology of the study by describing the samples, the survey, the vari-
ables, as well as the multi-step statistical analysis. We then show the results
in the five countries (Belgium, Denmark, Germany, South Korea, and the
US), emphasizing above all the relationship between the trust variables
and approval rates. We also compare the present results with earlier survey
waves in selected countries and provide an overall view of all surveys of
all our respective empirical studies. We discuss the results and limitations
of our study and conclude with comments on implications for nudging
research and behavioral public policy. Our ultimate emphasis, based on
our findings about trust, involve the importance of public participation and
consultation with respect to behaviorally informed policies.

The study

Sampling and survey

We employed an online representative survey in five countries representing
different levels of overall nudge approval as sketched in Chapter 5: The US
and Germany (principled pro-nudge nations), South Korea (overwhelm-
ingly pro-nudge), and Denmark (cautiously pro-nudge). As a new country
74 Trusting nudges

in our database, we included (the Flemish part of) Belgium.8 Again, we

employed the market research company Qualtrics9 to conduct the survey.
Field time was during six weeks in January and February 2018. We had
permanent access to the survey data and could monitor the survey and the
fulfilment of quota on a daily basis. Table A6.1 in the appendix provides an
overview of the different samples and sampling of this survey.

Survey instrument
We employed a questionnaire with 53 questions including socio-demo-
graphic variables; the list of 15 nudges in a randomized order as employed
in our earlier studies; a measure of political attitudes; questions measuring
psychological constructs (such as social trust and trust in government, per-
ceived freedom of choice) as well as variables describing individual factors
(such as perceived individual health, environmental concern, social trust
as well as consumption practices, such as smoking and drinking habits).
The complete survey instrument, as the descriptive statistics of the under-
lying data set, and the full list of variables employed have been published
­elsewhere.10
As before: The questionnaire was fully structured, and respondents
were required to follow them as provided. Each item was shown on a
single screen. Answering categories were adapted to the respective ques-
tions and ranged from Likert scales to binary schemes. Except for the
socio-­demographic questions, all items were randomized. As in all earlier
surveys, respondents were prompted with the question “Do you approve
or disprove of the following hypothetical policy?” The answer categories
were “approve” or “disapprove.”
With respect to “trust in institutions,” we used two different questions
to reduce the risk of methodological artifacts. The first was taken from the
World Value Survey: “How much do you trust in the following institu-
tions?” Then a set of public institutions was listed (namely: the armed
forces; the police; the courts; the government; political parties; parliament;

8 To ensure the same approach and level of quality of the Flemish sampling and survey
processes (and later on, the Mexican study), we developed a Standard Operation Proce-
dure on how to run our survey; it is available from the authors on request.
9 www.qualtrics.com.
10 See Appendix A1 and A2 respectively in Cass R. Sunstein, Lucia A. Reisch and Micha
Kaiser (forthcoming), Trusting Nudges?, supra note 1.
 Trusting nudges 75

the civil service; universities; the European Union; the United Nations).
The second item asked: “How much do you trust governmental institu-
tions?” We also asked whether people believe in the free market as the best
way to solve environmental and economic problems, a question used in
environmental research. All items were to be answered on a seven-point
Likert scale.
As in previous surveys, statistical equivalence of the questionnaire was
ensured by professional translation of the new questionnaire items from
English in the respective languages, followed by a back-translation into
English. The Flemish questionnaire was translated, back-translated, and
adapted in full. Due to the high internet penetration rates in the covered
countries, we could assume that answers were not systematically skewed
due to lack of internet access or proficiency.

Statistical analysis
The statistical analysis took place in several steps and with several meth-
odological approaches. (In the present chapter, we report only on the main
results.) In a first step, in order to get an overview of whether and how
this large number of variables (23) were interlinked, we drew a correlation
heatmap indicating correlations among all variables (Figure 6.1). On the
basis of the heatmap, we selected obviously correlating variables as identi-
fied by the map and looked into those more in depth. We then undertook a
multilevel regression of all variables and nudge approval (Table 6.1), tested
the robustness of the results with the help of a decision tree analysis, and
estimated the size of the probabilities. For the regressions and the machine
learning algorithms, the 15 nudges were categorized in five nudge clusters
as used in Chapter 5.

Results

Nudge approval, trust, and selected variables

The correlation heatmap as shown in Figure 6.1 suggests some expected
descriptive correlations between nudge approval and relevant variables.
As in prior studies, gender and age showed significant correlations with
approval. Moreover, the new variables “trust in institutions” and “envi-
ronmental concern” were found to correlate strongly with higher nudge
76 Trusting nudges

Figure 6.1  Correlation heatmap of relevant variables

Note: To download this figure in color please visit the e-resource at www.routledge.
com/9781138322783.

approval. “Belief in markets” was correlated with lower approval. Approval

rates by gender, conditional on trust in institutions (trustscore_inst), are
depicted in Figure 6.2. As shown, higher trust in institutions seems to be
linked to higher approval on average, and more so for women than for
men.
Interestingly, the concepts “social trust” and “trust in other people”
were not correlated with approval rates. But that is not entirely surprising;
our focus is on governmental policies, and higher trust in institutions is the
more relevant question. Furthermore, and perhaps surprisingly, the heat-
map did not suggest strong and significant correlations between overall
nudge approval and a large set of variables, notably health status and health
concern for oneself, subjective well-being, perceived freedom of choice,
risk aversion, and BMI.
 Trusting nudges 77

Figure 6.2  Overall nudge approval, conditional on trust in institutions

Note: The graph uses the World Wide Survey Question “On a scale of 1 to 7: How much do
you trust the following (10) institutions” as explanatory variable (trustscore_inst).

At the same time, the map does suggest some expected results. Meat con-
sumption seems to be negatively correlated with approval of “a meat-free
day in public canteens” (Nudge 15); smokers disapprove of government
campaigns (and subliminal advertising) against smoking (Nudge 12),
and people who drink alcohol more frequently disapprove of nudging in
­general. To that extent, behavior seems to play a role; people do not want
to be nudged to stop doing something that they like to do, and are now
doing. In a way, that should not be surprising, but it might have been
­predicted that people engaging in harmful behavior (such as smoking)
might be especially supportive of efforts to reduce that behavior.

Trust in institutions
The relationship between trust in institutions and nudge approval was
confirmed by weighted regression, where the effects were strong and
significant. As expected, we also found a significant negative relationship
between “belief in markets” and nudge approval. (Note parenthetically
that we might have tested nudges that promote reliance on markets, in
which case the relationship would be expected to be positive.) Column
(1) in Table 6.1 shows the regression results for all nudges together as well
as for the five nudge clusters.
Table 6.1  Weighted OLS regression for different nudge clusters (2018 survey)

Clusters
(1) (2) (3) (4) (5) (6)
Overall Government Information Default rules Subliminal ads Other
approval campaigns nudges mandates
GER 0.0316*** 0.0165 0.0494*** −0.0169 0.0986*** 0.1394***
(0.012) (0.014) (0.015) (0.014) (0.025) (0.020)
DEN −0.0689*** −0.0800*** −0.1041*** −0.0776*** −0.0540** 0.0193
(0.011) (0.014) (0.017) (0.014) (0.024) (0.020)
KOREA 0.1390*** 0.1289*** 0.1961*** 0.1257*** 0.3433*** 0.0064
(0.014) (0.017) (0.018) (0.018) (0.031) (0.027)
BE 0.0413*** 0.0334** 0.0219 0.0046 0.1728*** 0.1267***
(0.011) (0.014) (0.016) (0.014) (0.024) (0.020)
Male −0.0188*** −0.0130 −0.0169* −0.0104 −0.0124 −0.0590***
(0.007) (0.009) (0.010) (0.009) (0.016) (0.013)
Age −0.0006** 0.0007** −0.0000 −0.0014*** −0.0024*** −0.0004
(0.000) (0.000) (0.000) (0.000) (0.001) (0.000)
Yos −0.0032*** −0.0015 −0.0031*** −0.0034*** −0.0060*** −0.0040***
(0.001) (0.001) (0.001) (0.001) (0.002) (0.001)
City 0.0045** 0.0052* 0.0068** 0.0036 0.0079 0.0011
(0.002) (0.003) (0.003) (0.003) (0.005) (0.004)
Married 0.0055 0.0065 −0.0053 0.0053 0.0162 0.0154
(0.008) (0.010) (0.011) (0.010) (0.019) (0.014)
Noc 0.0060* 0.0003 0.0022 0.0082** −0.0003 0.0165***
(0.003) (0.004) (0.004) (0.004) (0.007) (0.006)
Income −0.0015 −0.0005 0.0009 −0.0036** −0.0038 0.0009
(0.001) (0.001) (0.002) (0.001) (0.003) (0.002)
Money left 0.0000 0.0077 0.0127 −0.0142 0.0215 0.0012
(0.008) (0.009) (0.010) (0.009) (0.016) (0.013)
Car 0.0047 −0.0086 0.0198 0.0011 0.0079 0.0116
(0.009) (0.011) (0.012) (0.011) (0.019) (0.015)
Politics −0.0053** −0.0068** −0.0047 −0.0065** 0.0088 −0.0077*
(0.002) (0.003) (0.003) (0.003) (0.005) (0.004)
Native −0.0349*** −0.0200 −0.0361** −0.0219 −0.0824*** −0.0705***
(0.012) (0.015) (0.016) (0.015) (0.028) (0.019)
Smoke −0.0100 −0.0331*** −0.0102 0.0133 −0.0749*** −0.0122
(0.008) (0.009) (0.010) (0.010) (0.017) (0.014)
(Continued)
 Trusting nudges 79

Clusters
(1) (2) (3) (4) (5) (6)
Overall Government Information Default rules Subliminal ads Other
approval campaigns nudges mandates
Alcohol −0.0092*** −0.0101** −0.0152*** −0.0085** −0.0032 −0.0038
(0.003) (0.004) (0.004) (0.004) (0.007) (0.006)
Meat −0.0120*** 0.0054 −0.0087* −0.0102** −0.0128* −0.0481***
(0.004) (0.004) (0.005) (0.004) (0.008) (0.006)
Health 0.0007 0.0075* −0.0040 −0.0021 0.0146** −0.0009
(0.003) (0.004) (0.004) (0.004) (0.007) (0.006)
Swb −0.0057 −0.0092** −0.0037 −0.0059 0.0003 −0.0057
(0.003) (0.004) (0.005) (0.004) (0.008) (0.006)

Notes: “Yos” means “years of schooling”. “Noc” means “number of children”. “Swb” means “subjec-
tive well-being”.

Trust in institutions is highly correlated with approval of nudges. To

predict marginal probabilities of nudge approval, i.e., the predicted size of
the effects, we estimated a logistic model for each nudge question as inde-
pendent variable separately. Predicted marginal probabilities for approval
conditioned on institutional trust—as shown in Figure 6.3—differ substan-
tially between the lowest possible trust score (10) and the highest possible
trust score (70). For instance, while (ceteris paribus) the probability to
accept the nudge “Encouraging green energy” (Nudge 3) is estimated to
be around 55 percent for individuals with the lowest possible value of
institutional trust (for an average individual in the sample), this probability
increases to almost 95 percent for the highest trust value.

Further results
The same is true of environmental concern: results are strong, as expected.
Other results from the regression analysis are worth reporting, though they
are less significant. A higher formal education (years of schooling) is cor-
related with lower approval rates toward nudges on average. City dwellers
tend to approve the tested nudges more than people who live in villages
or in the countryside. The number of children is positively correlated with
approval rates. Those who are left-of-center seem to approve of the tested
nudges more than conservatives do.
Revisiting our categorization of countries regarding nudge approval we
compiled the results from the three study waves (2015, 2016, and 2017/18),
80 Trusting nudges

including 16 countries. Table A6.3 in the appendix gives an overview of sam-

ples and sampling in all countries, with an overall N of 20,501 respondents.
We used the comparable data from these surveys (we lacked some data from
Mexico and Ireland, and also the US was not fully comparable) to run another
regression analysis. Table A6.4 in the appendix shows the weighed OLS
regression for the five nudge clusters. Confirming earlier results, the gender
factor was again found to be highly relevant for nudge approval in each of the
countries—except for China, where male respondents significantly approve
of the tested nudges more than women (this is also visible in Figure A6.1 in
the appendix). However, this should be interpreted in light of extremely high
approval in China from both genders, with rates between 80 and 90 percent.
N1 N2
.95 .9

.9 .8
P(approval=1)

P(approval=1)

.85 .7

.8
.6

.75
.5
.7
.4
10 20 30 40 50 60 70 10 20 30 40 50 60 70
Trust in institutions Trust in institutions

N3 N4
.9
.7

.8
P(approval=1)

P(approval=1)

.6

.7

.5
.6

.5 .4
10 20 30 40 50 60 70 10 20 30 40 50 60 70
Trust in institutions Trust in institutions

N5 N6
.75 .95

.7 .9
P(approval=1)

P(approval=1)

.65 .85

.6 .8

.55 .75

.5 .7
10 20 30 40 50 60 70 10 20 30 40 50 60 70
Trust in institutions Trust in institutions

Figure 6.3  Predicted marginal probabilities for approval, conditional on

institutional trust
 N7 N8
1 .7

.6
.9
P(approval=1)

P(approval=1)
.5

.8
.4

.7 .3

10 20 30 40 50 60 70 10 20 30 40 50 60 70
Trust in institutions Trust in institutions

N9 N10
.6
.9
P(approval=1)

.5

P(approval=1)
.8

.4 .7

.3 .6
10 20 30 40 50 60 70 10 20 30 40 50 60 70

Trust in institutions Trust in institutions

N11 N12
.45 .8

.4 .7
P(approval=1)
P(approval=1)

.35
.6
.3
.5
.25

.2 .4

10 20 30 40 50 60 70 10 20 30 40 50 60 70
Trust in institutions Trust in institutions

N13 N14
.9 .7

.8
P(approval=1)

P(approval=1)

.6
.7

.6 .5

.5
.4
10 20 30 40 50 60 70 10 20 30 40 50 60 70
Trust in institutions Trust in institutions

N15
.7

.6
P(approval=1)

.5

.4

.3
10 20 30 40 50 60 70

Trust in institutions

Figure 6.3  ( Continued)

82 Trusting nudges

We also compared approval rates of the 15 nudges over time in

Denmark, Germany, and South Korea—the three countries that we had
revisited two and a half years after the first survey. Overall, approval rates in
these countries were largely stable as compared to our earlier studies. We
found only small changes in magnitude between those waves, with modest
changes in both directions, i.e., less and more approval by both genders
(Figure A6.2 in the appendix). The country categorizations—Denmark as
a “cautiously pro-nudge country,” Germany as a “principled pro-nudge
nation,” and South Korea as “overwhelmingly pro-nudge” (as explained
in Chapter 5)—still applied three years later. This is particularly notable
for the latter, since the country has undergone a marked democratization
process in these past three years. Finally, the three countries that followed
our methodology to measure their national nudge approval rate—namely
Mexico, Ireland, and Belgium—turned out to be “overwhelmingly pro-
nudge” (in the case of Mexico) and “principled pro-nudge nations” (in the
cases of Belgium and Ireland).

Discussion
In some countries, policymakers have learned to tread around behavio-
ral interventions with caution, in order to avoid being accused of being
“national nannies” or even worse, of manipulating their citizens. Policy
measures that lack public endorsement may well turn out to be less likely
to succeed and to induce the intended behavioral changes without major
unintended side effects. There are also questions about legitimacy, in the
normative as well as the descriptive sense (see Chapter 9).
As in earlier studies, we have found general approval of nudges alongside
marked national differences in levels of support, with Denmark on the least
positive side, South Korea and Mexico the most positive, and Germany,
Belgium, Ireland, and the US somewhere in between. As expected, support
generally seems to decrease as the level of state intervention increases, as
estimated along our five “nudge clusters” of levels of intrusion. (We would
take this finding with caution in view of two facts: (1) much turns on the
substance of the nudge, that is, the direction in which people are being
nudged, rather than the level of intrusiveness; and (2) consistent with (1),
people will approve of high levels of intrusion for egregious kinds of mis-
conduct, e.g., murder, rape, assault, and theft.) In addition, approval is
 T rusting nudges 83

generally higher (or disapproval is lower) for those nudges that are targeted
to others—i.e., businesses—and lower for those that target people directly.
(We would also take this finding with caution. Many nudges applied to
people directly receive high levels of approval; consider calorie labels. At
the same time, some nudges applied to others would not receive approval;
consider a nudge designed to increase air pollution.) Those who engage in
the activity being nudged (e.g., smoking) are less likely to be supportive.
Our particular interest lay in the hypothesis that higher trust in public
institutions will be correlated with stronger support of nudges. This has
been confirmed. At the same time, people who believe in markets as the
best institution to solve environmental and economic problems are more
critical of nudges. Female gender was again found to be correlated with
approval of the tested nudges. Further, people’s own health concern and
health status had no influence on acceptance, and meat consumption only
on the (non)acceptance of the Nudge “meat-free days in cafeterias.” The
fact that approval rates in earlier tested countries have barely changed in
the past three years is noteworthy, particularly in the case of South Korea
where profound political changes have taken place.
For policymakers, our results convey relevant insights. Trust in public
institutions in general and environmental concern might be useful allies in
communicating about nudging and nudges. Endorsement of nudges in gen-
eral might increase when citizens are invited to participate, actively choose,
and offer feedback on planned interventions.11 If they can be expected or
reported, beneficial results in specific domains (health, environment, and
safety) or with respect to specific consumption habits (meat, alcohol, or
smoking) might be helpful in communicating with the public.
For purposes of both effectiveness and legitimacy, close engagement with
the public and attentiveness to its concerns can be exceedingly i­mportant.
It has been urged that a “one-nudge-fits-all” approach to behavioral public
policy is unlikely to be successful.12 Effective nudges, capable of receiv-
ing public acceptance, will more likely be developed with a process that
includes early participation of the affected groups, public scrutiny, and
deliberation—as well as transparent processes in governmental institutions.

1 See Peter John, How Far to Nudge? Assessing Behavioural Public Policy (2018).
1
12 See, e.g., OECD, Behavioural Insights and Public Policy: Lessons from Around the World (2018). See also
Xavier Troussard and René van Bavel, How Can Behavioural Insights be Used to Improve
EU Policy?, Intereconomics 53, 8 (2018).
84 Trusting nudges

In addition to public participation, the “test-learn-adapt-share” approach

called for by leading policy labs worldwide is a prerequisite for success.13
We offer four points by way of conclusion. First, we have confirmed
high levels of approval for nudges as policy tools across different countries
and cultures. Second, Belgium and Ireland join the large set of democratic
nations whose citizens generally embrace nudging, but with important
exceptions and qualifications; Mexico looks more like China and South
Korea, with overwhelmingly high approval rates. Third, levels of public
acceptance are reduced as nudges become more intrusive. Fourth, trust in
government institutions is highly correlated with approval of nudges.
We underline the last point. The best way to obtain trust is to earn it.
In that light, it is important not only to ensure that behaviorally informed
policies promote social welfare, but also to develop processes to ensure that
such polices are adopted transparently, with ample opportunity for public
engagement, and with openness to citizens’ objections and concerns.

Appendix to Chapter 6

Figure A6.1  Nudge approval by gender (all studies)

Note: White shading indicates average male approval.

13 See, e.g., Laura Haynes et al., Test, Learn, Adapt: Developing Public Policy with Randomised Controlled
Trials. Cabinet Office Behavioural Insights Team (2014); Joana Sousa Lourenco, Emanuele
Ciriolo, Sara Rafael Rodrigues Vieira de Almeida and Xavier Troussard, Behavioural Insights
Applied to Policy. Report No. EUR 27726 EN (2016).
Figure A6.1  (Continued)
Figure A6.2  Nudge approval among different time periods for South Korea (K),
Denmark (DK), and Germany (G)
Table A6.1  Samples and sampling in the different countries (2018 study): Types of representativeness and methodology (2018 survey)

Country Data Sample year Unmodified Representativeness Survey Weighting Sample Recruiting for Census/ Frame
provider sample size method method the panel Population of the
survey
Belgium GfK 2017/2018 1,002 Online CAWI No Quota Online 10 mio No
r­ epresentative for weighting sampling ­internet frames
gender, age, region users,18+
and education years
Denmark Qualtrics 2017/2018 966 Online CAWI RIM Quota Online 5.4 mio No
­representative for sampling ­internet frames
gender, age, region users,18+
and education years
Germany Qualtrics 2017/2018 1,535 Online CAWI RIM Quota Online 55 mio No
­representative for sampling ­internet frames
gender, age, region users,18+
and education years
South Qualtrics 2017/2018 1,017 Online CAWI RIM Quota Online 43.9 mio No
Korea ­representative for sampling internet users, frames
gender, age, region 18+ years
and education
USA Qualtrics 2017/2018 1,012 Online CAWI RIM Quota Online 272.4 mio No
­representative for sampling internet frames
gender, age, region users,18+
and education years

Note: “Mio” means million.

Table A6.2  Descriptive statistics—all variables (2018 survey)

N μ σ
Country 5,385.000 2.889 1.386
Gender 5,385.000 0.505 0.500
Age 5,385.000 46.676 16.391
Yos 5,385.000 12.297 4.989
N1 5,385.000 0.789 0.408
N2 5,385.000 0.668 0.471
N3 5,385.000 0.683 0.465
N4 5,385.000 0.537 0.499
N5 5,385.000 0.615 0.487
N6 5,385.000 0.843 0.364
N7 5,385.000 0.864 0.343
N8 5,385.000 0.429 0.495
N9 5,385.000 0.425 0.494
N10 5,385.000 0.777 0.416
N11 5,385.000 0.301 0.459
N12 5,385.000 0.558 0.497
N13 5,385.000 0.651 0.477
N14 5,385.000 0.542 0.498
N15 5,385.000 0.460 0.498
City 5,385.000 3.124 1.575
Married 5,385.000 0.480 0.500
Noc 5,385.000 1.179 1.250
Income 5,385.000 5.574 3.241
money_left 5,385.000 0.584 0.493
Car 5,385.000 0.753 0.431
Politics 5,385.000 3.945 1.360
Native 5,385.000 0.904 0.295
Weight 5,385.000 77.603 19.720
Smoke 5,385.000 0.286 0.452
Alcohol 5,385.000 1.949 1.070
Meat 5,385.000 3.342 1.110
Health 5,385.000 4.826 1.315
Swb 5,385.000 4.882 1.397
job_satisfaction 5,385.000 4.552 1.887
Friends 5,385.000 0.776 0.417
 N μ σ
trust_ggen 5,385.000 3.490 1.517
trust_pgen 5,385.000 4.101 1.322
Environment 5,385.000 4.940 1.439
health_concern 5,385.000 4.479 1.570
health_concernf 5,385.000 4.595 1.505
Markets 5,385.000 3.958 1.414
Risk 5,385.000 3.795 1.435
Freedom 5,385.000 5.012 1.374
Height 5,385.000 171.425 9.661
trustscore_inst 5,385.000 37.462 12.162
trustscore_priv 5,385.000 25.926 6.493
Infoscore 5,385.000 23.866 7.278
N 5,385.000 0.609 0.235
NC1 5,385.000 0.755 0.282
NC2 5,385.000 0.745 0.325
NC3 5,385.000 0.535 0.283
NC4 5,385.000 0.429 0.495
NC5 5,385.000 0.501 0.400
BMI 5,385.000 26.307 6.003

Notes: “Yos” means “years of schooling”. “Noc” means “number of children”. “Swb”
means “subjective well-being”. “Trust_ggen” means “trust in government, in general”.
“Trust_pgen” means “trust in people, in general”. “Health_concernf’” means “concern
about the future health status of friends and relatives”.
Table A6.3  Samples and sampling in the different countries: Types of representativeness and methodology (16 countries, all samples, all waves)

Country Data Sample year Unmodified Representativeness Survey Weighting Sample Recruiting for Census/ Frame of the survey
provider sample size method method the panel Population
Australia Qualtrics 2016 1,001 Online CAWI Target Quota Online 21 mio No frames
r­ epresentative for sampling internet
gender, age, region users,18+
and education years
Belgium GfK 2017/2018 1,002 Online CAWI No Quota Online 10 mio No frames
­representative for weighting sampling internet
gender, age, region users,18+
and education years
Brazil Qualtrics 2016 1,000 Online CAWI Target Quota Online 93 mio No frames
­representative for sampling internet
gender, age, region users,18+
and education years
Canada Qualtrics 2016 1,137 Online CAWI Target Quota Online 29.5 mio No frames
­representative for sampling internet
gender, age, region users,18+
and education years
China Qualtrics 2016 985 Online CAWI Target Quota Online 533 mio No frames
­representative for sampling internet
gender, age, region users,
and education 18+ years
Denmark Qualtrics 2017/2018 966 Online CAWI RIM Quota Online 5.4 mio No frames
­representative for sampling internet
gender, age, region users,18+
and education years
Denmark GfK 2015 1,000 F2f representative CAWI Target Quota Offline 4.54 mio About consumer
for gender, age, omni- sampling internet goods (soft drinks,
region bus users, coffee machines,
18+ years hearing aids) and
crossing the Great
Belt Bridge
France GfK 2015 1,022 F2f representative CAWI Target Quota Online 41.05 mio About views on
for gender, age, omni- sampling (population the
region bus of Ukraine
16–64 years)
Germany Qualtrics 2017/2018 1,535 Online CAWI RIM Quota Online 55 mio No frames
r­ epresentative for sampling internet
gender, age, region users,18+
and education years
Germany GfK 2015 1,012 Online CAWI RIM Quota Offline 55.06 mio About views on
­representative for omni- sampling and internet the
gender, age, region bus online users, economy
14+ years
Hungary GfK 2015 1,001 F2f representative CAWI RIM Quota Offline 7.35 mio, Ad hoc, no other
for gender, age, omni- sampling 15–69 years frames
region bus
Italy GfK 2015 1,011 Online CAWI No Quota Offline 35 mio No frames
­representative for omni- weighting sampling and internet
gender, age, region bus online users, 18–64
years
Japan Qualtrics 2016 1005 Online CAWI Target Quota Online 99 mio No frames
r­ epresentative for sampling internet
gender, age, region users,18+
and education years
(Continued)
Country Data provider Sample year Unmodified Representativeness Survey Weighting Sample Recruiting for Census/ Frame of the survey
sample size method method the panel Population
Russia Qualtrics 2016 918 Online CAWI Target Quota Online 70 mio No frames
r­ epresentative for sampling internet
gender, age, region users,18+
and education years
South Qualtrics 2016 949 Online CAWI Target Quota Online 43.9 mio No frames
Africa ­representative for sampling internet
gender, age, region users,18+
and education years
South Qualtrics 2017/2018 1,017 Online CAWI RIM Quota Online 43.9 mio No frames
Korea ­representative for sampling internet
gender, age, region users,
and education 18+ years
South Qualtrics 2016 932 Online CAWI Target Quota Online 11 mio No frames
Korea ­representative for sampling internet
gender, age, region users,18+
and education years
UK GfK 2015 2,033 F2f representative CAWI RIM Quota Online 50.9 mio About saving and
for gender, age, omni- sampling internet spending habits
region bus users,
18+ years
USA Qualtrics 2017/2018 1,012 Online CAWI RIM Quota Online 272.4 mio No frames
representative for
­
sampling internet
gender, age, region users,18+
and education years

Note: “Mio” means million.

Table A6.4  Weighted OLS regression for different nudge clusters

Clusters
(1) (2) (3) (4) (5) (6)
Overall approval Government campaigns Information nudges Default rules Subliminal ads Other mandates
Male −0.0304*** −0.0133*** −0.0240*** −0.0266*** −0.0544*** −0.0653***
(0.003) (0.004) (0.005) (0.004) (0.008) (0.006)
Age −0.0002 0.0008*** 0.0004*** −0.0011*** −0.0013*** 0.0009***
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
Higher education −0.0108*** 0.0065 −0.0022 −0.0192*** −0.0563*** −0.0015
(0.004) (0.005) (0.005) (0.005) (0.009) (0.007)
Belgium −0.0720*** −0.0508*** −0.1186*** −0.0780*** −0.0090 −0.0479**
(0.012) (0.014) (0.016) (0.015) (0.026) (0.020)
Brazil 0.0908*** 0.0643*** 0.0427*** 0.1394*** 0.1392*** 0.0326*
(0.009) (0.010) (0.011) (0.012) (0.023) (0.017)
Canada −0.0218** −0.0137 −0.0205 0.0029 −0.0756*** −0.0829***
(0.010) (0.012) (0.013) (0.013) (0.022) (0.017)
China 0.1578*** 0.0883*** 0.0799*** 0.2071*** 0.3974*** 0.1113***
(0.009) (0.010) (0.011) (0.011) (0.020) (0.017)
Japan −0.1419*** −0.0841*** −0.1485*** −0.1158*** −0.0710** −0.3320***
(0.012) (0.016) (0.015) (0.016) (0.029) (0.024)
Russia 0.0118 0.0737*** 0.0324** 0.0228* −0.0406 −0.1186***
(0.010) (0.012) (0.017) (0.013) (0.030) (0.022)
South Africa 0.0618*** 0.0562*** 0.0257** 0.0837*** 0.0708*** 0.0538***
(0.009) (0.010) (0.011) (0.012) (0.024) (0.018)
(Continued)
 Clusters
(1) (2) (3) (4) (5) (6)
Overall approval Government campaigns Information nudges Default rules Subliminal ads Other mandates
South Korea 0.0403 ***
0.0343 ***
0.0567 ***
0.0570 0.1989
***
−0.1050***
***

(0.010) (0.011) (0.011) (0.012) (0.021) (0.017)

Denmark −0.1829*** −0.1703*** −0.2395*** −0.1466*** −0.2597*** −0.1875***
(0.009) (0.011) (0.012) (0.012) (0.021) (0.016)
France −0.0385*** −0.0401*** −0.0162 −0.0644*** −0.1089*** 0.0434***
(0.009) (0.011) (0.012) (0.012) (0.023) (0.016)
Germany −0.0709*** −0.0652*** −0.0712*** −0.0879*** −0.1042*** −0.0109
(0.009) (0.011) (0.012) (0.012) (0.022) (0.016)
Hungary −0.1488*** −0.1819*** −0.1685*** −0.1077*** −0.1470*** −0.1941***
(0.011) (0.013) (0.014) (0.013) (0.023) (0.017)
Italy 0.0105 0.0015 −0.0246** 0.0337*** 0.0382 −0.0072
(0.010) (0.012) (0.013) (0.013) (0.023) (0.017)
United Kingdom −0.0217** −0.0350*** 0.0101 −0.0398 ***
−0.0445 **
0.0165
(0.009) (0.010) (0.011) (0.011) (0.021) (0.015)
USA −0.0847*** −0.0625*** −0.1090*** −0.0524*** −0.1494*** −0.1464***
(0.012) (0.014) (0.016) (0.015) (0.026) (0.021)
2015/2016 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
(.) (.) (.) (.) (.) (.)
2017/2018 −0.0142** −0.0250*** −0.0070 −0.0027 −0.0042 −0.0479***
(0.006) (0.008) (0.008) (0.008) (0.013) (0.011)
_cons 0.7333*** 0.8129*** 0.8429*** 0.6870 ***
0.6287 ***
0.6402***
(0.009) (0.010) (0.012) (0.011) (0.021) (0.016)
N 20,501 20,501 20,501 20,501 20,501 20,501
Adj. R2 0.148 0.087 0.089 0.116 0.089 0.088

Notes: “_cons” means the regression “constant” (intercept). It shows the value of the intercept of the model. “Adj. R2” means “adjusted R^2”, a measure
for the goodness of fit of the model. Values closer to 1 indicate a better fit.
 7
Educative nudges and
noneducative nudges

In terms of law and public policy, it is useful to distinguish between edu-

cative and noneducative nudges. Educative nudges include disclosure
requirements, reminders, and warnings, which are specifically designed
to increase people’s own powers of agency. Noneducative nudges include
default rules and uses of order effects (as on a menu or at a cafeteria),
which are designed to preserve freedom of choice, but without necessarily
increasing individual agency.
Our principal question here is whether people prefer educative nudges.
As we will see, the answer is complicated. The simplest version is: They
do. A more accurate and slightly more complex version is: They do, unless
they are told that noneducative nudges are more effective. But the full story
is far more interesting.

Two systems
Within behavioral science, some people have found it helpful to distin-
guish between two families of cognitive operations in the human mind:
System 1, which is fast, automatic, and intuitive; and System 2, which is
slow, calculative, and deliberative.1 When people recognize a smiling face,

1 See Daniel Kahneman, Thinking, Fast and Slow (2011). The idea of two systems is con-
troversial, and it is reasonable to ask what, exactly, the idea is meant to capture. For
96 Educative nudges and noneducative nudges

add three plus three, or know how to get to their bathroom in the middle
of the night, System 1 is at work. When people first learn to drive, when
they multiply 563 times 322, or when they choose a medical plan among
several hard-­to-distinguish alternatives, they must rely on System 2.
System 1 can and often does get things right. As Daniel Kahneman and Shane
Frederick write, “Although System 1 is more primitive than System 2, it is not
necessarily less capable.”2 Through fast and frugal heuristics, people can perform
exceedingly well. Any professional athlete has an educated System 1; Serena
Williams knows what shot to hit in an instant. As a result of years of practice,
an experienced lawyer, judge, doctor, or engineer has a well-­trained System 1,
and trained intuitions are often on the mark. At the same time, System 2 is
hardly unerring. On multiplication problems, or in ­choosing among health
care plans, people often make mistakes, even if they are ­trying very hard.3
Nonetheless, System 1 is distinctly associated with identifiable behavio-
ral biases, producing a wide range of problems for policy and law. People
sometimes show “present bias,” focusing on the short ­term and downplay-
ing the future. Most people tend to be unrealistically optimistic.4 People use
heuristics, or mental shortcuts, that usually work well, but that sometimes
lead them in unfortunate directions. With respect to probability, people’s
intuitions may go badly wrong, in the sense that they produce serious
mistakes, including life-­altering ones.5 To be sure, our intuitions are both
adequate and helpful in the situations in which we ordinarily find ourselves.
But there is no question that intuitions can badly misfire, and that a good
nudge, and good choice architecture, will often provide indispensable help.
Educative nudges, offered by government agencies, attempt to strengthen
the hand of System 2 by improving the role of deliberation and people’s

example, something very different from a two-­system account is offered in E. A. Phelps

et al., Emotion and Decision Making: Multiple Modulatory Neural Circuits, Ann. Rev of
­Neuroscience 263, 37 (2014). Following Kahneman, we understand the idea as a useful fic-
tion, not referring to “systems in the standard sense of entities with interacting aspects
or parts.” Kahneman, supra note, at 29. For those who reject the terminology, or are skep-
tical of it, it might be helpful simply to distinguish between noneducative and educative
nudges, and to see the survey here as asking when people prefer one or another.
2 Daniel Kahneman and Shane Frederick, Representativeness Revisited: Attribute Substitu-
tion in Intuitive Judgment, in Heuristics and Biases: The Psychology of Intuitive Judgement 49, 51
(Thomas Gilovich et al., eds., 2002).
3 See Eric Johnson et al., Can Consumers Make Affordable Care Affordable? The Value of
Choice Architecture, PLOS One 8, 1 (2013).
4 See Tali Sharot, The Optimistic Bias (2011).
5 For a powerful demonstration, see Daniel L. Chen et al., Decision-­Making under the
Gambler’s Fallacy: Evidence from Asylum Judges, Loan Officers, and Baseball Umpires,
The Quarterly J. Econ. 131, 1181 (2016).
 Educative nudges and noneducative nudges 97

considered judgments. The most obvious example is disclosure of relevant

information. Some kinds of nudges, sometimes described as “boosts,”6
attempt to improve people’s capacity to make choices for themselves, for
example by improving statistical literacy.
Noneducative nudges are designed to appeal to, or to activate, System 1.
Graphic health warnings can be seen as an example, at least if they are
not understood as having the purpose or effect of education. We might
distinguish between System 2 disclosures, designed simply to give people
factual information and to ask them to process it, and System 1 disclosures,
designed to work on the automatic system (for example, by inculcating fear
or hope). Some nudges do not appeal to System 1, strictly speaking, but
turn out to work because of its operation—as, for example, where default
rules have large effects in part because of the power of inertia, or where
the ordering of items on some kind of menu affects what people choose,
because of the selective nature of attention.7 Nudges of this kind can be seen
as “exploiting” the operations of System 1, though it would be more neu-
tral to say that they take account of those operations, acknowledging that
some form of choice architecture, affecting System 1, may be inevitable.
As we understand them here, System 2 nudges are specifically designed
to increase people’s capacity to exercise their own agency. On ethical
and other grounds, they might seem better for that reason. Thus Jeremy
Waldron writes: “I wish, though, that I could be made a better chooser
rather than having someone on high take advantage (even for my own
benefit) of my current thoughtlessness and my shabby intuitions.”8
In the abstract, that is an honorable wish. As a matter of principle,
the challenge arises when it is costly and difficult to make people b ­ etter
­choosers—and when the net benefits of a System 1 nudge are far higher than
the net benefits of a System 2 nudge. System 1 nudges, such as a­ utomatic
enrollment, make life much simpler, and that is no small gain. There is also
evidence that System 2 nudges can affect beliefs without affecting behavior,
and that System 1 nudges can be more effective in altering what people actu-
ally do. The choice between System 1 nudges and System 2 nudges raises
pervasive and fundamental questions about agency, freedom, and welfare.

6 Ralph Hertwig, When to Consider Boosting: Some Rules for Policymakers, Behavioural
Public Policy 1, 143 (2017).
7 Eran Dayan and Maya Bar-­Hillel, Nudge to Nobesity II: Menu Positions Influence Food
Orders, Judgement and Decision Making 6, 333 (2011); Daniel R. Feenberg et al., It’s Good to
be First: Order Bias in Reading and Citing, NBER Working Papers (Working Paper No. 21141)
(2015), available at www.nber.org/papers/w21141.
8 See Jeremy Waldron, It’s All For Your Own Good, NY Rev of Books (2014), available at www.
nybooks.com/articles/archives/2014/oct/09/cass-­sunstein-its-­all-your-­own-good/.
98 Educative nudges and noneducative nudges

The project here

Our major goal here is to report the results of a nationally representative
survey in the United States, designed to elicit people’s preferences between
System 1 nudges and System 2 nudges in diverse contexts. We emphasize
that our study is limited to Americans.9 Administered by Survey Sampling
International, the survey included more than 2,800 people, who were paid
for their participation.
In brief, seven different groups, each consisting of more than 400 p ­ eople,
were asked to register their preferences with respect to different pairs of
nudges. Four of those pairs involved areas in which nudges have often been
used as policy tools: savings, smoking, clean energy, and water conserva-
tion.10 In many ways, these pairs can be seen as standard, illustrating, as they
do, dilemmas that can be found in multiple domains. Three of the pairs
involved areas that raise highly distinctive issues and concerns: voter regis-
tration, childhood obesity, and abortion.
With respect to the four initial issues, the first finding is that in a neutral
condition, in which participants receive no information about effectiveness,
majorities prefer System 2 nudges. Notably, Americans are also divided, with
between 26 percent and 45 percent favoring System 1 nudges. In the neutral
condition, two of the four issues produce no significant differences among
Democrats, Republicans, and independents. And while two of the issues did
produce such differences, with a higher percentage of Democrats favoring
System 1 nudges, the differences are relatively small.
The second finding is that when people are asked to assume that the
System 1 nudge is “significantly more effective,” many of them shift in
its direction—but the shift is relatively modest, usually in the vicinity of

9 Several other studies, with different designs, have explored this question. See Gidon
Felsen et al., Decisional Enhancement and Autonomy: Public Attitudes Toward Overt
and Covert Nudges, Judgement and Decision Making 8, 203 (2012) (testing people’s at-
titude toward employment prospects, and finding generally high levels of approval
of System 2 nudges); Janice Jung and Barbara Mellers, American Attitudes Toward
Nudges, Judgement and Decision Making 11, 62 (2016) (finding that on bounded scales,
people generally prefer System 2 nudges); Ayala Arad and Ariel Rubinstein, The
­People’s Perspective on Libertarian Paternalistic Policies, J. Law and Economics 61(2),
311–333 (2018) (finding evidence of “reactance” against System 1 nudges and some
inclination to prefer System 2 nudges). The findings here are broadly compatible
with those in these earlier papers.
10 We borrow the water conservation example from Janice Jung and Barbara Mellers,
American Attitudes Toward Nudges, Judgement and Decision Making 11, 62 (2016).
 Educative nudges and noneducative nudges 99

about 12 percentage points. The third finding is that when people are asked
to assume specific numbers, offering a quantitative demonstration that
System 1 nudges are more effective, the shift in their direction is essentially
the same in magnitude. The fourth and final finding is that when people
are asked to assume that System 2 nudges are “significantly more effec-
tive,” there is no shift in the direction of those nudges. This is an especially
surprising finding, and we will attempt to explain it.
The most obvious interpretation of these findings is that in important
contexts, most people want to protect and promote people’s agency, and
so they will favor System 2 nudges—but they also care about effectiveness,
and so will turn to System 1 nudges if the evidence shows that they are sig-
nificantly better. At the same time, there is significant heterogeneity within
the American population. Many people prefer System 1 nudges, perhaps on
the ground that they are more effective, perhaps on the ground that they
make life simpler and easier. Some people appear not to have any abstract
preference between System 1 nudges and System 2 nudges; apparently they
care only about effectiveness.
By contrast, some people have a strong preference for the latter, and
will require compelling evidence of superior effectiveness in order to
favor System 1 nudges. Because significant numbers of Americans show
no inclination to prefer System 1 nudges even when asked to assume that
they are clearly more effective, we can safely say that some segment of
the population would demand very powerful evidence to favor System 1
nudges—and perhaps no evidence would be sufficient.
With respect to voter registration, childhood obesity, and abortion,
the patterns are illuminatingly different. For the first two, majorities do
not favor System 2 nudges. On the contrary, automatic voter registration
has clear majority support, and for childhood obesity, cafeteria design is
preferred to parental education. Asking people to assume the significantly
greater effectiveness of the System 1 nudge does increase the level of sup-
port, but it is high even without that information. The best explanations
for the preference for System 1 nudges involve people’s judgments about
protection of the franchise (arguing in favor of automatic registration) and
protection of children (favoring cafeteria design).
With respect to reducing the number of abortions, majorities consist-
ently favor System 2 nudges, and that preference does not shift when people
are asked to assume that System 1 nudges are more e­ ffective—undoubt-
edly because of a belief, on the part of many, that it is not appropriate for
1 00 Educative nudges and noneducative nudges

public officials to appeal to System 1 to d ­ iscourage women from making

their own choices. Notably, Republicans, Democrats, and independents all
favor System 2 nudges in the abortion setting, though in most conditions,
the level of support for System 1 nudges is significantly lower among
Democrats. The sharp distinction between majority approval of a System 1
nudge for voting and majority approval of a System 2 nudge for abortion
attests to the importance of people’s judgments about whether a right is at
stake—and whether a nudge is promoting or undermining it.
These findings support a variety of conclusions. In significant domains,
majorities will prefer System 2 nudges to System 1 nudges, but there is
likely to be real division on that issue. If System 1 nudges are shown to
be more effective, there will be a shift in their direction, but it will not
be as dramatic as might be anticipated, apparently because some people
put a high premium on personal agency. Insofar as children are involved,
System 1 nudges will be more welcome, and the same is true if System
1 nudges facilitate people’s ability to enjoy something that qualifies as
a right. If, on the other hand, any kind of nudge is compromising what
people regard as a right, it will be rejected, and a System 2 nudge will be
preferred, because it shows greater respect for individual agency.
In important respects, the survey findings are consistent with what
emerges from a more sustained analysis of the normative issues; we sketch
the central ingredients of that analysis here. Both the findings and the analysis
bear on a variety of issues in law and policy. They suggest reasons to prefer
System 2 nudges, such as disclosure of statistical information or some kind
of education, but they also suggest that in many contexts, System 1 nudges,
such as default rules, are preferable. System 1 might be inclined to favor
System 2 nudges, and System 2 will frequently concur, but for many rights
and interests, System 2 will ultimately decide that System 1 nudges are best.

Savings, smoking, and the environment

The first four questions asked people to choose among some familiar interven-
tions, in areas that are familiar in law and policy.11 Here are the four pairs:

1
1 See, e.g., Christine Jolls, Product Warnings, Debiasing, and Free Speech: The Case of Tobacco
Regulation, J. Institutional and Theoretical Econ. 53, 169 (2013); Raj Chetty et al., Active vs. Passive De-
cisions and Crowd out in Retirement Savings Accounts: Evidence from Denmark (Working Paper No. 18565)
(2012), www.nber.org/papers/w18565; Felix Ebeling and Sebastian Lotz, Domestic Up-
take of Green Energy Promoted by Opt-­Out Tariffs, Nature Climate Change 5, 868 (2015).
 Educative nudges and noneducative nudges 101

Which of these policies do you prefer, as part of an antismoking campaign?

1) Graphic warnings, with vivid pic- 2) Purely factual information, giv-
tures of people who are sick from ing people statistical information
cancer. about the risks from smoking.
Which of these policies do you prefer, as part of a campaign to encourage
people to save for retirement?
1) Automatic enrollment of employ- 2) Financial literacy programs at the
ees in savings plans, subject to workplace, so that employees
“opt out” if employees do not are educated about retirement
want to participate. options.
Which of these policies do you prefer, as part of a program to reduce
pollution?
1) Automatic enrollment of customers 2) Educational campaigns so that
in slightly more expensive “green” consumers can learn the advan-
(environmentally friendly) energy, tages of green (environmentally
subject to “opt out” if customers friendly) energy.
want another, slightly less expen-
sive energy source.
Which of these policies do you prefer, as a way of encouraging water
conservation?
1) The government requires hotels 2) The government requires hotels
to select a default policy of to provide guests with informa-
“environment-­friendly rooms” in tion about an “environment-­
which towels left on the racks are friendly” policy in which towels
not washed. If people want their left on the racks are not washed.
towels washed, they can tell the People are encouraged to choose
front desk, and their towels will to take part, but if they do not
be washed daily. choose to do so, their towels will
be washed every day.

Neutral condition
In a neutral condition, in which people were provided with no informa-
tion about effectiveness, majorities consistently showed a clear preference
for System 2 nudges. The aggregate data is presented in Table 7.1.
It should be noted that the preference for System 2 nudges is strongest
in the cases of green energy and water conservation. With respect to green
energy, the likely theory is that it is better for people to learn, and to make
1 02 Educative nudges and noneducative nudges

Table 7.1  Support for System 1 and System 2 nudges

Percentage of respondents (N=430) who:

Question Prefer System 1 nudge Prefer System 2 nudge
Smoking 45% 55%
Saving 43% 57%
Energy 26% 74%
Water 32% 68%

their own choices, than for them to be defaulted into an energy source that
might turn out to be more expensive or less reliable. Participants might
well have been concerned that people would not take the trouble to opt
out and might thus have faced higher electricity bills without their explicit
consent. In the case of water conservation, money is not involved, but
more people also favored System 2 nudges, perhaps because of a concern
about defaulting guests into a situation that might not be in their interest
(involving unwashed and perhaps dirty towels).
It should also be noted that while the preference for System 2 nudges is
consistent, large numbers of people favor System 1 nudges in all four con-
texts. One reason might be that they believe them to be more effective, so
long as no information is provided on that question. The 45 percent level
of support for graphic warnings for cigarettes might well have been based
on a judgment that if the goal is to address a serious public health problem,
such warnings are better than purely factual information. Another reason
might be that some System 1 nudges might seem to impose lower decision-­
making burdens on choosers, as in the cases of default rules for saving,
energy, and water conservation. If a System 1 nudge makes things easier for
people, and does not require them to act, it might appear to be preferable.

System 1 nudges “significantly more effective”

In the neutral condition, people’s preferences could have any number of
sources. To obtain some understanding of what motivated those prefer-
ences, a different group of people was asked to assume that the System 1
nudge was “significantly more effective.” The hypothesis was that these
three words would lead to a major shift in the direction of System 1 nudges.
The hypothesis was confirmed, but in a qualified way (see Table 7.2).
There are two noteworthy results here. First, the shift in the direction of
System 1 nudges is significant for all four issues (using chi-­square analysis,
 E ducative nudges and noneducative nudges 103

Table 7.2  Preference when System 1 nudge “significantly more effective”

Percentage of respondents who prefer System 1 nudge

Question Neutral condition Told that System 1 nudge is “significantly more
(N = 430) effective” (N = 407)
Smoking 45% 57%
Saving 43% 55%
Energy 26% 38%
Water 32% 42%

two-­tailed p < 0.05 for each question); but it is not massive. Second, the
shift is essentially the same for all four questions. Indeed, it is remarkably
consistent, with no significant differences across questions. Informed of
the greater effectiveness of System 1 nudges, there is a general movement
of 10 to 12 percentage points.
We do not have enough data to speak of anything like an iron law here, but
it is not too speculative to say that many Americans think that System 2 nudges
will be more effective or instead believe that System 2 nudges are a better way
of respecting people’s agency—but they will shift when they receive informa-
tion about the comparative effectiveness of System 1 nudges. At the same time,
many people (usually 40 percent or more) will have a strong commitment,
visceral or otherwise, to the superiority of the System 2 nudge. The addition of
three words (“significantly more effective”) will not change that commitment.

Numbers
The words “significantly more effective” have a high degree of opacity. It
is not clear what they mean. Once they are specified in quantitative terms,
they might have a stronger or weaker impact. Suppose, for example, that
people were told to assume that automatic enrollment would increase par-
ticipation in savings plans from 40 percent to 90 percent, or that graphic
warnings would save 200,000 lives annually—but that the System 2 alterna-
tives would have essentially no impact. Under those assumptions, it would
not be easy to reject the idea that System 1 nudges are better. To reject that
idea, one would have to have concerns about the outcomes (saving lives is
good, but increased participation in savings plans is less obvious), or to put
a very high premium indeed on a certain conception of personal agency.
To understand the effects of quantitative information, participants
were asked to assume specified numerical disparities in favor of System
1 04 Educative nudges and noneducative nudges

1 nudges—not as stark as in the examples just given, but nonetheless

lopsided.

•• Smoking: “Assume that [the System 1 nudge] is far more effective. It

reduces smoking by 20 percent, while [the System 2 nudge] reduces
smoking by 5 percent.”
•• Saving: “Assume that [the System 1 nudge] is far more effective. It
leads 90 percent of workers to enroll in savings plans, whereas [the
System 2 nudge] leads only 55 percent to enroll in such plans.”
•• Energy: “Assume that [the System 1 nudge] is far more effective. It
cuts pollution by 40 percent, whereas [the System 2 nudge] cuts pol-
lution by just 5 percent.”
•• Water: “Assume that [the System 1 nudge] is far more effective. On aver-
age it cuts water use from washing towels by 70 percent, whereas [the
System 2 nudge] cuts water use from washing towels by 10 percent.”

The results are illustrated in Table 7.3.

It should be clear that the quantitative information did not have a larger
effect than the words “significantly more effective.” Quite surprisingly,
that information produced no statistically significant changes. One reason
may be that the numerical differences were not so extreme; they plausibly
reflected the kind of disparity that a purely qualitative account (“significantly
more effective”) would suggest. If so, the numbers provided no additional
information. Another reason may be that the people who favored System
2 nudges, even in the face of a qualitative explanation that it would be less
effective, did so because of a strong preference for what they saw as personal
agency, and hence could not be moved even by fairly impressive numbers.

Table 7.3  Preference for System 1 nudges with quantitative information

Percentage of respondents who prefer System 1 nudge

Question Neutral Told that System 1 nudge is Told that System 1 nudge is more
condition “significantly more effective” effective, with quantification
(N = 430) (N = 407) (N = 435)
Smoking 45% 57% 58%
Saving 43% 55% 56%
Energy 26% 38% 43%
Water 32% 42% 47%
 Educative nudges and noneducative nudges 105

System 2 nudges “significantly more effective”

If it is assumed that System 2 nudges are “significantly more effective,” we
might expect that very large majorities would endorse them. If a nudge
increases people’s own capacities and also produces the desired result, it
would seem far preferable to a less effective intervention that does not
educate people in any way. The principal qualification is that if a nudge
is effective in producing a result that people do not like, then they will of
course reject it for that very reason. (Most people would not like a nudge
that is effective in encouraging people to use illegal drugs or to text while
driving.) We will return to this point. The results are presented in Table 7.4.
Most surprisingly, the assumption of the comparatively greater effec-
tiveness of System 2 nudges does not produce any shift in their direction.
The numbers are essentially identical—a highly unexpected finding. Any
explanation remains speculative, but it is possible that those who supported
System 2 nudges already assumed that they would be more effective, so
that the three additional words added no new information.
Alternatively, some people might think that System 1 nudges have some
independent advantage, or that System 2 nudges have some independent
disadvantage. System 1 supporters might have stronger preference than
System 2 supporters and thus are less likely to be persuaded by effectiveness
arguments. Automatic enrollment in a savings plan might be more desir-
able than financial literacy programs, simply because they do not impose
the costs and burdens of the latter. The same is true of automatic enroll-
ment in green energy. It is also true that most people already prefer System
2 nudges and hence fewer people are available to be moved.
If so, then the lack of an effect from the System-­2-nudge-­is-more-­
effective assumption is similar to the lack of effect of quantitative

Table 7.4  P reference for System 1 nudges when System 2

nudge is “significantly more effective”

Percentage of respondents who prefer System 1 nudge

Question Neutral condition Told that System 2 nudge is
“significantly more effective”
Smoking 45% 43%
Saving 43% 44%
Energy 26% 26%
Water 32% 29%
1 06 Educative nudges and noneducative nudges

information. Some people prefer System 1 nudges even if they believe

them to be less effective. They might be engaged in some kind of infor-
mal cost–­benefit analysis—a point to which we will return.

Political divisions
Do political affiliations explain people’s preferences for System 1 or System
2 nudges? The results are presented in full in Tables 7.512, 7.6, 7.7, 7.8.13
There are many numbers here, but the basic story is straightforward.
Republicans, Democrats, and independents all favor System 2 nudges, with
just one qualification: Democrats are evenly split with respect to anti-­smoking
nudges. Both qualitative and quantitative information about the greater effec-
tiveness of System 1 nudges produces an increase of about 10 to 20 percent in
favor of System 1 nudges—and essentially the same degree of change is observed
for all three groups. For all three groups, assuming that System 2 nudges are

Table 7.5  Support for System 1 and System 2 nudges by partisan affiliation

CONDITION 1 Percentage of respondents11 who:

Question Prefer System 1 nudge Prefer System 2 nudge
Dem. Rep. Indep. Dem. Rep. Indep.
Smoking 50% 44% 40% 50% 56% 60%
Saving 42% 48% 39% 58% 52% 61%
Energy 34% 24% 19% 66% 76% 81%
Water 42% 27% 26% 58% 73% 74%

Table 7.6  Preference when System 1 nudge “significantly more effective” by

partisan affiliation

CONDITION 2 Percentage of respondents12 who:

Question Prefer System 1 nudge Prefer System 2 nudge
Dem. Rep. Indep. Dem. Rep. Indep.
Smoking 62% 57% 52% 38% 43% 48%
Saving 60% 55% 49% 40% 45% 51%
Energy 48% 31% 34% 52% 69% 66%
Water 51% 36% 38% 49% 64% 62%

12 163 Democrats; 142 Republicans; 125 independents.

13 163 Democrats; 142 Republicans; 125 independents.
 Educative nudges and noneducative nudges 107

Table 7.7  Preference with quantitative information by partisan affiliation

CONDITION 3 Percentage of respondents13 who:

Question Prefer System 1 nudge Prefer System 2 nudge
Dem. Rep. Indep. Dem. Rep. Indep.
Smoking 61% 56% 56% 39% 44% 44%
Saving 58% 51% 57% 42% 49% 43%
Energy 47% 38% 42% 53% 62% 58%
Water 52% 41% 48% 48% 59% 52%

Table 7.8  Preference when System 2 nudge is “significantly more effective” by

partisan affiliation

CONDITION 4 Percentage of respondents14 who:

Question Prefer System 1 nudge Prefer System 2 nudge
Dem. Rep. Indep. Dem. Rep. Indep.
Smoking 53% 35% 39% 47% 65% 61%
Saving 47% 37% 47% 53% 63% 53%
Energy 28% 24% 25% 72% 76% 75%
Water 41% 20% 23% 59% 80% 77%

significantly more effective produces results quite similar to those in the neutral
condition. Perhaps most remarkably: None of the differences between condi-
tion 1 and condition 4, for any partisan affiliation, is ­statistically significant.
The largest and most important finding here is that in a majority of the
conditions, the differences among Democrats, Republicans, and independ-
ents are not significant. Their judgments, as between System 1 and System
2 nudges, are broadly in line with one another. But in some conditions,
Democrats are more inclined to System 1 nudges than are Republicans and
independents. In all conditions, for example, Democrats are more favora-
bly disposed than Republicans or independents to a System 1 nudge for
water conservation; the difference is significant (p <0.05). In condition
2, Democrats are more favorably disposed than Republicans or independ-
ents to a System 1 nudge for energy. In condition 4, Democrats are more
favorably disposed than Republicans to a System 1 nudge for smoking.1415

14 165 Democrats; 138 Republicans; 132 independents.

15 169 Democrats; 131 Republicans; 133 independents.
1 08 Educative nudges and noneducative nudges

We can offer some plausible explanations for these differences.

Democrats are comparatively more enthusiastic about green energy and
water conservation, and very possibly anti-smoking efforts as well; to
them, a System 1 nudge might seem more appealing if it is thought to
be more effective. Republicans might be more likely to favor a System
2 nudge, especially for green energy or water conservation, in order to
preserve personal agency. Notably, however, there are no significant
differences among the three groups in terms of movements across condi-
tions.

Voting, children, and abortion

The range of System 1 and System 2 nudges is of course exceptionally
wide. For example, some nudges promote rights, by making them eas-
ier to enjoy; consider, for example, simplified voter registration. Some
nudges involve children. Teachers impose mandates on elementary
school children, but they also nudge them, in various ways, to do their
homework, to act courteously, and to avoid disrupting classes. Some
nudges discourage the use of rights. We could easily imagine efforts to
steer people away from certain religious practices, or to discourage them
from exercising their right to sexual privacy; pro-­abstinence nudges are
an example.
As illustrations of these distinctive kinds of nudges, we tested people’s
judgments about voting, childhood obesity, and abortion. The three pairs
looked like this:

Which of these policies do you prefer, as part of a program to increase voter

registration?
1) Automatic voter registration, so 2) A public education campaign to con-
that when people receive their vince people to register to vote.
driver’s licenses, and show they are
domiciled in your state, they are
automatically registered as voters.
Which of these policies do you prefer, as part of a program to combat
childhood obesity?
1) Redesigning school cafeterias so 2) Educating parents about the problem
that healthy, low-­calorie options of childhood obesity and how to
are in the most visible locations. combat it.
 Educative nudges and noneducative nudges 109

Which of these policies do you prefer, as a means of discouraging abortions?

(Please indicate which you prefer even if you do not like either.)
1) Requiring pregnant women, 2) Requiring pregnant women, before
before having an abortion, to see having an abortion, to speak briefly
vivid photos of fetuses, designed with a doctor about whether they
to show that they are merely very really believe, on reflection, that an
young children. abortion is the right choice, in light of
the moral issues involved.

General results
Most Americans prefer automatic voter registration to efforts to encourage
people to register. With the three variations, the changes in assumptions
are not statistically significant (condition 1 is neutral; condition 2 assumes
that System 1 nudge is significantly more effective; condition 3 assumes
quantitative evidence of the superior effectiveness of System 1 nudges;
condition 4 assumes that System 2 nudge is significantly more effective)
(see Table 7.9).
There are two noteworthy findings here. The first is that a majority pre-
fers the System 1 nudge. The reason is probably a judgment or intuition
that people should be voters “by default”; they should not have to take
steps to attain that status. The second is that with respect to automatic voter
registration, movements do not occur across the three conditions, as they
sometimes do for the four nudges previously discussed. In particular, we
observe movements from condition 1 to conditions 2 and 3 for the four
standard nudges; no such movements are found here.
For childhood obesity, a small majority favors cafeteria design over
parental education, except when people are asked to assume that the latter
is significantly more effective, as is shown in Table 7.10.

Table 7.9  Voter registration

Voter registration Percentage of respondents who:

Condition Prefer System 1 nudge Prefer System 2 nudge
1 57% 43%
2 62% 38%
3 61% 39%
4 52% 48%
110 Educative nudges and noneducative nudges

Table 7.10  Childhood obesity

Childhood obesity Percentage of respondents who:

Condition Prefer System 1 nudge Prefer System 2 nudge
1 53% 47%
2 53% 47%
3 63% 37%
4 48% 52%

Table 7.11  Abortion

Abortion Percentage of respondents who:

Condition Prefer System 1 nudge Prefer System 2 nudge
1 25% 75%
2 34% 66%
3 33% 67%
4 29% 71%

In the three “informed” conditions, only one shift is significant: With

quantitative information, there is an increase in support for cafeteria design.
In general, we do not find the same kinds of shifts as are observed for the
four more standard nudges.
For abortion, the System 2 nudge is preferred by a substantial majority.
(Note that for this question, it was specifically requested that participants
chose one even if they do not like either, acknowledging that on that highly
sensitive issue, some respondents might reject both nudges.) The prefer-
ence for the System 2 nudge does not shift significantly across the four
conditions (see Table 7.11).

Political divisions
What is the role of political divisions? We might well expect that it would
be larger than in the four standard cases, and in some respects it is, but the
full story is not entirely straightforward, the results for which are in Tables
7.12, 7.13, and 7.14.
The most consistent difference can be found in the area of abortion,
where higher percentages of Republicans are more inclined to favor the
 Educative nudges and noneducative nudges 111

Table 7.12  Voter registration by partisan affiliation

Voter registration—by party Percentage of respondents who:

Condition Prefer System 1 nudge Prefer System 2 nudge
Dem. Rep. Indep. Dem. Rep. Indep.
1 65% 53% 52% 35% 47% 43%
2 63% 53% 57% 37% 47% 43%
3 71% 54% 58% 29% 46% 42%
4 55% 53% 48% 45% 47% 52%

Table 7.13  Childhood obesity by partisan affiliation

Childhood obesity—by party Percentage of respondents who:

Condition Prefer System 1 nudge Prefer System 2 nudge
Dem. Rep. Indep. Dem. Rep. Indep.
1 61% 45% 51% 39% 54% 49%
2 59% 48% 52% 41% 52% 48%
3 70% 52% 67% 30% 48% 33%
4 51% 43% 49% 49% 57% 51%

Table 7.14  Abortion by partisan affiliation

Abortion—by party Percentage of respondents who:

Condition Prefer System 1 nudge Prefer System 2 nudge
Dem. Rep. Indep. Dem. Rep. Indep.
1 20% 32% 23% 80% 68% 77%
2 30% 42% 29% 70% 58% 71%
3 25% 43% 31% 75% 57% 69%
4 24% 34% 30% 76% 66% 70%

System 1 nudge. In three of the four conditions, the difference for that
question is statistically significant16 between Democrats and Republicans
(condition 4 is the exception). For both voting and childhood obesity,
the difference between Democrats and Republicans is significant for

16 Using chi-­square analysis, two-­tailed p < 0.05 for each question.

112 Educative nudges and noneducative nudges

conditions 1 and 3. Democrats and independents show a significant dif-

ference for voting in conditions 1 and 3. Republicans and independents
show such a difference for abortion in conditions 2 and 3; Republicans
and independents show such a difference for childhood obesity in con-
dition 3. Interestingly, there is no significant difference of any kind in
condition 4.
Here as well, the details should not obscure the basic story. For the
four standard cases, political affiliation usually did not explain ­people’s
choices between System 1 and System 2 nudges (with interesting excep-
tions, especially in the case of green energy). Political affiliation mattered
more for abortion, voting, and childhood obesity. It would not, of
course, be surprising to find that Democrats are less supportive of pro-­
life nudges than are Republicans, or that Republicans are less enthusiastic
than Democrats about automatic voter registration or efforts to combat
childhood obesity (see Chapter 2). The point is that in politically con-
tested issues, there is a partisan difference in terms of choice between
System 1 and System 2 nudges. Apparently it is the case that if people
strongly support a particular end, they will be more likely to support a
System 1 nudge to attain it.

Within-subjects
The foregoing findings involved a “between subjects” design. Different
groups of participants saw different conditions, rather than all of them at
once. That design has significant advantages, because it prevents contami-
nation by previous answers. If questions are seen in isolation, responses
cannot be affected by order effects, or by a particular factor that becomes
highlighted only by virtue of its clear difference from a previous question.
Nonetheless, there are advantages to a “within-subjects” design, by
which participants see, and answer, all of the questions in the same survey.
The principal advantage is that with a within-­subjects design, it is possible
to test whether people’s original answers shift after they are given informa-
tion about comparative effectiveness. That question is important to test,
because it reveals whether some people are strongly committed to System
1 or System 2 nudges—so committed that they will stick with one or the
other even when effectiveness information stares them in the face.
To explore that issue, we used Amazon Mechanical Turk to ask about
400 people 24 questions, involving all of the areas tested above (with the
 E ducative nudges and noneducative nudges 113

exception of abortion17). Notably, this is not a random sample, and hence

it would be hazardous to draw conclusions about the sources of differences
between this group and the nationally representative one. With that caveat,
the results are in Table 7.15.
Several things are relatively clear. In general, the answers in the neutral
condition are fairly close to what was found in the nationally representative
sample. At the same time, and as expected, the movements across conditions
are somewhat greater. In particular, we observe movements of at least 25 per-
cent from condition 1 to condition 3 for four more standard questions. And
for all questions, movements of at least 11 percent, and sometimes of more
than 20 percent, can be found from condition 1 to condition 4. In this survey,
the differences among the conditions produced larger shifts in people’s views.
There are two important qualifications. First, at least one-­third of the
population continued to favor the System 2 nudge in the within-­subjects
condition, even when they were given numbers to support the compara-
tively greater effectiveness of the System 1 nudge. Second, large percentages
of people (usually around one-­quarter) continued to favor the System 1
nudge in the within-­subjects condition, even when they were informed
that the System 2 nudge was significantly more effective.
Although it does not involve a nationally representative sample, the within-­
subjects study provides useful information. It suggests that the number of

Table 7.15  Within-subjects results

Percentage of respondents who prefer System 1 nudge

Question Neutral Told that System 1 Told that System 1 Told that System 2
condition nudge is “sig- nudge is “significantly nudge is “significantly
nificantly more more effective,” with more effective”
effective” quantification
Smoking 41% 57% 67% 30%
Saving 45% 58% 72% 28%
Energy 36% 50% 69% 19%
Water 42% 55% 67% 21%
Obesity 61% 71% 78% 29%
Voting 60% 66% 76% 34%

17 This area was excluded on the ground that it is highly controversial, and it is not clear
how much more would be learned in a within-­subjects design.
114 Educative nudges and noneducative nudges

people who shift to System 1 nudges will probably be greater in a within-­

subjects design—and that the appeal of the System 2 nudge will be heightened,
in that design, with evidence of greater effectiveness. At the same time, it for-
tifies the general conclusion that a certain percentage of the population will
favor System 2 nudges even if they are significantly less effective, in large part
because of a commitment to a certain conception of individual agency.

Popular opinion, law, and public policy

As we have emphasized, survey evidence can tell us a great deal about what
kinds of policies will produce public approval or disapproval. We know
that when people are not asked to choose between System 1 nudges and
System 2 nudges, and are simply asked whether they approve of a nudge,
they tend to favor both, including graphic warnings for smoking and
distracted driving, and automatic enrollment in savings plans and green
energy. We also know that Americans reject nudges that reflect what they
see as illicit ends (such as religious favoritism) or that are inconsistent with
the values and interests of most choosers (such as automatic contributions
to particular charities). There is a concern about manipulation, at least in
extreme cases, such as subliminal advertising to discourage smoking or use
of visual illusions to encourage drivers to slow down.
We have also emphasized that the results of surveys may or may not
track what would emerge from a sustained analysis. In fact we do not
know exactly how people are thinking when they respond to survey ques-
tions. Consider an admittedly speculative hypothesis: System 1 prefers System
2 nudges. On this view, the automatic system favors System 2 nudges, and
the deliberative system is necessary to override that form of favoritism. The
data here are not sufficient to support that hypothesis; people’s preference
for System 1 nudges might be deliberative rather than automatic. But the
hypothesis cannot be ruled out of bounds.
Consider another hypothesis: System 2 favors System 1 nudges. The data here
are also insufficient to support that hypothesis, though it is imaginable that
a careful analysis of various situations would suggest that System 1 nudges
often or generally work better.
How should regulators think about the choice, having as they do a tool-
box of instruments? It is possible, of course, that mandates will be better
than nudges of any kind; perhaps they will be more effective and have
higher net benefits. It is also possible that inaction is best, because any new
 Educative nudges and noneducative nudges 115

intervention would have costs in excess of benefits. Economic incentives

might be the best approach of all. But in many contexts, policymakers must
specifically decide between System 1 nudges and System 2 nudges. To pro-
mote savings, they might engage in an educational campaign or opt for
automatic enrollment; to promote access to public programs, they might
rely on education or new default rules; to discourage smoking or distracted
driving, they might rely on graphic warnings or a statistical presentation of
some kind. How should they decide?
To come to terms with that question, it would be helpful to specify
the foundations for any answer. Suppose that we are welfarists, believing
that any evaluation has to turn on the effects of an intervention on social
welfare.18 If so, helpful questions are: What are the costs and what are the
benefits of a System 1 or System 2 nudge? Which has higher net benefits?
For these questions, information about effectiveness is relevant, but it is
hardly sufficient. We need cost information as well. A maximally effective
nudge might be too costly to be worthwhile, or it might have lower net
benefits than a somewhat less effective but far less costly nudge.
In addition, the effectiveness information does not, by itself, give a full
account of benefits. If 90 percent of people end up in savings plans, or if
automatic enrollment in green energy cuts pollution by 20 percent, what
exactly are the welfare consequences? Increases in participation rates and
reductions in pollution seem desirable, but a great deal of further work
would be needed to understand exactly how desirable they are. Are increases
in participation rates important? How important? What are the mortality
and morbidity consequences of cutting pollution levels by 20 percent? In
this respect, the survey questions, even in the various conditions, failed to
provide respondents with relevant information. It would be interesting,
for example, to see how people’s view would switch if they were asked
to assume that for smoking, a System 1 nudge would prevent 4,000 more
premature deaths than a System 2 nudge.
At first glance, welfarists would have no systematic reason to prefer System
1 nudges to System 2 nudges, or vice-­versa. Everything turns on their costs
and benefits.19 But second-­order considerations might cut in one or another

18 A welfarist approach is used in Ryan Bubb and Richard Pildes, How Behavioral Econom-
ics Trims Its Sails and Why, Harv. L. Rev. 127, 1593 (2014).
19 If a System 1 nudge causes a welfare loss because people resent it, that loss would
of course have to be included. Cf. Sarah Conly, Against Autonomy (2012), at 156–59
(­cataloguing welfare effects of soda regulation).
116 Educative nudges and noneducative nudges

direction. Part of the welfare calculation involves the cost of nudging itself.
Under imaginable conditions, System 1 nudges can be far simpler to imple-
ment (as, for example, when they involve a mere default rule). At the same
time, it is relevant to ask about the long-­term effects of a nudge. If a System 2
nudge would educate people, and have beneficial effects in multiple domains
of their lives, then it would have ancillary benefits, and they might turn out
to be significant. It is doubtful that survey responses are adequately capturing
these points, though some respondents might be attentive to them.
Suppose that we are not welfarists and that we believe that for reasons that
involve dignity or autonomy, people ought to be active agents, affirmatively
responsible for outcomes that affect their lives. To be sure, this idea has con-
siderable ambiguity, but something of this sort undergirds the judgment that
even if automatic enrollment of some kind can promote people’s welfare, it
is more respectful to them, and therefore best for them, to become informed
and then to choose.20 Perhaps what is wrong with paternalism, even of the
choice-­preserving kind, is that it is insulting to people’s capacity for agency;
perhaps it shows a form of disrespect.21 Why not educate people, rather than
enrolling them in a program that government thinks is in their interest?
Different people who press this question might accept diverse kinds of
answers. Some people might agree that if automatic enrollment is signifi-
cantly better on welfare grounds, it is not necessary or preferable to educate
people; but they would insist that the government must meet the burden
of demonstrating that it is significantly better. Other people would adopt a
strong presumption in favor of educative approaches and would demand
an exceptionally strong demonstration of higher net benefits. Still others
might believe that in at least some contests, no such demonstration could
justify a System 1 nudge. A continuum of beliefs might well be imagined.
Such a continuum would, of course, fit with the results here.

Agency
In some circles, there is a strong preference for interventions that augment
people’s capacities, and skepticism about forms of choice architecture that

20 Cf. Nicholas Cornell, A Third Theory of Paternalism, Mich. L. Rev. 113, 1295 (2015) (argu-
ing that paternalism shows such disrespect).
21 Id. A powerful response can be found in Conly, supra note 105, arguing that it is not dis-
respectful for government to act on the basis of an accurate understanding of people’s
capacities.
 Educative nudges and noneducative nudges 117

seem to exploit or take advantage of people’s fallibility. If a default rule

works because of inertia, for example, it might be seen to be a form
of manipulation, and even if that charge is far too strong, some people
might contend that it is best to rely on education. On one view, the choice
between System 1 and System 2 nudges depends on an assessment of
comparative welfare effects, which requires a form of cost–benefit analy-
sis.22 On another view, concerns about autonomy and dignity deserve a
central place.
Our major goal here has been to investigate what people actually think
about these questions. A central finding is that most people usually do pre-
fer System 2 nudges, at least in the class of cases that were tested. Moreover,
the preference cuts across partisan lines. When participants are told to
assume that System 1 nudges are more effective, many of them are in
their direction—usually between 10 and 14 percent. When they are given
quantitative information, specifying the greater effectiveness, the shift is
essentially identical. And when people are asked to assume that System 2
nudges are significantly more effective, their judgments are about the same
as in the neutral condition—a most unexpected finding.
Political differences emerge in several contexts. For example, Democrats
are more inclined to favor System 1 nudges in the contexts of green energy
and water conservation. But the more dramatic finding is that in general,
Democrats, Republicans, and independents show strikingly similar patterns
of responses. They tend to favor System 2 nudges, at least in the standard
cases: to shift by the same percentages when they are asked to assume that
the System 1 nudge is significantly more effective and that it has specified
numerical advantages; and to show the same numbers as in the neutral
condition when they are asked to assume that the System 2 nudge is sig-
nificantly more effective.
As we have seen, different subject areas elicit different responses. If
­people care greatly about the end, perhaps effectiveness is all that matters,
and the issue of agency will seem beside the point. For example, a System
1 nudge to reduce criminal violence might be preferred purely on effec-
tiveness grounds, and people will not much care that a System 2 nudge
preserves people’s capacity to exercise their own agency (to murder or to
rape). We can easily go further: If people are outraged by the conduct that
is being targeted (murder, rape), and if they want to eliminate it, a man-

22 See Cass R. Sunstein, The Cost–Benefit Revolution (2018)

118 Educative nudges and noneducative nudges

date will be entirely acceptable, and a System 1 nudge, complementing that

mandate, will be unobjectionable in principle.
To the extent that the issue is polarizing on political grounds, we might
also expect to see polarized judgments about which kind of nudge to favor.
The abortion example is exemplary on that ground. We have seen that while
most Democrats and most Republicans favor System 2 nudges to reduce
abortions, the percentage of Republicans who favor System 1 nudges is sig-
nificantly higher. For green energy and water conservation, there is a similar
difference, but in the opposite direction. If people question or do not like
the ends of those who deploy nudges, they might end up preferring System
2 nudges, because they seem better on autonomy grounds.
It should therefore be unsurprising that in some conditions we also find
partisan differences with respect to voter registration and childhood obe-
sity. These findings have large implications for judgments about nudging
in general. They suggest that we will find comparative receptivity to System
1 nudges when the ends seem desirable and when people trust the officials
who seek to secure them—and comparative skepticism about System 1
nudges when the ends seem questionable or the officials untrustworthy.
The most interesting question involves the precise tradeoff between sac-
rificing a degree of personal agency (as System 1 nudges might be taken
to do) and increasing effectiveness. People put different weights on agency
and understand it in different ways, and some of them will demand a steep
price, in terms of effectiveness, in order to compromise it. Here as well,
context matters, and so the value placed on agency will be high for some
populations (with respect to, say, the abortion right), whereas it will be
low for those very populations (with respect to, say, voter registration).
The value of agency varies across persons and contexts.
Less obviously, and more intriguingly, the same is true for its sign. For
some of the subjects of nudging, the exercise of agency is a cost rather
than a benefit; voter registration is the most prominent case in point. A
form of choice architecture that respects rights, and that does not require
people to take action to enjoy them, might be strongly favored, simply
on the ground that it makes things easy. The example suggests a larger
point. System 1 might tend to prefer System 2 nudges, and System 2
might agree, but after sustained analysis, System 2 will often conclude
that System 1 nudges are best.
 8
Misconceptions

Our principal goal here has been to present information about public
opinion. We have only glanced at normative questions. But by way of
clarification, we think that it will be useful to catalogue some common
mistakes and misconceptions. Unfortunately, they continue to divert atten-
tion both in the public domain and in academic circles, and hence to stall
progress. The good news is that survey evidence suggests that when people
are asked concrete questions, they do not fall victim to these misconcep-
tions. Abstractions appear to cause the trouble.
Without further ado:

1. Nudges are an insult to human agency.

In free societies, people are treated with respect. They are allowed to go
their own way. Some people object that nudges are troublesome because
they treat people as mere objects for official control.
The objection is off the mark. One of the main points of nudging is to pre-
serve freedom of choice—and thus to maintain people’s capacity for agency.
Many nudges are self-consciously educative, and hence they strengthen that
very capacity: consider calorie labels, or warnings about risks associated
with certain products. With information, warnings, and reminders, people
120 Misconceptions

are in a better position to choose their own way. Noneducative nudges, such
as uses of healthy choice architecture at cafeterias or in grocery stores, also
allow people to choose as they wish. Survey evidence, suggesting widespread
approval of both educative and noneducative nudges, testifies to general
appreciation of these points.
Perhaps it could be argued that if the goal is to promote agency, default
rules are problematic. But because such rules are omnipresent in human
life, it is not easy to make that argument convincing. Would it make sense
to excise default rules from the law of contract? To say that employers,
hospitals, and banks are forbidden from using default rules? In practice,
what would that even mean? Those who are inclined to reject default rules
out of respect for individual agency would do well to ponder the countless
contexts in which such rules make life simpler and easier to navigate. (On
the immense importance of navigability, more in a moment.)

2. Nudges are based on excessive trust in government.

An intuitive objection to nudging is rooted in fear of government. To put that

objection in its sharpest form: Suppose that public officials are incompetent,
self-interested, reckless, or corrupt. Suppose that your least favorite leaders are
or will be in charge. Would you want them to nudge? Or suppose that you are
keenly alert to public choice problems, emphasized by James Buchanan and his
followers, or “the knowledge problem,” emphasized by Friedrich Hayek and
his followers.1 If interest groups are able to push government in their preferred
directions, and if public officials lack crucial information, then you might
insist: Do not nudge! Reliance on private markets might seem far better.
Indeed, behavioral science itself might be taken to put this conclusion in
bold letters. There is no reason to think that public officials are immune to
behavioral biases. In a democratic society, the electoral connection might
mean that they will respond to the same biases that affect ordinary people. To
be sure, structural safeguards might help, especially if they ensure a large place
for technocrats, insistent on science and on careful attention to costs and ben-
efits. But in any real-world polity, behavioral distortions are difficult to avoid.
These are fair and important points, but if they are taken as an objection to
nudging, they run into a logical problem: a great deal of nudging is inevita-
ble. So long as government has offices and websites, it will be nudging. If the
law establishes contract, property, and tort law, it will be nudging, if only

1 Friedrich Hayek, The Use of Knowledge in Society, Am. Econ. Rev. 35, 519 (1945).
 M isconceptions 121

because it will set out default rules, which establish what happens if people
do nothing. As Hayek himself wrote, the task of establishing a competitive
system provides “indeed a wide and unquestioned field for state activity,”
for “in no system that could be rationally defended would the state just do
nothing. An effective competitive system needs an intelligently designed and
continuously adjusted legal framework as much as any other.”2
As Hayek understood, a state that protects private property and that
enforces contracts has to establish a set of prohibitions and permissions,
including a set of default entitlements, establishing who has what before
bargaining begins. For that reason, it is pointless to exclaim, “do not
nudge!”—at least if one does not embrace anarchy.
The second answer to those who distrust government is that because
nudges maintain freedom of choice, they offer a safety valve against official
error. Those who favor nudges are keenly alert to the public choice prob-
lem and the knowledge problem, and to the possibility that public officials
will show behavioral biases. Many of them are influenced by Buchanan and
(especially) Hayek. If one distrusts government, the real focus should be on
mandates, bans, subsidies, and taxes. To be sure, nudges ought not to be
free from scrutiny, but they should be a relatively lower priority.
It is true, of course, that some nudging is optional. Government can
warn people about smoking, opioid addiction, and distracted driving, or
not. It can seek to protect consumers against deception and manipulation,
or not. It can undertake public education campaigns, or not. If you think
that government is entirely untrustworthy, you might want it to avoid
nudging whenever it can.
In the abstract, that position cannot be ruled out of bounds. Public choice
problems, and the knowledge problem, are real and important. On highly pes-
simistic assumptions about the capacities and incentives of public officials, and
highly optimistic assumptions about the capacities and incentives of those in
the private sector, nudging should be minimized. But private actors nudge,
and sometimes it is very much in their interest to exploit cognitive biases, thus
causing serious harm to countless people. Would it be a good idea to forbid
public officials from taking steps to reduce smoking and distracted driving? In
any case, the track record of real-world nudging includes impressive success
stories, if success is measured by cost-effectiveness.3

2 Friedrich Hayek, The Road to Serfdom 88 (1943).

3 Shlomo Benartzi et al., Should Governments Invest More in Nudging?, Psychol. Sci. 28, 1041
(2017).
122 Misconceptions

To be sure, nudges, like other interventions from such officials, should

be constrained by democratic requirements, including transparency, public
debate, and independent monitoring (including continuing evaluation of
how they work in practice). Constraints of this kind can reduce the risks
(without eliminating them). The fundamental point is that those risks are
far larger with other tools, above all mandates and bans.

3. Nudges are covert.

Some people have argued that mandates, bans, and taxes have one advan-
tage: They are transparent. People know what they are. No one is fooled.
By contrast, nudges are covert and in that sense sneaky, a form of trickery.4
They affect people without their knowledge.
For countless nudges, this objection is hard to understand. A GPS device
nudges, and it is entirely transparent. Labels, warnings, and reminders are
not exactly hidden; if they are, they will not work. When an employer auto-
matically enrolls employees into a savings plan, subject to opt out, nothing
is hidden. (If it is, there is a problem; the right to opt out should be clear.)
Why, then, have intelligent people objected that nudges are covert? Is
there anything at all to that objection? One possibility is that when people
participate in a randomized controlled trial, they may not be informed of
that fact. (A randomized trial might not work if people are told about the
various conditions.) But we suspect that the real answer is that some nudges
work even though those who are affected by them do not focus on them,
or even think about them. While such nudges are hardly hidden, people
may be unaware of them, or at least unaware of their purposes and effects.
For example, a cafeteria might be designed so that the healthy foods are
most visible and placed first, and people might choose them for that very
reason. Such a design is not hidden—on the contrary, it should be o ­ bvious—
but people may not be aware that their cafeteria has been designed so as to
promote healthy choices. To be sure, they know that the fruits are more vis-
ible than the brownies, but they might not know why, and their d ­ ecision to
select a fruit might be quick and automatic rather than reflective. Or ­people
might not think much about the default rules that come with (say) an agree-
ment with a rental car company. If people are automatically enrolled into
some kind of insurance plan and allowed to opt out, they might say, “yeah,
whatever,” and simply go along with the default.

4 Edward Glaeser, Paternalism and Policy, U. Chi. L. Rev. 73, 133 (2006).
 M isconceptions 123

In that sense, it is correct to say that some nudges can work even if or
perhaps because people are unaware that they are being nudged. Note,
however, that emerging evidence finds that the effects of such nudges are
not diminished even if people are told that nudging is at work. Though
research continues, transparency about the existence and justification of
default rules appears not to reduce their impact in general.5 For some
people, such clarity may even increase that impact, by amplifying the infor-
mational signal that some default rules offer.6 On plausible assumptions,
drawing attention to the healthy design of a cafeteria will actually increase
the effect of that design, because it will convey valuable information. (To
be sure, it may produce “reactance” in some consumers.)

4. Nudges are manipulative.

In a variation on the claim that nudges are covert, some people have objected
that nudges are a form of manipulation. But return to the points we have
just explored: If people are reminded that they have a doctor’s appointment
next Thursday, no one is manipulating them. The same is true if people are
given information about the caloric content of food or if they are warned
that certain foods contain shellfish or nuts, or that if they take more than the
recommended dosage of Benadryl, something bad might happen.
To be sure, most people reject subliminal advertising, apparently on the
ground that it is manipulative. And we could imagine a graphic warning
about opioid addiction, or about the use of cell phones while driving, that
would create immediate fear or revulsion, or intensely engage people’s
emotions; it might be objected that nudges of this kind count as a form of
manipulation. To know whether they do, we need a definition of manipula-
tion. To make a (very) long and complex story short, philosophers and others
have generally converged on the view that an action counts as manipulative
if it bypasses people’s capacity for rational deliberation.7 On any view, most
nudges do not qualify. True, some imaginable nudges might cross the line,
but that is very different from saying that nudges are manipulative as such.

5 Hendrick Bruns et al., Can Nudges Be Transparent and Yet Effective?, Journal of Economic
Psychology (2018); George Loewenstein, Warning: You Are About to be Nudged. Behavioral
Sci. and Pol’y, 1, 35 (2015).
6 Craig McKenzie et al., Recommendations Implicit in Policy Defaults, Psychol. Sci. 17, 414
(2006).
7 Anne Barnhill, What is Manipulation? In Manipulation: Theory and Practice 50, 72 (Christian
Coons and Michael Weber, eds., 2014). Barnhill’s own account is more subtle.
124 Misconceptions

5. Nudges exploit behavioral biases.

Some people object that nudges “exploit” or “take advantage of” behavio-
ral biases. Indeed, some people define nudges as exploitation of behavioral
biases.8 That does sound nefarious. But the objection is mostly wrong, and
while people can define terms however they wish, this particular definition
is a recipe for confusion.
Many nudges make sense, and help people, whether or not a behavioral
bias is at work. A GPS is useful for people who do not suffer from any such
bias. Disclosure of information is helpful even in the absence of any bias.
A default rule simplifies life and can therefore be a blessing whether or not
a behavioral bias is involved. As the GPS example suggests, many nudges
have the goal of increasing navigability—of making it easier for people to get to
their preferred destination. Such nudges stem from an understanding that
life can be either simple or hard to navigate, and a goal of helpful nudging
is to promote simpler navigation.
At the same time, it is true that some nudges counteract behavioral biases,
and that some nudges work because of behavioral biases. For example, many
human beings tend to suffer from present bias, which means that they give
relatively little weight to the long term; many of us suffer from unrealistic
optimism, which means that we tend to think that things will turn out better
for us than statistical reality suggests. Some nudges try to counteract present
bias and optimistic bias—for example, by emphasizing the long-term risks
associated with smoking and drinking, or by suggesting the importance of
retirement planning. Similarly, default rules work in part because of inertia,
which undoubtedly counts as a behavioral bias. But it is misleading—a form of
rhetoric, in the not-good sense—to suggest that nudges “exploit” such biases.

6. Nudges wrongly assume that people are irrational. Some critics object
that nudges are based on a belief that human beings are “irrational,”
which is both insulting and false.9 This objection takes different forms.

8 Ricardo Rebonato, Taking Liberties (2012).

9 The most peculiar version of this claim comes from a psychologist: “The interest in
nudging as opposed to education should be understood against the specific political
background in which it emerged. In the US, the public education system is largely con-
sidered a failure, and the government tries hard to find ways to steer large sections of
the public who can barely read and write. Yet this situation does not apply everywhere.”
Gerd Gigerenzer, On the Supposed Evidence for Libertarian Paternalism, Rev. Phil. and
Psychol. 3, 361 (2016). We offer no comment, except to add that we are unaware of any
public officials in the US who have tried hard, or at all, or ever, “to find ways to steer
large sections of the public who can barely read and write.”
 M isconceptions 125

In one form, the objection is that while people rely on simple heuris-
tics and rules of thumb, nothing is wrong with that; those heuristics and
those rules work well, and so nudging is not needed, and can only make
things worse. In another form, the objection urges that the whole idea of
nudging is based on weak psychological research and on an assortment
of supposed laboratory findings that do not hold in the real world. In yet
another form, the objection is that people can and should be educated
rather than nudged.
In its best form, the objection urges that people’s utility functions are
complex and that outsiders may not understand them; what seems to be
“irrationality” may be the effort to trade off an assortment of goals. A mun-
dane example: People might eat fattening foods not because they suffer
from present bias, but because they greatly enjoy those foods. A less mun-
dane example: People might fail to save for retirement not because they
suffer from optimistic bias, but because they need the money now.
No one should doubt that heuristics generally work well (that is why
they exist); but they can also misfire. When they do, a nudge can be
exceedingly helpful. Many nudges are developed with reference to well-
established behavioral findings, demonstrating that people depart from
perfect rationality. For example, default rules work in part because of the
power of inertia; reminders are necessary and effective in part because
­people have limited attention; information will be more likely to influ-
ence behavior if it is presented in a way that is attentive to people’s
imperfect information-processing capacities. These and other claims
are based on evidence, both in the laboratory and the real world. (It is
always possible that they will be found to be imprecisely stated, or
wrong in important settings.) But those who embrace nudges do not
use the term “irrationality.” In fact they abhor it; “bounded rationality”
is much better. Nor does anyone doubt that education can work. As we
have emphasized, many nudges are educative. More ambitious educative
efforts, such as efforts to help people to assess risks and to teach statisti-
cal literacy, are usually complements to nudges, and rarely substitutes or
alternatives.
It is also true (and exceedingly important) that people’s utility functions
are complex and that outsiders might not understand them; that is one rea-
son why nudgers insist on preserving freedom of choice. To the extent that
nudging is inevitable, it is pointless to contend that because of the com-
plexity of people’s utility functions, nudging should be avoided. To the
extent that nudging is optional, it should be undertaken with an apprecia-
126 Misconceptions

tion of the risk of error and with careful efforts to ensure that it promotes,
and does not undermine, people’s welfare. A GPS device does not decrease
welfare. In general, information about health risks and potential financial
burdens should increase welfare.
Of course nudges must be tested to ensure that they are doing what they
are supposed to do. Some nudges fail. When they do, the right conclusion
may be that freedom worked—or that we should nudge better.

7. Nudges work only at the margins; they cannot achieve a great deal.

If experts are asked to catalog the world’s major problems, many of them
would single out poverty, malnutrition and hunger, unemployment, cor-
ruption, diseases, terrorism, and climate change. On one view, nudges are
an unfortunate distraction from what might actually help. With an under-
standing of nudging, we might have some fresh ideas about how to tweak
letters from government to citizens, producing statistically significant
increases in desirable behavior. But that is pretty small stuff. If behavioral
economists want to make a contribution, shouldn’t they focus on much
more important matters?
It is true that behaviorally informed approaches are hardly limited to
nudges; mandates, bans, and incentives may well have behavioral justi-
fications. The policy program of behavioral science is not exhausted by
nudges.10 It is also true that some nudges produce only modest changes.
But in multiple domains, nudges have proven far more cost-effective than
other kinds of interventions, which means that per dollar spent, they have
had a significantly larger impact.11
By any measure, the consequences of some nudges are not properly described
as modest. As a result of automatic enrollment in free school meals programs,
more than 11 million poor American children are now receiving free breakfast
and lunch during the school year. Credit card legislation, enacted in 2010, is
saving American consumers more than ten billion US dollars annually; signifi-
cant portions of those savings come from nudges and nudge-like interventions.12

10 Richard Thaler, Much Ado About Nudging, Behavioral Public Policy Blog (2017), https://­
bppblog.com/2017/06/02/much-ado-about-nudging/.
11 Shlomo Benartzi, et al., Should Governments Invest More in Nudging?, Psychological Science
28, 1041 (2017).
12 Sumit Agarwal et al., Regulating Consumer Financial Products: Evidence from Credit Card. Work. Pap.,
NBER (2013).
 Misconceptions 127

With respect to savings, automatic enrollment in pension programs has pro-

duced massive increases in participation rates.13
New nudges, now in the early stages or under discussion, could also
have a major impact. If the goal is to reduce greenhouse gas emissions,
automatic enrollment in green energy can have large effects in many
nations.14 The Earned Income Tax Credit and its variations rank among the
world’s most effective anti-poverty program, but many eligible people do
not take advantage of it. Automatic enrollment would have large conse-
quences for the lives of millions of people. With respect to the most serious
­problems, the use of nudges remains in its preliminary stages. We will see
far more in the future, and the impact will not be small.
It is true, of course, that for countless problems, nudges are hardly
enough. They cannot eliminate poverty, unemployment, and corruption.
But by itself, any individual initiative—whether it is a tax, a subsidy, a
mandate, or a ban—is unlikely to solve large problems. Denting them
counts as an achievement.

13 Raj Chetty et al., Active vs. Passive Decisions and Crowdout in Retirement Savings
­Accounts: Evidence from Denmark, available at www.nber.org/papers/w18565 (2012);
Richard Thaler, Much Ado About Nudging., Behavioral Public Policy Blog, https://bppblog.
com/2017/06/02/much-ado-about-nudging/ (2017).
14 Felix Ebeling and Sebastian Lotz, Domestic Uptake of Green Energy Promoted by ­Opt-Out
Tariffs, Nature Climate Change 5, 868 (2015), available at doi: 10.1038/nclimate2681;
­Daniel Pichert and Konstantinos Katsikopoulos, Green Defaults: Information Presentation
and Pro-environmental Behaviour, Journal of Environmental Psychology 28, 63 (2008).
 9
A Bill of Rights for
Nudging

The idea of “legitimacy,” central to contemporary legal theory and political

philosophy, can be taken in two different ways. It can be seen as a purely
descriptive term: Does the citizenry actually believe that a government or a
policy is legitimate? It can also be seen as a normative term: Is a government
or a policy legitimate in principle? Under either understanding, of course, the
notion of legitimacy needs further specification. But however the notion is
specified, it captures the view that governments and policies need to receive,
and should deserve to receive, some kind of consent from those who are sub-
ject to them. Constitutions are often designed to promote legitimacy in both
the descriptive and the normative sense. Bills of Rights, frequently included
in constitutions, sometimes define the idea of political legitimacy.
Discussions of legitimacy sometimes focus on the topic of individual rights,
emphasizing freedom of speech, freedom of religion, and the right to due
process of law. We could easily imagine nudges that would compromise such
rights; consider, for example, a default rule that presumed that citizens were
Christians, and that they intended to vote for the current leader.
Discussions of legitimacy also focus more broadly on the topic of
coercion, and in particular on coercion from government, which has
 A Bill of Rights for Nudging 129

a monopoly on the legitimate use of force. How can government legitimately

require citizens to act or to refrain from acting? Serious answers draw on a range
of philosophical traditions. Whether Kantian, Aristotelian, Lockean, or
Benthamite, those answers converge on the view that it is legitimate to
prevent harm to others (and thus to forbid murder, assault, or rape) and to
solve collective action problems (and thus to provide for national defense
and to combat pollution). Many nudges help in those endeavors, but it is
generally agreed that they are not sufficient; they are complementary to
more aggressive approaches, including uses of the criminal law.
We could easily distinguish among nudges in terms of the kind of
problem that they are intended to solve: harm-to-others nudges; collec-
tive action problem nudges; coordination problem nudges; harm-to-self
nudges. Some of the nudges tested here involve harm to others; consider
automatic enrollment in green energy. It is also important to see that
behaviorally informed approaches need not be nudges.
For example, a tax on sugary beverages might be defended as a way of
counteracting present bias and unrealistic optimism on the part to consum-
ers, and thus of protecting them against their own errors. Fuel economy and
energy efficiency standards might be justified as a way of protecting con-
sumers against their tendency to neglect the long term. In a succinct account
in 2012, the United States Environmental Protection Agency referred to
“inadequate consumer attention to long-term effects of their decisions, or a
lack of salience of benefits such as fuel savings to consumers at the time they
make purchasing decisions.”1 In a longer account in 2010, the EPA said this2:

The central conundrum has been referred to as the Energy Paradox in this
setting (and in several others). In short, the problem is that consumers
appear not to purchase products that are in their economic self-interest.
There are strong theoretical reasons why this might be so:
– Consumers might be myopic and hence undervalue the long-term.
– Consumers might lack information or a full appreciation of informa-
tion even when it is presented.

1 See 2017 and Later Model Year Light-Duty Vehicle Greenhouse Gas Emissions and
­Corporate Average Fuel Economy Standards, Final Rule, 77 Fed. Reg. 62624, 63114,
(2012), available at www.gpo.gov/fdsys/pkg/FR-2012-10-15/pdf/2012-21972.pdf.
2 See Light-Duty Vehicle Greenhouse Gas Emission Standards and Corporate Average Fuel
Economy Standards; Final Rule, Part II, 75 Fed. Reg. 25,324, 25,510–11 (May 7, 2010),
available at www.gpo.gov/fdsys/pkg/FR-2010-05-07/pdf/2010-8159.pdf. Under
President Donald Trump, the EPA has proposed to repudiated this analysis and asked for
public comments on how to think about consumer savings.
130 A B ill of R ights for Nudging

– Consumers might be especially averse to the short-term losses

associated with the higher prices of energy efficient products relative to
the uncertain future fuel savings, even if the expected present value of
those fuel savings exceeds the cost (the behavioral phenomenon of “loss
aversion”).
– Even if consumers have relevant knowledge, the benefits of energy-
efficient vehicles might not be sufficiently salient to them at the time of
purchase, and the lack of salience might lead consumers to neglect an
attribute that it would be in their economic interest to consider.

We offer these examples only to suggest the wide range of approaches

that might emerge from an engagement with behavioral findings. Our
own surveys explore only a subset of those approaches. Nonetheless, we
think that they tell us a great deal about principles of legitimacy—most
obviously in the descriptive sense, but if we have any faith at all in the
wisdom of crowds, in the normative sense as well. Many governments
are bound by constitutions, of course, and constitutional understandings
might well restrict behaviorally informed approaches. For example, the
German Constitution requires respect for “human dignity,” and in that
nation, all such approaches must be consistent with that requirement. The
United States Constitution requires states to provide “equal protection of
the laws,” understood as a broad prohibition on discrimination; nudges,
like mandates and bans, could run afoul of that prohibition.

Toward a Bill of Rights for Nudging

Drawing on our findings, we suggest six principles of legitimacy—a kind
of Bill of Rights for Nudging. We emphasize that as for other Bills of
Rights, the items on the list should be taken as broad principles, rather
than as concrete specifications. They remain to be given concrete content
through engagement with particular cases. In some cases, they should be
treated only as presumptive, and as subject to override on the basis of a
compelling justification. For example, the right to freedom of speech does
not include bribery, perjury, or criminal conspiracy, and under emergency
conditions, the police do not need to get a warrant before searching a
home. Nonetheless, the rights to freedom of speech, and to protection
from unreasonable searches and seizures, have immense importance. We
think that the same is true of the items on this list—understanding them
as directed at public officials who nudge, and not for judicial enforcement.
 A Bill of Rights for Nudging 131

Public officials must promote legitimate ends

People approve of nudges that promote legitimate ends; they disapprove
of nudges that promote illegitimate ends. In terms of designing any Bill of
Rights, that simple intuition is an excellent place to start.
A mandate, a ban, or a tax might be an effort to insulate public officials
and to protect their power; consider a prohibition on dissent. A nudge
could easily fall in the same category. We could easily imagine a nudge that
would be designed to discourage dissent. As we have seen, the vast major-
ity of Americans reject nudges of this kind, and we expect that the same
would be true in most and perhaps all of the nations explored here. Apart
from self-insulation, nudges could be designed to protect racial, ethnic, or
religious majorities, or to favor men over women. They could be designed
to undermine liberty. Such nudges would run afoul of the requirement of
legitimate ends.
The issue becomes more interesting, of course, when there is a dispute
about what counts as a legitimate end. Is it legitimate to nudge people to
be heterosexual? To believe in God? Different people, and majorities in dif-
ferent nations, are likely to offer different answers to such questions, and
to many people, the answers are obvious; they essentially need no defense.
Our goal here is not, of course, to reach conclusions about what count as
legitimate ends. It is only to suggest that insofar as there is legal or social
clarity on that topic, it creates a barrier against some nudges.

Nudges must respect individual rights

This idea can easily be seen as a corollary of the prohibition on the pursuit
of legitimate ends, but it deserves separate recognition. It is meant to create
a kind of second-order right—a right to ensure respect for rights. As we
have emphasized, nudges respect freedom of choice, which will often make
it harder to see them as rights violations. But our surveys, and some imagi-
nable variations, show that even when nudges allow people to go their own
way, they may violate rights. We would not support the design of ballots
by which current leaders print their name in big, bold, pleasing letters, and
their opponents’ names in small, obscure, ugly letters. A political leader
may certainly campaign on his own behalf, and use behavioral strategies
to nudge people to vote for him; but he may not create a voting system
by which those who fail to vote at all are counted as having voted for him.
132 A B ill of R ights for Nudging

It remains necessary, of course, to specify the category of individual

rights, and to decide whether nudges interfere with them. Different nations
may arrive at different judgments on that question. But the general point
is clear.

Nudges must be consistent with people’s values and interests

Most of the nudges tested here are designed to protect people against their
own errors; consider calorie labels, anti-smoking campaigns, and automatic
enrollment in savings programs. The overwhelming majority of respond-
ents embrace nudges when they are consistent with people’s values and
interests—and reject them when they are not.
This principle too can be specified in many different ways, and we could
easily imagine cross-cultural variations in the preferred specification. Is it
legitimate to attempt to nudge women to embrace traditional gender roles?
We do not think so, but there is hardly an international consensus on that
score. What matters, for our purposes, is the apparent consensus on the gen-
eral principle, though different specifications would be intriguing to elicit.
We add as well that the principle is emphatically not designed to forbid
public officials from leading rather than following public opinion. In some
cases, officials might believe, for example, in protection against discrimi-
nation on the basis of ethnicity or religion, and might insist on nudging
people not to discriminate on those grounds even if they believe that such
discrimination is consistent with their values and very much in their inter-
ests. The question whether and when public officials should be able to
depart from what (some or many) people believe to be their values or their
interests is obviously a delicate one, turning on the grounds on which the
departure might be justified. Such officials certainly have a burden of justi-
fication. We notice the issue without resolving it here.

Nudges must not manipulate people

In most nations, official manipulation of citizens encounters a kind of
taboo, and it often has a constitutional source. In diverse nations, we have
found widespread (though not universal) disapproval of subliminal adver-
tising, even when it is meant to achieve legitimate ends. The disapproval
might well be rooted in a commitment to individual agency: People are
entitled to make up their own minds, exercising their own faculties, and it
 A Bill of Rights for Nudging 133

is not permissible for officials to try to manipulate them. Indeed, it is possi-

ble that some people would disapprove of manipulation where they would
approve of coercion. We expect that many people who would be willing
to accept, or even embrace, a prohibition on the use of drugs (cocaine or
heroin) would disapprove of an official effort to use subliminal advertising
for that purpose.
We have stressed that the idea of manipulation is not self-defining. There
is an elaborate philosophical literature on the subject, with an emphasis on
efforts to subvert or undermine, or at least a failure to respect, p ­ eople’s
capacity for reflective choice.3 In addition, some people, committed to
some version of utilitarianism, would not impose a taboo on manipula-
tion; they would allow it if it is appropriate or necessary to increase social
welfare.4 Let us bracket the complexities here and note only that in most of
the nations explored here, manipulation creates serious concern, and there
is at least a presumptive principle against it.

In general, nudges should not take things from people, and give them
to others, without their explicit consent
A principle of this general kind does emerge from our findings; con-
sider widespread disapproval of charitable donations by default and of
making people organ donors by default. For that reason, it must be
included, but it should be treated with considerable caution as we do
not know its boundary conditions. For example, most people do not
oppose the tax system, even though people do not pay taxes voluntarily. In
some nations, military service is obligatory, and that obligation seems
to run afoul of the principle. We are not sure whether majority opposi-
tion to charitable donations by default is a product of a principle of this
kind, or a narrower understanding that charitable donations must be
understood as donations, that is, an intentional gift from one person to
another. If so, the idea of charitable donations by default is objection-
able for that very reason.

3 For a number of instructive treatments, see Manipulation (Christian Coons and Michael
Webster, eds., 2014).
4 See Jonathan Baron, A Welfarist Approach to Manipulation, Journal of Marketing Behavior 1,
283 (2016).
134 A B ill of R ights for Nudging

Nudges should be transparent rather than hidden

The widespread opposition to subliminal advertising is most naturally
taken as a prohibition on manipulation, but it can also be taken to support a
related principle, which is that nudges, like other interventions, should not
be hidden or covert. While our findings do not compel the conclusion that
citizens would embrace that principle, we like it, and so we include it here.
Like the previous principles, it requires specification. Transparent about
what? A sweet-free cashiers area is obviously transparent in a relevant sense;
the area has no sweets. Must it also be transparent in the sense that pub-
lic officials justify the nudge by reference to the behavioral findings that
underpin it? We think that the answer is “yes”—both to ensure that the
justification is subject to public scrutiny (and so corrected if it is wrong)
and also to treat citizens with respect.
An important qualification involves randomized controlled trials. Many
nudges, and many other interventions, are tested in that way, by compar-
ing the outcome of a control condition with that of a treatment condition
(in which people might be nudged). While a trial is ongoing, it would
likely be self-defeating to disclose its existence to participants. In that sense,
transparency is not required. But after the trial has taken place, it is impor-
tant not to hide it, and to ensure that people generally (and participants in
particular) are able to learn what happened. Some kind of public registry
might be a good idea for that purpose.

Of welfare and autonomy

A Bill of Rights for Nudging grows fairly directly out of our empirical findings.
There are, of course, broader questions about how to evaluate behaviorally
informed policies, going well beyond the idea of a Bill of Rights. We close
with two candidates. The first involves autonomy. The second involves
welfare. The two candidates have powerful theoretical foundations in the
Western political tradition; they resonate in other political traditions as well.
The idea of autonomy is of course sharply contested, and we want to
avoid controversial philosophical claims here. Does autonomy require free-
dom of choice? Always? Does it require us to attend to the background
conditions under which people form their preferences and values?5 If so,

5 For relevant discussion, see Jon Elster, Sour Grapes (1983).

 A Bill of Rights for Nudging 135

what exactly does that mean? Does it mean that we should dismiss or fail
to respect judgments that are based on a lack of information or that come
from a problem of self-control?6 Are some judgments nonautonomous?
Let us bracket these questions. Note first that insofar as we are speaking
of nudges, autonomy is preserved in an important sense: People are allowed
to do as they wish. They are not forbidden or coerced. But the discussion
thus far—and some of our central findings—should be sufficient to show
that this is not enough. If people are deceived or manipulated, it is fair to
say that their autonomy has been violated. We can also imagine a default
rule that would be questionable on grounds of autonomy, and it is now
possible to see why. If people lose some right (say, to religious freedom)
or some interest (say, to their property) by default, it might well be right
to say that their autonomy has been compromised. And if people have not
been clearly informed that they have a right to opt out, a default rule might
not be altogether different from a mandate or a ban. This point suggests
that if autonomy matters, we need to attend closely to the circumstances
that allow opt-in or opt-out. Insofar as opt-out rights are concealed, and
insofar as opting out is difficult or surrounded by serious burdens, a default
rule might undermine autonomy even if it preserves freedom of choice.
The idea of welfare is also contested, and here too, we bracket philo-
sophical disagreements.7 Does it refer to utilitarianism, narrowly conceived?
How shall we define “utility”? Does it refer to pleasures and pains?8 Does it
make distinctions among qualitatively different goods (a beach, a house, a
dog, a friendship)? If we can answer such questions, how do we measure
welfare? If it is a broader concept than utility (as many think), does it cap-
ture everything that ought to matter in human life? If wealthy people lose
more than poor people gain, has welfare been reduced? Is that decisive?
Here again we bracket fundamental questions. We suggest that as a gen-
eral rule, all nudges should have to pass a social welfare test, which means
that they should produce welfare gains on net. They should also (and this is
an independent point) maximize social welfare, which means that of those
approaches that would produce net gains, they should have the highest net
gains.

6 A superb collection, with many behavioral insights, is Addiction and Choice (Nick Heather
and Gabriel Segal, eds., 2017).
7 See Utilitarism and Beyond (Amartya Sen and Bernard Williams et al., 1982).
8 A valuable treatment is Paul Dolan, Happiness by Design (2016).
136 A B ill of R ights for Nudging

To be more concrete: Suppose that in some nation (say, Norway) a default

in favor of green energy would impose significant costs on consumers,
because the cost of green energy is higher than that of (say) coal. Suppose
too that the green energy default would produce significant reductions in
pollution (including greenhouse gas emissions). The questions would be:
Are the costs lower than the benefits? If so, are there ways to produce higher
net benefits? We could easily imagine cases in which these questions would
be easy to answer, because the numbers are so clear. We could also imagine
cases in which these questions would be hard. If they turn out to be hard,
at least we know why they are hard—and we know where we need to seek
more information (or perhaps make some controversial judgments).
We are acutely aware that these points raise immediate questions.
Consider five:

1. Nudges should be effective and cost-effective. That seems important.

Does our analysis include those requirements?
2. Social welfare is one thing; cost–benefit analysis is another. While
cost–benefit analysis is often defended as the most administrable way
to test the question whether a policy promotes social welfare, some
people are not persuaded at all by that argument.9 Monetizing various
welfare effects can be a serious challenge.
3. Some nudges are designed to overcome present bias and inertia, and the
welfare analysis may not be straightforward in such cases. Automatic
enrollment in savings programs is an example. If people lose something
in take-home pay but have more money in retirement, are they better off?
4. Some nudges are designed to serve distributive purposes. For example,
they might be meant to help those at the bottom of the economic lad-
der. Does the welfare analysis capture that goal?
5. Some nudges are designed to prevent discrimination or to reduce vari-
ous forms of unfairness. For example, they might be meant to reduce
discrimination on the basis of race or sex, or to ensure fair treatment in
the workplace. Does the idea of welfare include or exclude that goal?

Some of these questions are easier than others. If a nudge is ineffective, it is

unlikely to be delivering significant benefits—which means that it will be
exceedingly hard to justify on welfare grounds. The advantage of welfare

9 See Matthew Adler, Welfare and Fair Distribution (2011).

 A Bill of Rights for Nudging 137

analysis is that it forces us to ask the right question: How effective, exactly? The
idea of cost-effectiveness is important, and the idea of maximizing social wel-
fare captures it in exactly the right way. If an intervention is not cost-effective,
it is unlikely to maximize welfare. Some other approach would be better.
Cost–benefit analysis is indeed a proxy for welfare, not the thing itself. In
some cases, it might mislead us, or it might prove incomplete—as, for exam-
ple, when we are dealing with variables that are hard to quantify (such as the
effects of disclosure of information).10 When the goal is to combat present
bias or inertia, it remains necessary to ask: Are people being helped or hurt
by the nudge, on balance? We cannot evaluate automatic enrollment in pen-
sion programs, or think about default contribution rates, without asking that
question. We should be aware that despite our best efforts, our answers may
be fallible—which presents a reason to engage the public (see Chapter 6).
When the goal is to help those at the bottom of the economic ladder,
or to combat discrimination and unfairness, it remains important to know
that the intervention is effective and cost-effective. If it would accomplish
little or nothing, and if another approach would do as much for a lower
cost, then we know enough not to undertake it. But it is not unreasonable
to wonder whether the welfare criterion allows us to take account of efforts
to help those who face economic deprivation, or sex discrimination, or
unfair treatment of workers.
There is a large literature on how to understand welfare and welfarism,
with particular attention to exactly these questions. Once more, we do not
mean to take a stand on the philosophical questions.11 We agree that if
welfarism is unable to embrace the relevant goals, so much the worse for
welfarism; it stands condemned as incomplete.
But our goal here is not to give a full account of the occasions for
using nudges or of the proper scope of behaviorally informed tools. We
emphasize more modestly the importance of respecting autonomy and
of promoting social welfare. These ideas fit, broadly speaking, with our
empirical findings. We have not said that the results of surveys should be
taken as decisive. On the contrary, we have emphasized that judgments
about potential policies demand a careful study of their likely effects—and
that public judgments, uninformed as they may be, are no substitute for
that kind of study.

0 For detailed discussion, see Cass R. Sunstein, The Cost–Benefit Revolution (2018).
1
11 See Adler, supra note 130.
138 A B ill of R ights for Nudging

Nonetheless, we confess that we have been singularly impressed with

what we have found—and in important respects, surprised by our findings.
The judgments of ordinary citizens suggest an intuitive understanding, in
diverse nations, of the importance of both autonomy and welfare. They
suggest that in this domain, as in many others, any Bill of Rights will not
be a top-down imposition from a self-appointed political elite. It is more
likely to be an outgrowth of powerful strands in national cultures, and per-
haps even the human heart.
 A c kno w ledgm en t s

This book has been years in the making, and we have many institutions and
individuals to thank.
For financial support, we are grateful to the Behavioral Economics and Public
Policy Program at Harvard Law School; we are also grateful to the Governing
Responsible Business Cluster at Copenhagen Business School and the Center
for Consumption, Markets and Politics at Zeppelin University in Germany. For
valuable discussions at various stages, we thank Eric Posner, Eldar Shafir, and
Richard Thaler. For invaluable research support we thank Micha Kaiser and
Julius Rauber. For editorial assistance and research support, we thank Andrew
Heinrich. Special thanks go to Roger Franz for his valuable suggestions.
We also thank the cooperators and governments of Flanders (Belgium),
Ireland, and Mexico for sharing their national survey data with us.
In producing this book, we have drawn on a series of studies on public
opinion and behaviorally informed approaches. Though we have made signifi-
cant changes and additions, we are grateful for permission from the following
journals. For Chapters 1, 2, and 3, “Do People Like Nudges?,” Administrative Law
Review; for Chapter 4, “Do Europeans Like Nudges?,” Judgment and Decision Making
11, 310 (2016); for Chapter 5, “A Global Consensus on Nudging? Almost
But Not Quite,” Regulation and Governance 12, 3 (2018); for Chapter 6, “Trusting
Nudges? Lessons From An International Survey,” Journal of European Public Policy
140 A cknowledgments

(DOI: 10.1080/13501763.2018.1531912; for Chapter 8, “Misconceptions

About Nudges,” Journal of Behavioral Economics for Policy 2, 61 (2018).
Early discussions of some of our findings are also reported in Cass R.
Sunstein, The Ethics of Influence (2016) and Cass R. Sunstein, Human Agency and
Behavioral Economics (2017); in all cases, we draw on the original sources, not
on the material that appears there.
 Index

Entries in bold denote tables; entries in italics denote figures.

abortion: attitudes to nudges on 9, behavioral biases 96, 120–1, 124

23; educative and noneducative Behavioural Insights Team (UK) 1–2
nudges on 98–100, 108–12, Belgium 71, 73–4, 82, 84
110, 118 benefit-cost analysis 43, 115, 117, 136–7
active choice, mandatory 40 Bill of Rights for Nudging x, 30, 130–4
affirmative consent see explicit BMI (Body Mass Index) 71–2, 76
consent boosts 97
age, and attitude to nudging 65 bounded rationality 125
agency: and educative nudges 95, Brazil: attitudes to nudging in 63; use
97, 99–100, 103–4, 108, 116–18; of nudging in 66n9
nudges as insult to 119–20 BRICS countries 53–4
alcohol consumption 71–2, 74, Buchanan, James 120–1
77, 83
American exceptionalism 29 cafeteria design 22, 99, 108–10, 122
attention filtering 57 calorie labels 2n6: European
Australia, attitudes to nudging in 63 attitudes to 35, 38, 39; partisan
autonomy, and nudging 4–5, 7, 116, differences on 24; targeting of 83;
134–5, 137–8 US attitudes to 9, 10
142 I ndex

Cameron, David 28 Denmark, attitude to nudges in 31, 36,

Canada, attitudes to nudging in 39–40, 42–3, 45–7, 55, 66–7, 82, 86
62, 66 descriptive statistics 88–9
carbon emissions charge 15–16, 35, dignity 4–5, 116–17, 130
39, 40, 66 disclosures 22
cautiously pro-nudge nations 55, 66, discrimination, combating 14, 20,
70, 73, 82 130, 132, 136–7
charitable donations, default rules distracted driving: global attitudes
for 11, 16, 22, 30, 35, 39, 40, 44, 60, to 59; graphic warnings about 114;
66, 133 public education against 10, 23–4,
children: educative and noneducative 35, 37
nudges for 99–100, 108–12; distributive purposes 136
number of 57, 79
China, attitudes to nudging in 54–5, Earned Income Tax Credit 127
60–3, 67–9, 80 economic incentives 115
choice architecture 2, 4; European educative nudges 31; and agency
attitudes to 35, 43; global attitudes 117–19; preference for 98–103,
to 62; mandatory active choice as 102, 105, 106–7, 111, 112–4, 113;
40; motivations behind 19, 28; US regulators’ use of 114–16; and
attitudes to 12–13 thinking systems 96–7; use of
choice editing 35 term 95
cigarette package warnings 9, 24 effectiveness information 98–9, 102,
Citizens Score 68 103–7, 112, 113, 115
clean energy see green energy environmental concern, and nudge
coercion 128–9, 133 approval 72, 75, 83
Communism, warning labels for EPA (Environmental Protection
17–18, 22 Agency) 129–30
consumption habits 72, 74, 83 ethical issues 4–6
correlation heatmap 75, 76 Europe: attitudes to nudges in 34,
credit card legislation 126 35–44, 37–41; items in nudge survey
cultural clusters 54, 71 32–3; national variations in 29–31,
45–6; political parties of 49–50
default rules 2, 7; and agency 120; explicit consent, to takings 22, 28, 30,
European attitudes to 30, 35, 39, 44, 47, 60, 133
40–1, 43; global attitudes to 60, 61;
law as 121; as noneducative nudge food labeling see calorie labels; traffic
95, 97; people disadvantaged by light food labeling
21; transparency about 123; US formal education, and nudge
attitudes to 8, 15, 16 approval 58, 79
 I ndex 1 43

framing, of ethical questions 5 inertia: and default rules 97, 117,

France, attitude to nudges in 124–5; negative consequences of 7,
31, 46–7 22, 28, 44
free school meals programs 126 information nudges 35, 37–8, 39–40,
freedom of choice 119–21, 125, 131, 42, 46, 65
134; perceived 74, 76 Internet penetration rates 54
fuel economy and energy efficiency intrusiveness, levels of see nudge
standards 129–30 clusters
Ireland 82, 84
gender: and attitude to nudges 42, irrationality, nudges as assuming
47, 63, 75–6, 80, 83, 84–5; changing 124–6
18, 21 Italy, attitude to nudges in 31, 36, 40,
Germany, attitude to nudges in 31, 42, 45
82, 86
GFK (Gesellschaft für Japan, attitudes to nudges in 55, 60,
Konsumforschung) 33 62–3, 66–7, 69–70
GMOs (genetically modified
organisms) 9n2, 12–13, 22 Kahneman, Daniel 95–6
green energy: educative and knowledge problem 120–1
noneducative nudges 98, 100–8,
112; encouraged or mandated labor violations, labels for 14, 22
use of 12, 20, 35, 39, 40–1, 60, 79, legitimacy: of goals 9, 23, 30, 43, 47,
114, 127 131; of nudging 7, 82–3, 128–30
green nudges 72 loss aversion 30, 60, 130
Green Parties 46–7
mandates: and nudges 114, 117; US
Hayek, Friedrich 120–1 attitudes to 24–5
health warnings, graphic 9n4, 97, manipulation: nudges as acceptable
101–3, 114–15, 123 123; protection against 121;
healthy food placement 12, 39, 40, unacceptable 22–3, 30, 44, 53, 114,
58, 61 132–3
heuristics 27, 67, 96, 125 markets, belief in 76–7, 83
Hungary, attitudes to nudges in meat-free days 35, 41, 42, 63, 65, 77, 83
31, 36, 39, 42–3, 45, 47, 55, Mexico 82, 84
66–7, 70
name changes, automatic 13–16,
illicit goals 4, 7, 19–20, 53 20, 53
India 54n1 noneducative nudges 31; and agency
individual rights 6, 128, 131–2 117–18; and freedom of choice 120;
144 I ndex

preference for 98–100, 102–6, 33, 36, 46, 51; in global survey 57–8,
111, 112–4, 113; regulators’ use of 64, 65, 73; in US survey 9, 11–12,
114–16; and thinking systems 97; 20, 22–4
use of term 95 political parties, nudges favoring 15,
nudge clusters: in European survey 18–19
35–6, 43; in global survey 59, 64, political valence, of nudges 7, 26–7
78–9, 82–4, 93–4 preference falsification 68–9
nudges: costs of 116; covertness present bias 96, 124–5, 129, 136–7
of 122–3; effectiveness of 126–7; principled pro-nudge nations 55,
institutional use of 2–4; national 65–6, 70, 73, 82
categories of support for 55, 65–9; public choice problems 120–1
objectives of 129; public opinion public education campaigns:
on 4–7; use of term 1–2 European attitudes to 36, 37–8;
global survey on 59–60; US
Obama, Barack 3, 26 attitudes to 8, 10, 11, 17, 18, 20–1
obesity, educative and noneducative
nudges on 98–9, 109, 110–11 Qualtrics 56
obesity education: European quantitative information 103–6, 104,
attitudes to 35, 37; global attitudes 107, 110, 117
to 59; partisan differences on 24;
US attitudes to 10–11, 20, 23–4 randomized controlled trials 122, 134
opioid addiction 121, 123 reactance 98n8, 123
optimism, unrealistic 124, 129 Red Cross see charitable donations,
order effects 14, 95, 112 default rules for
organ donors: default rule for 13n13; relevant information 6, 43, 97, 115
mandatory active choice 12, 21, 35, religion, nudges favoring 15–16, 19
39, 40, 61–2, 66n9 risk aversion 72–3, 76
overeating, public education about Russia: attitudes to nudging in 54–5,
11, 35, 37 62–3, 66; use of nudging in 66n9
overwhelmingly pro-nudge nations
55, 67, 69, 73, 82 salt content labels 12, 13, 22, 35, 38, 39
sampling: for European survey 31, 48;
Park Geun-hye 69 for global surveys 56–7, 70, 73–4,
partisan nudge bias 25–8, 73 87, 90–2
paternalism 30, 46 savings plans: automatic enrolment
political attitudes: educative and in 9–10, 20, 24–5, 101, 114,
noneducative nudges and 106–8, 122, 127, 136; educative and
110–2, 111, 117; in European survey noneducative nudges 98, 100–8
 I ndex 1 45

self-government 4, 6–7 traffic light food labeling 12, 24, 32,

sexual orientation discrimination 38, 39, 66n9
10–11 transparency 122–3, 134
‘significantly more effective’ see trust: in government 47, 67, 70, 73–5,
effectiveness information 120–2; institutional 71–2, 74–6, 77,
smoking: educative and 79, 80–1, 83; in other people 76;
noneducative nudges 98, 100–8; social 71–2, 74, 76
and nudge approval 71, 74, 77, 83;
public education about 11, 35, 37 United Kingdom: attitudes to nudges
social welfare 5–6, 84, 115, 133, 135–7 in 31, 36, 40, 42, 45; partisan
sociodemographic variables: in differences on nudging 46; use of
Europe 35–6, 42–3, 47, 51–2; in nudges in 2, 28, 30
global survey 57–8, 63, 64 United States: attitudes to nudging
South Africa, attitudes to nudging in 8–19, 10–13, 17, 82; preference
in 63 for educative and noneducative
South Korea, attitude to nudges in nudges in 98–112; use of nudging
55, 60–2, 67–70, 82–3, 86 in 2–3
statistical literacy 97, 125 utilitarianism 133, 135
subliminal advertising: European utility functions, complexity of 125
attitudes to 30, 35, 40, 42, 44, 47;
global attitudes to 62, 63, 114, 123, values and interests, consistency
132, 134; US attitudes to 9, 17, 114 with 8, 22, 114, 132
sugar taxes 129 visual illusions 9, 22, 114
sweet-free cashier zones 31, 35, 41, voter registration: automatic
42, 45, 63, 65, 134 13–14, 20; educative and
System 1 nudges see noneducative noneducative nudges 98–9,
nudges 108–12, 109, 111
System 2 nudges see educative
nudges warnings 2; US attitudes to 8
Systems 1 and 2 thinking see thinking water conservation, educative and
systems noneducative nudges 98, 100–8,
117–18
takings 21 welfare analysis, of nudging 115–16,
Tannenbaum, David 25–7, 45n8 135–8
terrorism, warning labels for 14, 22 women’s surnames see name
test-learn-adapt-share approach 84 changes, automatic
thinking systems 95–7; and nudge World Bank, use of nudging 3
preference 114 WVS (World Values Survey) 72, 74