Polit Behav (2015) 37:221–251

DOI 10.1007/s11109-014-9269-1

ORIGINAL PAPER

Explaining Group Influence: The Role of Identity

and Emotion in Political Conformity and Polarization

Elizabeth Suhay

Published online: 8 March 2014

 Springer Science+Business Media New York 2014

Abstract Evidence has accumulated that people often conform to political norms.
However, we know little about the mechanisms underlying political conformity.
Whose norms are people likely to follow, and why? This article discusses two
phenomena—social identity and ‘‘self-conscious’’ emotions—that are key to
understanding when and why people follow the crowd. It argues that adherence to
in-group norms is a critical basis of status among in-group peers. Conformity
generates peer approval and leads to personal pride. Deviance generates disapproval
and causes embarrassment or shame. These emotional reactions color an individ-
ual’s political perspectives, typically generating conformity. These same mecha-
nisms can spur between-group polarization. In this case, differentiation from the
norms of disliked out-groups results in peer approval and pride, and conformity to
out-group norms disapproval and embarrassment or shame. This framework is
supported by the results of two experiments that examine the influence of group
opinion norms over economic and social aspects of citizens’ political ideologies.
One exogenously varies the social identity of attitudinal majorities; the other primes
the relevant emotions. In addition to contributing to the study of political conformity
and polarization, this article adds to our growing understanding of the relevance of
social identity and emotion to political life.

Keywords Conformity
Polarization
Social identity
Emotion
Norms

Political opinion

As of July 1, 2014, the author will be Assistant Professor of Government at American University.

E. Suhay (&)
Department of Government and Law, Lafayette College, 100 Kirby Hall, Easton, PA 18042, USA
e-mail: [email protected]

123
222 Polit Behav (2015) 37:221–251

After a long hiatus, political scientists have returned in force to the study of social
influence. Researchers in this area share the perspective that individuals’ political
decisions are not made in a social vacuum but, rather, are continually influenced by
the words and actions of others. Most political scientists who study social influence
focus on one of two topics: political discussion within social networks (e.g.,
Huckfeldt and Sprague 1995; McClurg 2006; Sinclair 2012; Sokhey and Djupe
2011) or conformity to social norms (e.g., Bolsen 2013; Gerber et al. 2008; Mutz
1998; Nadeau et al. 1993; Nickerson 2008; Noelle-Neuman 1993; Walsh 2003). The
accumulated findings from studies in both areas leave little doubt that people
influence one another’s political attitudes and behaviors.
While scholars have successfully demonstrated social influence in a variety of
settings and with respect to many different political outcomes, considerably less
attention has been paid to the mechanisms of such influence, i.e., how and why
political decisions are affected by a person’s social network or group. Drawing from
rational choice theories (e.g., Downs 1957), most social network scholars assume
that influence between discussants occurs due to information exchange or logical
argument (e.g., Huckfeldt and Sprague 1995). While this familiar assumption may
be appropriate for the study of political discussion, it does not explain conformity to
social norms. In fact, the conformity, or ‘‘social pressure,’’ phenomenon is in part
defined by the absence (or irrelevance) of substantive information exchange or
argument.
What fills this mechanistic void? Research in psychology and sociology suggests
that identification with peers (Turner et al. 1987, 1991) and ‘‘self-conscious’’
emotions—especially pride, embarrassment, and shame (Lewis 2000; Scheff
1988)—are worth consideration. While some studies of social influence in political
science have discussed these mechanisms and tested them indirectly (e.g., Gerber
et al. 2010; Sinclair 2012), surprisingly few studies have provided persuasive
evidence of their causal role in influence by directly manipulating them
experimentally and/or measuring them via participant self-report. The result is that
speculation as to the role of identity and emotion in politically relevant social
influence far outweighs the evidence. In addition, while some political scientists
have discussed these mechanisms independently, none have combined them into an
integrated theoretical framework.
The framework presented here focuses on the mechanisms underlying conformity
to social norms1 and argues that self-conscious emotions encourage individuals to
adopt the norms of groups with which they closely identify.2 More specifically,
pride stemming from perceived approval of in-group peers signals to the individual
that a conforming viewpoint or action is valid, whereas embarrassment or shame
stemming from perceived disapproval of in-group peers generates doubt. The result

1
While political scientists often associate norms with behavior, group norms may be behaviors, thoughts,
or even feelings that are typical within a group (Hogg and Reid 2006, p. 8).
2
Many types of groups are relevant, ranging from face-to-face (e.g., workplace, neighborhood) to
broader demographic (e.g., race, sex, religious) groups. A person who identifies with a group perceives it
to be an important element of his or her personal self-concept (Tajfel and Turner 1986).

123
Polit Behav (2015) 37:221–251 223

is conformity to in-group norms. These same mechanisms contribute to between-

group polarization as well. Individuals tend to distance themselves from out-group
norms in part because they perceive that in-group peers approve of opinions and
actions that are different from those of disliked out-groups and disapprove of those
that resemble disliked out-groups. This framework is tested with two experiments
that examine the role of identity and emotion in the influence of group opinion
norms over economic and social aspects of citizens’ political ideologies. Overall,
the evidence supports the theoretical framework.
This article contributes to research in political science beyond the study of
political conformity and polarization. First, this study broadens our understanding of
the political relevance of emotions. While considerable evidence on this score has
accumulated, the role of emotions in social influence has remained underexplored,
as has the political relevance of the particular emotions examined in this study—
pride, embarrassment, and shame. Second, this article unpacks the somewhat
mysterious phenomenon of shared group norms. It is common wisdom in political
science that political opinions and actions are correlated with social identity; the
social–psychological dynamics described in this article provide one explanation for
why this is the case.

Theoretical Framework

Social Influence Over Political Opinion and Behavior

Over a half century ago, the Columbia and Michigan schools of voting behavior
were quite taken with social influence and, in particular, the idea of social
conformity. Berelson et al. (1954, p. 122) argued that ‘‘[d]uring a campaign political
preferences are ‘contagious’ over the range of personal contacts.’’ Campbell et al.
(1960) showed that voting preferences of citizens mirrored those of their racial,
religious, and union affiliations, even after controlling for characteristics that
citizens were likely to share with fellow group members (e.g., income, education,
occupation). The authors attributed these findings to forces less ‘‘rational’’ than
ordinary persuasion. Berelson et al. argued that ‘‘[f]or many voters political
preferences may better be considered analogous to cultural tastes,’’ with origins in
ethnic, sectional, class, and family traditions (1954, p. 311). Campbell et al. wrote:
‘‘[N]orms and values attributed to a generalized ‘group’: these are the expectations
concerning appropriate behavior for [e.g.] the ‘loyal’ Catholic or union member. It
is the group standards that are psychologically real and are responsible for influence
when it occurs’’ (1960, p. 296).
These voting scholars had been influenced by the findings of experimental
psychologists studying group dynamics who, in the preceding decades, had been
working to understand a particular social influence phenomenon, variously labeled
‘‘group pressure,’’ ‘‘conformity,’’ or ‘‘majority influence.’’ For example, Sherif
([1936] 1966) demonstrated with his ‘‘auto-kinetic effect’’ studies that a group of

123
224 Polit Behav (2015) 37:221–251

previously unacquainted individuals would converge on a group norm that would

influence the individuals’ judgments even when they were separated from the group.
Asch (1951), in his line-length studies, discovered that norms created by the
experimenter, via confederates, were sometimes influential over participants’
judgments even when they were obviously ‘‘wrong.’’ Hundreds of similar
experimental studies followed, many explicitly addressing the power of social
and political norms (see Turner 1991 for a review). Together, the studies suggest
that individuals frequently change their beliefs or behaviors to mirror a perceived
majority3 even when no accompanying explanations, information, or arguments are
given.
However, the reasons for conformity remained murky, even among psycholo-
gists. The lack of a clear mechanism of influence cast doubt on the conformity
phenomenon. Coupled with an overall decline in interest in social influence,
political scientists seemed to quickly lose interest in the topic of conformity. When
the study of politically relevant social influence reemerged several decades later,
interest had shifted to discussion networks (e.g., Beck 2002; Huckfeldt and Sprague
1987, 1995; McClurg 2006). One benefit of this influence paradigm is that it fits
more comfortably within standard rationality assumptions. Scholars in this tradition
tend to argue that citizens exchange—and are persuaded by—information and
logical arguments.
Until very recently, only a handful of contemporary studies had tackled the
subject of political conformity. The most prominent are Mutz’s Impersonal
Influence (1998) and Noelle-Neuman’s Spiral of Silence (1993). Although each
builds on the conformity research tradition in psychology, these works differ from
that tradition in important ways. For example, similar to social network scholars,
Mutz fits conformity into a rationalistic framework; she argues that some people
construct arguments as to why the majority believes as it does, persuading
themselves in the process. Mutz’s framework represents an alternative, although
ultimately complementary, explanation for political conformity to the one offered
here. Noelle-Neuman moves in a different direction. She explains her ‘‘spiral of
silence’’ findings by arguing that people are motivated to avoid embarrassment;
however, she never directly tests this assumption.
Interest in conformity has noticeably increased more recently in political science,
with scholars quickly amassing considerable evidence that peers often influence one
another’s political opinions and behaviors even when substantive information
exchange and arguments are absent (e.g., Bolsen 2013; Gerber et al. 2008; Gerber
and Rogers 2009). While this more recent scholarship has focused more on
demonstrating the influence of social norms than on investigating the mechanisms

3
This article focuses on descriptive norms (an opinion held or behavior engaged in by the majority), not
injunctive norms (opinions or behaviors considered socially desirable or even morally correct). Note,
however, that the line between descriptive and injunctive norms is unclear. Like injunctive norms,
descriptive norms typically suggest to people how they ought to behave (Theiss-Morse 2009), signaling
opinions and behaviors ‘‘appropriate’’ for group members (Turner et al. 1987). Also similar to injunctive
norms, descriptive norms can be enforced via social-psychological rewards and sanctions (Scheff 1988).

123
Polit Behav (2015) 37:221–251 225

that underlie it, scholars have begun a search for mechanisms of influence in line
with those proposed in this article. Reflecting the thinking of early voting scholars,
Sinclair (2012) argues that social identity plays an important role in political
influence (also see Alwin et al. 1991; Walsh 2003). Supporting this idea, she
demonstrates that neighborhood canvassers are more successful at getting people to
vote than outsiders. One limitation of this study, however, is that social identity is
manipulated only roughly (assuming neighbors identify with one another). Similar
to Noelle-Neuman (1993),4 Gerber et al. (2008, 2010) and Panagopoulos (2010)
argue that the desire to feel pride and avoid embarrassment or shame motivates
citizens to conform to voting norms in their communities. They demonstrate that
people are more likely to vote when reminded voting records are public. While the
evidence that this type of experimental treatment increases voter turnout is highly
persuasive, the emotions assumed to cause this relationship are not directly
manipulated or measured, leaving the precise emotional mechanisms uncertain. In
sum, while strong evidence for political conformity has accumulated and scholars
have named social identity and particular emotions as mechanisms of conformity,
research demonstrating a causal role for these mechanisms via exogenous
manipulation and/or careful measurement is all too rare.

Explaining Conformity and Polarization with Social Identity and Emotion

The primary goal of this article is to better understand the mechanisms underlying
political conformity, with a secondary goal of relating these same mechanisms to
polarization.5 Psychologists and sociologists have paid considerable attention to
these phenomena, incorporating the themes of identity and emotion into a variety of
prominent theoretical frameworks relevant to the study of political conformity and
polarization.

The Social Identity Perspective

This article draws important insights into conformity and polarization from ‘‘self-
categorization theory’’ (SCT) (Turner 1985, 1991; Turner et al. 1987). SCT builds
upon—and expands the scope of—social identity theory, developed by Henri Tajfel
in conjunction with Turner (e.g., Tajfel 1978; Tajfel and Turner 1986).6 While most
political scientists are more familiar with social identity theory, this article focuses
on SCT for two reasons. First, unlike social identity theory, SCT develops a well-
4
See also work by Elster on adherence to social norms (e.g., 1999).
5
Many definitions of polarization exist. In this article, polarization refers to the phenomenon whereby
the norms of two or more groups increasingly differ from one another over time.
6
The borders of social identity theory are ambiguous, in large part because it has spawned an enormous
literature. A common error is to attribute theoretical propositions and empirical insights associated with
self-categorization theory to social identity theory, particularly in the arena of social influence (see
Haslam et al. 2010). Such confusion is understandable given the overlapping themes (social identity) and
authorship (Tajfel and Turner). And, in some cases, authors explicitly use the ‘‘social identity theory’’
label to refer to both theories, while clarifying that there exist two separate branches of the theory (e.g.,
Huddy 2001; Huddy and Khatib 2007). In this article, the term ‘‘social identity perspective’’ is used as an
umbrella term to refer to both theories together along with newer theories that build on their insights.

123
226 Polit Behav (2015) 37:221–251

specified theory of the psychological group, one that allows for investigation of
intragroup, as well as intergroup, dynamics (Haslam et al. 2010). Second, SCT has
been used to explain a wider range of group-related phenomena than has social
identity theory. Most notably, SCT offers arguably the most prominent explanation
for social conformity and polarization in contemporary social psychology (Haslam
et al. 2010).
Self-categorization theory argues that the key to in-group conformity and
between-group polarization is social identity, i.e., a person’s incorporation of a
group membership into his or her self-concept (Turner 1985; Turner et al. 1987).
When a relevant social identity is salient, individuals will seek to be ‘‘prototypical’’
group members (Turner 1985; Turner et al. 1987). Prototypical group members are
characterized by a high ‘‘meta-contrast ratio,’’ meaning their attitudes and actions
not only resemble in-group members’ but are also easily distinguished from those of
out-group members. Numerous experimental studies support this notion (Turner
1991). People tend to conform to the attitudinal and behavioral norms of in-groups,
not out-groups (Abrams et al. 1990; Clark and Maass 1988; Wood et al. 1996).
Further, taking into account the fact that identity strength varies (Huddy 2001),
group members who identify more strongly with the group than other members are
more influenced by its norms on average (Terry and Hogg 1996; Terry et al. 1999;
also see Theiss-Morse 2009). Finally, when people learn of a norm held by an out-
group, particularly a derogated one, they tend to shift their attitudes or actions in the
opposite direction (Clark and Maass 1988; Nelson et al. 2011; Wood et al. 1996).
These researchers, by employing randomized and controlled experiments, have
isolated social identification as one cause of conformity (and lack of identification
as a cause of polarization). Thus, the effects are not confounded with competing
causal influences on within-group similarity, such as homophily (i.e., like tending to
affiliate with like, see Lazer 2011).7
What is it about identification that opens a person up to group influence?
Psychologists have long argued that peers provide the individual with helpful
‘‘social reality tests’’ when peers indicate whether they agree with the individual’s
point of view (Festinger 1950; Turner et al. 1987, 1991). The people with whom one
identifies provide the best social reality tests: ‘‘The perceived, expected or believed
agreement of similar others in the same situation implies that our behaviour is a
function of the objective world’’ (Turner 1991, p. 161). Agreeing with in-group
members generates ‘‘subjective validity,’’ the perception of truth; disagreeing with
them generates ‘‘subjective invalidity,’’ the perception of falsehood. Note, however,
that the logic is reversed for out-groups: agreeing with out-group members should
generate subjective invalidity.

7
These studies do not argue that homophily does not contribute to within-group similarity; rather, they
argue that group influence over the individual and homophily both are reasons for within-group similarity
but also that their effects can be disentangled through experimentation.

123
Polit Behav (2015) 37:221–251 227

Self-categorization theory offers an appealing explanation for political confor-

mity and polarization; however, it seems incomplete. Namely, it largely ignores
emotion.8 Empirical accounts of people’s encounters with group norms are thick
with emotion, especially when people disobey norms (e.g., Asch 1951; Elster 1999;
Milgram 1992; Noelle-Neuman 1993). And, while SCT—at least as originally
formulated by Turner—says little about emotion, its forebear social identity theory
considers emotion to be highly relevant to social identity and inter-group processes.
In particular, a key tenant of social identity theory is that ‘‘individuals seek to
achieve or maintain positive self-esteem by positively differentiating their ingroup
from a comparison outgroup on some value dimension’’ (Haslam et al. 2010, p. 343;
see, e.g., Tajfel and Turner 1986).9 Is emotion—specifically, the desire for positive
self-esteem—relevant to social influence?

Self-Conscious Emotions

A group of emotions called ‘‘self-conscious emotions’’—especially embarrassment,

shame, and pride—deserve our attention in seeking to understand political
conformity and polarization. Not only are they closely linked to individuals’ levels
of self-esteem, but they also have been explicitly named by many researchers as
responsible for social conformity.
Although he does not explicitly discuss emotion, Turner argues that those who
achieve a high meta-contrast ratio (again, not only resembling the in-group, but also
being distinct from the out-group) are evaluated most highly by in-group peers
(Turner 1985; Turner et al. 1987; also see Theiss-Morse 2009). According to
other researchers, those who exemplify shared community ideals tend to be treated
with respect and admired (Lewis 2000), whereas norm breakers are often subject to
derogation (or shaming) (Marques et al. 2001). While the former case is likely to
lead to positive self-esteem and pride, the latter tends to cause embarrassment and
shame10 (Asch 1951; Lazarus 1991; Lewis 2000; Milgram 1992). The result is often
conformity at the individual-level to social norms and values (Lewis 2000; Scheff
1988). This phenomenon has been described most fully by sociologist Thomas
Scheff: ‘‘[P]ride and shame make up a subtle and pervasive system of social
sanctions. This system leads to experiencing social influence as compelling’’ (Scheff
1988, p. 396). The system functions continuously, even when we are alone, because

8
Defining emotion is notoriously difficult. Lazarus (1991) says that ‘‘emotion is an integrative,
organismic concept that…unites motivation, cognition, and adaptation in a complex configuration’’ (40).
Cognitive appraisals of whether and how a situation is relevant to an individual’s goals set in motion
(ideally) adaptive action tendencies and coping mechanisms. Much of this psychological and
physiological activity occurs automatically and subconsciously, but some may be conscious, including
subjectively felt ‘‘feelings.’’
9
Self-categorization theory also includes this idea, but Turner casts it in cognitive terms (Turner 1985,
p. 261) and does not incorporate it into his explanation for conformity and polarization. Turner’s de-
emphasis of the self-esteem plank of social identity theory may stem from uneven empirical support for
this proposition (see Brown 2000). The self-esteem hypothesis discussed in this article is related but
clearly distinct from that discussed as a part of social identity theory.
10
While Lazarus argues that the main difference between these two emotions is intensity level (Lazarus
1991), note that some argue for differences in kind as well as degree (e.g., see Miller 2007).

123
228 Polit Behav (2015) 37:221–251

we can imagine others’ reactions to our behavior in detail (Lazarus 1991, p. 241;
Scheff 1988).
The explanation for conformity offered by the emotions literature is likewise
compelling but also underspecified in its lack of attention to identity. Certainly not
every majority norm of which a person is aware will influence him or her.
Americans are unlikely to feel embarrassed by their tendency to wear shoes indoors,
even if made aware that many other cultures disdain the practice. A longtime
Democrat is unlikely to gloat when she reads a public opinion poll and discovers
that her opposition to gay marriage places her within the Republican mainstream.
While the literature on self-conscious emotions does specify that emotion-driven
conformity occurs in response to ‘‘peers,’’ ‘‘important others,’’ and the like
(suggesting social identity is relevant), this component is undertheorized and
generally untested empirically.

An Integrated Perspective on Social Influence

It seems clear that an integration of SCT and the literature on self-conscious

emotions would benefit the study of social influence generally, and political
influence more specifically. While two existing theories have made moves in this
direction—intergroup emotions theory (Mackie et al. 2009) and collective self-
esteem (Luhtanen and Crocker 1992)—both offer more insight into intergroup
dynamics (e.g., prejudice) than the intragroup dynamics that are responsible for
conformity.11 This should not be surprising. As Theiss-Morse (2009, p. 66) argues,
the extent to which group boundaries affect internal group dynamics has generally
been understudied.
Thus, this article draws directly from the self-categorization and self-conscious
emotions literatures. Focusing on their applicability to political variables, a model
of conformity and polarization is introduced. In group contexts, interpersonal
comparative judgments are common among in-group members (Turner 1985,
p. 255). The basis of much of this comparison is the extent to which individuals
conform to in-group norms and differ from out-group norms, particularly norms
held by derogated out-groups. In other words, the extent to which individuals
resemble in-group prototypes is a key basis of status within a group. Prototypicality
(in general or in a specific domain) tends to generate pride in individuals because
they perceive respect and approval from in-group members; non-prototypicality

11
Intergroup emotions theory, a broadly applicable theory developed by Smith and Mackie (see, e.g.,
Mackie et al. 2009), argues that, when social identity is salient, people will appraise situations and
experience relevant emotions in accord not with their personal identities but, rather, with their social
identities. For example, when a social identity is highly salient, an out-group attack on the in-group is
experienced as an attack on the self, and fear or anger directed at the out-group is generated as a result.
Luhtanen and Crocker (1992) provide an updated take on social identity theory’s ‘‘need for positive
distinctiveness’’ with their construct ‘‘collective self-esteem,’’ i.e., that part of an individual’s self-esteem
that is derived from the status of one’s in-group(s) within society at large. This concept is distinct from
what they call ‘‘membership esteem,’’ that part of self-esteem stemming from one’s status as an individual
within the group. While collective self-esteem may be the cause of many important intergroup
phenomena (including prejudice and people’s desire to ‘‘exit’’ low status groups), emotions scholars have
made clear that within-group conformity hinges on membership esteem.

123
Polit Behav (2015) 37:221–251 229

tends to generate embarrassment or shame because individuals perceive a lack of

respect and disapproval. Note that peer approval and disapproval may be
communicated explicitly or may be assumed or imagined on the part of the
individual.
These self-conscious emotional reactions color an individual’s perspective. When
people perceive in-group approval in response to their adherence to in-group norms
(or their differentiation from out-group norms), the resulting pride strengthens their
internal commitment to in-group norms. When people perceive in-group
disapproval in response to their deviance from in-group norms (or adherence to
out-group norms), the resulting embarrassment or shame weakens their commitment
to those deviant attitudes or behaviors. In this way, norms shared by a majority of
in-group peers often become internalized over time by group identifiers.12 Finally,
note that these conforming and polarizing influences are limited to the in-group.
Evaluative feedback from individuals outside the boundaries of the in-group is
largely orthogonal to a person’s self-worth and, thus, his or her self-conscious
emotional experiences.
Given the greater relevance of in-group status to self-worth among high
identifiers, these mechanisms should be more pronounced among high identifiers,
leading to greater conformity and polarization among them. Although for the most
part outside the empirical scope of this article, note that the likelihood and extremity
of conformity and polarization will also vary according to additional factors,
including personality traits, group cultures, and social contexts that may increase or
decrease the salience of particular group prototypes, the tendency to make
interpersonal comparative judgments, or the extremity of emotional reactions to
perceived judgments.
In sum, the theory—which I refer to as ‘‘social–emotional influence theory’’—
provides one explanation for why political conformity occurs. It also argues that
these same mechanisms that underlie conformity to in-group norms underlie
between-group polarization. In other words, it is argued that polarization—while
admittedly dependent on the existence of two groups—is at least in part motivated
at the individual level by intragroup social–psychological processes. While the
theory is broadly relevant to political conformity and polarization, this article
narrows the subject of empirical inquiry in two respects. First, it investigates
‘‘impersonal’’ conformity, as opposed to conformity to individuals with whom one
is interacting in person. Impersonal conformity represents a more difficult test of the
theoretical framework, as in-group peers are distant from study participants. Second,
it focuses on conformity to opinion, not behavioral, norms. This allows for the
elimination of certain confounding factors, such as a calculated concern for one’s
reputation due to deviance that is easily observed by peers. In addition, a focus on
group influence over private opinion rather than public behavior (or opinion

12
Many distinguish between ‘‘informational influence’’ (‘‘true’’ influence based on information) and
‘‘normative influence’’ (surface compliance due to an effort to ingratiate oneself with peers) (Deutsch and
Gerard [1955] 1965) but others have challenged this dichotomy (e.g., Turner et al. 1987; Hogg and Reid
2006). This article suggests that what many would call ‘‘normative influence’’ can in fact cause the ‘‘true’’
influence often associated with informational influence.

123
230 Polit Behav (2015) 37:221–251

expressed in public) may carry more normative weight as it implies peer influence
has been internalized.

Hypotheses

This theory results in six key hypotheses regarding political influence that are tested
below:
In-group conformity (H1) When encountering a new or unexpected in-group
norm, individuals will shift their political views in the direction of conformity with
the majority.
Out-group polarization (H2) When encountering an out-group norm, individuals
will not change their views in the direction of conformity with the (out-group)
majority; and, if the out-group is disliked by their in-group peers, individuals will
shift their views away from those of out-group members.
Identity strength and conformity (H3) Due to the accumulation of identity-based
influence over time, in-group identification will be positively correlated with
adherence to in-group opinion norms.
Identity strength moderation (H4) The influence patterns described in H1 and H2
will be moderated by the strength of individuals’ in-group identification.
Emotional reactions to group norms (H5) Agreement with the in-group majority
generates pride among individuals, and disagreement embarrassment/shame;
agreement with a derogated out-group majority generates embarrassment/shame
among individuals, and disagreement pride.
Emotion moderation (H6) When encountering group norms, individuals who
experience elevated levels of pride or embarrassment/shame are more likely to
change their political views (in the direction of the in-group or away from the out-
group, as relevant).13

Experimental Evidence

Given that experiments are the gold standard for assessing causation (Morton and
Williams 2010), two experiments were designed to test these mainly causal
hypotheses. The first experiment focuses on religious group influence over citizens’
stands on social issues and emphasizes the role of social identity. Citizens’ opinions
on issues such as abortion and gay marriage depend to a significant extent on their
religious faith (Leege et al. 2002). To the extent that religious affiliation represents a
social identity (Djupe and Gilbert 2008; Wald et al. 1988), individuals’ opinions on
these issues are likely to be affected by the opinions held by religious groups—both

13
It is common for causal variables to be both mediators and moderators in a theoretical framework. This
is the case with respect to self-conscious emotions in the model proposed herewith (see the ‘‘An
Integrated Perspective on Social Influence’’ section). Because the critical test of emotions’ causal
influence in Study 2 below involves exogenous arousal, a moderation test is performed. However, one
could argue that—taken together—the two studies test mediation: Group norms arouse self-conscious
emotions in Study 1, and their causal effects are demonstrated in Study 2 (see Bullock et al. 2010; Imai
et al. 2012).

123
Polit Behav (2015) 37:221–251 231

in-groups and disliked out-groups.14 The second experiment focuses on in-group

conformity with respect to social welfare attitudes and investigates the moderating
role of self-conscious emotions. In this study, the social welfare attitudes examined
evoke the value economic individualism, which occupies a central place in
American political culture (Feldman 1999). The relevant group in the experiment is
college peers, a highly relevant social identity for many undergraduate students
(Alwin et al. 1991; Moskalenko et al. 2006).
Taken together, these two experiments offer several advantages. First, each key
causal agent suggested by the theoretical framework (identity and emotion) is
exogenously varied in one experiment, distinguishing causation from correlation.
Second, group influence is examined over two factors—broad preferences with respect
to economic and social policy, i.e., economic and social values—thought to explain most
of the variance in political ideology (Feldman and Johnston 2014). Third, these groups
are not formal political groups. As is likely true of a great many social groups, despite the
likelihood that these groups influence their members’ political views, probably few
people choose to affiliate with them initially for political reasons (see Huckfeldt and
Sprague 1995). This represents a normative problem worth investigating further.

Study 1: Religious Identity and Social Conservatism Experiment

Study Sample

The study was conducted with 220 church-going Catholics from five churches in a
large Archdiocese in the Midwest. Via a combination of flyers placed in church
bulletins and announcements from the pulpit, participants were asked to take part in
a short, anonymous, on-line study of Catholic opinion being conducted by
University of Michigan researchers in return for a $5 donation to their church and a
$5 donation to a Catholic charity of their choice.
Most participants had a high level of commitment to Catholicism. In response to
the pre-test question, ‘‘How important is being Catholic to you?’’ 34 % said ‘‘very’’
and 44 % said ‘‘extremely.’’ Most participants also indicated that they held socially
conservative values, with 71 % indicating in the pre-test that they agreed with
traditional church teachings on abortion, divorce, and homosexuality. These and
other sample characteristics, such as an older mean age (54) and more females
(69 %), likely reflect the church-going nature of the sample.15

Study Design

The experiment had a 1 9 4 between-subjects design. Participants began the on-line

study by filling out a pre-test that addressed Catholic identity, basic political views,
and demographics. Participants were then randomly assigned to one of three
14
While a religious person’s opinions on social issues are also influenced by religious doctrine, religious
leaders, etc., the focus here will be specifically on religious peer influence.
15
Reflecting the demographic attributes of the participating Archdiocese, the sample is more racially
homogeneous (98 % white) and upper-income (50 % middle-class; 40 % upper-middle class) than U.S.
Catholics as a whole. (National statistics from Pew Research Center 2008.)

123
232 Polit Behav (2015) 37:221–251

treatments or the control group. The identity of the influencing group was
exogenously varied by exposing some participants to the views of the Catholic in-
group and others to the views of a disliked out-group, Evangelicals.16
More specifically, those in the ‘‘Catholics are conservative’’ condition learned that
a majority of American Catholics shun divorce, oppose abortion, and oppose gay
marriage. Those in the ‘‘Catholics are progressive’’ condition learned that, to the
contrary, most Catholics believe one can be a good Catholic without following
conservative Church teachings on divorce, abortion, and gay marriage. Finally, those
in the ‘‘Evangelicals are conservative’’ condition learned that most Evangelicals have
socially conservative views on divorce, abortion, and gay marriage. The control group
received no such information. Information presented to participants originated in
reputable public opinion surveys (D’Antonio 2007; Davis et al. 2008; Pew Research
Center 2008). The Catholics are conservative vs. progressive contrast was achieved by
presenting responses to different survey questions. See Appendix 1 for the stimuli. No
‘‘Evangelicals are progressive’’ stimulus was included because survey information
that could be interpreted in this fashion could not be found.
Following the stimulus, participants filled out a post-test questionnaire that
included questions on social and political attitudes as well as their emotions. They
then were debriefed.

Measures

The dependent measure is a socially conservative attitude scale made up of nine

questions touching on five (equally weighted) salient themes—divorce, pre-marital
sex, homosexuality, sexuality and teens, and abortion (a = .83). Variation in
participants’ strength of identification with other Catholics was measured in the pre-
test with an additive scale made up of two questions adapted from Huddy and
Khatib (2007) (r = .71).17 Emotional responses to the stimuli were assessed at the
end of the study with two questions that asked treated participants whether they felt
pride or shame in response to the opinion information (i.e., the stimulus). All
variables range from 0 to 1, with 1 representing the most conservative, identified, or
emotional response. See Appendix 1 for exact wording and response options.

Data Analyses

Expectations with respect to the main treatment effects are as follows. The general
conformity hypothesis (H1) is tested with the ‘‘Catholics are progressive’’
16
American Evangelicals and Catholics tend to be mutually exclusive groups (Putnam and Campbell
2010). Only one participant, removed from the sample, identified as Evangelical. In the post-test,
participants rated Evangelicals 25 points lower than Catholics on average on a 100-point scale. (Ratings
were not influenced by experimental treatments.)
17
The questions that make up the identity measure reflect the social identity perspective’s definition of
‘‘identification’’: ‘‘the extent to which the category is valued and contributes to an enduring sense of self’’
(Haslam et al. 2010, p. 349). By design, the measure is somewhat out of sync with self-categorization
theory’s exclusively cognitive focus (the ‘‘importance’’ question has an affective component). However,
the measure admittedly does not emphasize affiliative, emotional attachments to the same extent as those
of some authors, such as Theiss-Morse (2009).

123
Polit Behav (2015) 37:221–251 233

treatment; participants in this condition are expected to express more progressive

values in the post-test than the control group. The polarization hypothesis (H2) is
tested by comparing reactions to the ‘‘Catholics are conservative’’ and ‘‘Evangel-
icals are conservative’’ treatments. In the former, participants’ values should be at
least as conservative as the control group.18 However, in the latter, values should be
more progressive than the control group, given Evangelicals’ low status among
participants.
Figure 1 displays the means and 90 % confidence intervals19 for the four
experimental groups with respect to social conservatism. The patterns are as
predicted. Relative to the control group, participants who learned that most Catholics
were unsupportive of conservative Church stands were more progressive in the post-
test. And, while participants who learned that other Catholics were socially
conservative were equivalent to the control group, those who learned that Evangelicals
were socially conservative were more progressive than the control group.
We can assess significance levels using OLS regression. The Social Conserva-
tism Scale was regressed onto dummy variables representing the three treatment
groups; the control group was excluded as the comparison group. The model is as
follows: Social Conservatism = b0 ? b1Catholics Conservative ? b2Catholics
Progressive ? b3Evangelicals Conservative ? e.20 The coefficients for the ‘‘Cath-
olics are progressive’’ (b2 = -.077) and ‘‘Evangelicals are conservative’’ (b3 =
-.070) conditions are both statistically significant (p B .05 one-tailed).21 See the
first column of Table 1.
We turn next to evaluating the hypotheses related to identity strength. With
respect to H3, we would expect there to be a positive relationship between Identity
Strength and Social Conservatism given the longstanding association between the
American Catholic community and opposition to abortion, gay marriage, and the
like. To avoid contamination by the experimental treatments, this hypothesis is
evaluated by analyzing the control group only. With respect to H4, we expect strong
identifiers to be more influenced by the ‘‘Catholics are progressive’’ and
‘‘Evangelicals are conservative’’ stimuli than weak identifiers. To test these
hypotheses, Identity Strength is added to the above equation along with interaction
terms between Identity Strength and each of the three treatment groups. The results
are in the last column of Table 1.
The coefficient on Identity Strength, which represents the relationship between
this variable and the dependent variable in the excluded control group, is very large
and highly significant (b = .611, p B .001). On average, the strongest Catholic
identifiers were 61 percentage points higher on the Social Conservatism Scale than
18
Because the information does not challenge de facto assumptions, it is unlikely to influence
participants.
19
Note that confidence intervals surrounding two estimates that are significantly different from one
another may still overlap somewhat. (Confidence intervals are wider than standard errors.)
20
Statistical tests assessing experimental group balance on demographic and political variables showed
that randomization was successful and, therefore, no control variables were used.
21
One-tailed tests are employed throughout this section given the directional nature of the hypotheses.
Note p-values for the following additional contrasts: Catholics conservative vs. Catholics progressive
(p = .07); Catholics conservative vs. Evangelicals conservative (p = .08).

123
234 Polit Behav (2015) 37:221–251

.8

.7

Social Conservatism
.6

.5

.4

.3
o l e e e
ntr tiv siv tiv
Co rva re
s rva
nse rog nse
co sp co
cs lic ls
th oli tho eli
ca
Ca Ca an
g
Ev

Fig. 1 Social Conservatism Scale means by experimental group (Study 1). Note Figure displays
experimental group means and 90 % confidence intervals

Table 1 Effects of treatments on participants’ opinions on social issues (Study 1)

Social conservatism Social conservatism
(w/identity terms)

Constant .619*** (.032) .192** (.073)

Catholics conservative -.013 (.046) .035 (.123)
Catholics progressive -.077* (.043) -.094 (.107)
Evangelicals conservative -.070* (.043) .026 (.110)
Identity strength – .611*** (.098)
Catholics conservative 9 Identity – -.100 (.162)
Catholics progressive 9 Identity – .007 (.142)
Evangelicals conservative 9 Identity – -.148 (.148)
N 209 206

Table entries are unstandardized OLS regression coefficients with standard errors in parentheses
*** p B .001; ** p B .01; * p B .05 (one-tailed tests)

the lowest identifiers, supporting H3. With respect to the identity moderation
hypothesis, the coefficient on Catholics Progressive*Identity Strength is essentially
zero, against expectations. The coefficient on the Evangelicals Conservative*Iden-
tity Strength interaction is substantial and in the expected direction (b = -.148) but
misses statistical significance (p = .16, one-tailed). In Fig. 2, this interactive effect

123
Polit Behav (2015) 37:221–251 235

Control Evangelicals cons.

.8

Social Conservatism
.6

.4

.2

0
0 .25 .5 .75 1
Identity Strength

Fig. 2 ‘‘Evangelicals conservative’’ treatment interacted with identity (Study 1). Note Figure displays
predicted values for experimental groups at five levels of identity with 90 % confidence intervals at each
level. Values calculated and graphed using the margins set of commands in Stata 12

is graphed along with its confidence intervals at each level of the Catholic identity
variable. While the treatment effect is clearly not significant at the first three levels
of identity, it is significant at the top two levels. Thus, there is some limited support
for H4.22
Finally, we can turn to the emotion measures to test H5. We would expect typical
sample members—who began the study indicating they were socially conserva-
tive—to feel the most pride in the ‘‘Catholics are conservative’’ condition, as they
reflect on the fact that their values are in line with a majority of American Catholics,
and less pride in the other two conditions, perceiving their values to be out-of-step
with a progressive majority or in sync with a disliked out-group. The mirror
opposite pattern is expected for shame, with the least shame expected in the
‘‘Catholics are conservative’’ condition and more shame in the other two conditions.
Note that, all else equal, people tend to express much more pride than shame;
therefore, contrasts among treatment groups are examined for each emotion
separately.
The expected patterns emerge in Fig. 3. Difference-in-means tests were
conducted and most of the relevant comparisons reach statistical significance. With
respect to pride, the difference is greatest between those in the Catholics vs.
22
Similar results are obtained if this treatment effect is estimated separately for those with Catholic
identities above vs. at/below the scale midpoint. Those with strong Catholic identities appeared to shift
their views in the progressive direction in response to the ‘‘Evangelicals are conservative’’ stimulus
(b = -.126, p B .01) but weak identifiers did not.

123
236 Polit Behav (2015) 37:221–251

.5 .5

.4 .4

Average Shame
Average Pride
.3 .3

.2 .2

.1 .1

0 0
Catholics cons. Catholics prog. Evangelicals cons. Catholics cons. Catholics prog. Evangelicals cons.

Fig. 3 Average levels of pride and shame in response to treatments (Study 1)

Evangelicals conservative conditions (p B .01); a near-significant difference

emerges for the Catholics are conservative vs. progressive contrast (p = .13). With
respect to shame, those in the Catholics are conservative condition said they
experienced none whatsoever; this is in contrast to the greater shame in the
Catholics are progressive (p B .05) and Evangelicals are conservative (p = .10)
conditions.23

Study 1 Discussion

Overall, these findings support the relevant hypotheses. Catholics who learned that
other Catholics held progressive family values expressed more progressive views in
the post-test than the control group, and Catholics who learned that Evangelicals
were socially conservative also expressed more progressive views in the post-test.
These findings are especially compelling when one considers that most practicing
Catholics are respectful of a conservative Church hierarchy and likely less open to
progressive change than others, and that the conformity and polarization described
occurred in a sample with a median age of 54, an age when values and attitudes are
at their most stable (Sears and Levy 2003). Also, as expected, the strongest Catholic
identifiers in the sample were overwhelmingly more likely to be socially
conservative. It is all the more intriguing, then, that high identifiers appeared to

23
The total N available is too small for a test of H6 (emotion moderation). The reason for the small N is
as follows: (1) For methodological reasons, the analysis cannot include the control group (because the
emotion questions asked for reactions to the treatments) or participants who said in the pre-test that they
disagreed with socially conservative Church teachings (a different emotional pattern is expected from
such individuals, and there are too few to analyze separately). (2) There was significant attrition prior to
the emotion questions because they followed a difficult screening question at the end of the study. This
attrition is statistically unrelated to treatment group and, thus, does not threaten the causal inferences.
The final N is 31.

123
Polit Behav (2015) 37:221–251 237

be more likely than low identifiers to shift in a progressive direction when they were
exposed to conservative out-group norms.24
In addition, the data support the idea that self-conscious emotions are
automatically evoked when group identifiers encounter group norms. On average,
study participants who expressed support for traditional values at the outset of the
study felt the most proud and the least ashamed when they learned that most
Catholics supported those values (as they did); by contrast, they felt less proud and
more ashamed when they learned that most Catholics opposed those values or that
most Evangelicals supported them. This said, these findings do not demonstrate a
causal role for these emotions in the observed influence. To test this expectation, we
turn to a second experiment that exogenously varies the emotions pride and
embarrassment.

Study 2: Self-Conscious Emotions and Economic Individualism Experiment

Study Sample and Design

Approximately 100 participants were recruited from the University of Michigan

Introductory Psychology subject pool.25 Students participated in person, in groups.
Each student received a paper-and-pencil questionnaire containing a pre-test,
stimulus, and post-test. Students were debriefed regarding the study aims and
methodology after completing the study.
The experiment had a 2 9 2 between-subjects design with random assignment of
treatments. Two factors were varied: information regarding whether most fellow
University of Michigan undergraduates supported or opposed economic individu-
alism, and whether or not participants’ self-conscious emotions were primed. The
stimuli included the following. With respect to the opinion information, half of
participants read poll results that most University of Michigan undergraduates
supported economic individualism; the other half read poll results that most
University of Michigan undergraduates opposed economic individualism. All poll
results were fabricated for the purposes of experimental control.26 With respect to
emotion priming, approximately half of the participants read emotional scenarios
intended to prime either pride or embarrassment. (Embarrassment was chosen over
shame for ethical reasons.) In accord with standard practice in psychology, the
emotion primes preceded the opinion stimulus (e.g., see Small and Lerner 2005;
24
The identity moderation hypothesis (H4) was only partially supported, however; strong identifiers
were not more likely than low identifiers to shift in a progressive direction when exposed to progressive
in-group norms. One explanation for this null result is that some of the most devoted Catholics, who also
tended to be the most conservative, may have dismissed the progressive Catholics depicted in the study as
not ‘‘true Catholics.’’ Borrowing again from Turner (1991), if an in-group norm is too different from a
person’s personal beliefs, he or she may choose to redraw group boundaries—separating him or herself
from the former in-group—rather than conform. Exploring when such identity redefinition occurs is an
important topic for further study.
25
Demographics are as follows: 47 % Democratic, 23 % Republican, and 31 % Independent or ‘‘other.’’
78 % white, 8 % African American, 7 % Asian, 5 % Hispanic, and 2 % Native American. Men made up
52 % of the sample. The mean age was 19.
26
A post-test probe did not turn up any skepticism with regard to the veracity of these stimuli.

123
238 Polit Behav (2015) 37:221–251

Table 2 Experimental design of economic individualism experiment (Study 2)

No emotion Emotion primed
(pride or embarrassment)

Majority supports economic individualism 23 20

Majority opposes economic individualism 22 20

Zajonc 1980). Primes were presented in the form of a question to decrease

awareness of the manipulation and were unrelated to economic individualism (as
well as to politics in general) to decrease the likelihood of confounding effects.27
See Table 2 for the basic design. The stimuli are available in Appendix 2.
The logic of how the opinion and emotion stimuli were expected to work together
requires some explanation. First, it was expected that nearly all participants would
support economic individualism in the abstract (as it is a very popular value in the
US). Thus, the experimental design assumes participant commitment to this general
value. Two Likert-type statements at the start of the questionnaire—‘‘individuals
should strive to be financially self-reliant’’ and ‘‘one ought to work hard in life’’ (the
same statements used in the opinion stimulus)—effectively screened out the small
percentage of participants (15 %) who did not support economic individualism.28
Thus, all analyzed participants who received the pro-individualism poll results
learned they were in the majority; those who received the anti-individualism poll
results learned they were in the minority. Second, the emotion primes were tailored
to these specific circumstances: those in the pro-individualism condition also
assigned an emotion prime read pride-evoking scenarios, intended to increase the
likelihood that they would feel pride after finding themselves in step with peers;
those in the anti-individualism condition also assigned an emotion prime read
embarrassment-evoking scenarios, intended to increase the likelihood that they
would feel embarrassment after finding that their opinions were deviant.

Measures

The dependent measure is the average of responses to two questions on government

social welfare policy which heavily implicate the value economic individualism.
The first question asked participants whether the federal government ought to give
college scholarships to all low-income high school graduates or only to those who
have taken rigorous courses in high school. The second asked participants whether
the federal government should provide welfare benefits to all low-income single
mothers or only to those who are willing to work for those benefits. (See Appendix 2

27
In other words, ‘‘incidental affect’’ (orthogonal to study content) rather than ‘‘integral affect’’ (arising
in response to related content) is examined (see Blanchette and Richards 2010). The former allows one to
more cleanly isolate the causal influence of emotion on the dependent variable; the latter is usually
intertwined with cognitive content related to the study.
28
In most cases, these individuals expressed mild opposition to just one of the statements. While there
are too few cases for separate analysis here, note that adding these individuals to the analyses that follow
does not considerably alter the results.

123
Polit Behav (2015) 37:221–251 239

for exact question wording.) Answers to these questions are correlated at .35.29 The
additive scale Social Welfare Policy was re-coded to range from 0 to 1 (m = .54) so
that relatively individualistic answers were represented by higher numbers. The
format of both questions was modeled on a format frequently used by the American
National Election Studies.
The post-test questionnaire also included measures of participants’ feelings. A
battery of emotion items was asked following the opinion stimulus to provide a
secondary test of emotional moderation. The instructions read: ‘‘Describe how you
feel right now by indicating the extent to which you feel each emotion.’’ Answer
choices, arrayed on a five-point scale, ranged from ‘‘not at all’’ to ‘‘extremely.’’30
The emotions were listed in alphabetical order. These questions were used to create
a Self-Conscious Emotion Intensity scale, also placed on the 0 to 1 interval, as well
as another emotion measure described in the Discussion and Conclusion. The Self-
Conscious Emotion Intensity scale is scored to reflect responses in the ‘‘pride’’
family (‘‘proud’’ and ‘‘self-confident’’) for participants in the pro-individualism/
majority treatment groups and responses in the ‘‘embarrassment’’ family (‘‘embar-
rassed’’ and ‘‘insecure’’) for participants in the anti-individualism/minority treat-
ment groups. This variable construction reflects the previous logic of the emotion
primes, where the specific emotion prime depended on whether the participant is in-
step with the majority (pride primed) or out-of-step with the majority (embarrass-
ment primed).31

Data Analyses

2 9 2 experiments are most easily analyzed and interpreted with ANOVAs, which
allow for straight-forward interpretation of the two ‘‘main effects’’ as well as the
interaction effect. Two separate ANOVAs were conducted—one with the emotion
treatment variable and one with the measured emotion variable.32 The experimental

29
Patterns of results presented below are similar when the variables are assessed separately.
30
This battery of questions was based on a widely-used emotion measure called the ‘‘Profile of Mood
States’’ (POMS) created by McNair and Droppleman (1971).
31
Note that it is inappropriate to test the moderating effect of emotion in this study by comparing the
opinion treatment effect across levels of embarrassment and pride separately and for all participants,
regardless of treatment. To illustrate, take the case of embarrassment. The meaning and effects of a high
level of embarrassment differ depending on whether it occurs in participants who received the pro-
individualism or the anti-individualism treatment. In response to the anti-individualism treatment, where
the participant is in the minority, embarrassment signals a participant is likely to conform, as expected;
however, in response to the pro-individualism treatment, the unusual circumstance of embarrassment in
response to being in the majority suggests a participant may instead deviate from the perceived norm
because he or she is, evidently, uncomfortable being in the mainstream. (A parallel, opposite, result would
occur if one concentrated on pride.) Thus, despite a focus on one emotion, the changing context means
such an analysis is not comparing like to like. Interacting the opinion treatment variable with the above-
described Self-Conscious Emotion Intensity scale better tests the emotion moderation hypothesis.
32
Randomization successfully balanced the treatment groups with respect to age and political ideology
but not sex or race. These two variables are therefore added to the analyses as controls; however, their
addition does not substantially alter the results of the analyses.

123
240 Polit Behav (2015) 37:221–251

Table 3 ANOVA results for three models (Study 2)

Model 1: Model 2: Model 3:
basic measured arousal
model emotion added

Opinion treatment .90 .72 .75

Emotion treatment .02 – –
Opinion treatment 9 emotion treatment 4.46* – –
Self-conscious emotion intensity – 1.83 .36
Opinion treatment 9 self-conscious emotion intensity – 6.06* 4.51*
Emotional arousal – – 2.64
Opinion treatment 9 emotional arousal – – .01
Sex control .18 .22 .13
Race control 5.12* 5.65* 5.76*
N 82 82 81

Cells contain F-statistics and are marked

** p B .01; * p B .05; ^ p B .10

set-up allows for tests of H1 (conformity) as well as H6 (emotion moderation).33 For

H1 to be fully supported, participants’ post-test views in the pro-individualism
treatment groups should be more ‘‘conservative’’ than participants’ post-test views in
the anti-individualism groups. For H6 to be supported, this difference—evidence of
majority influence—should be greater when the relevant emotions are greater. In
other words, there should be a positive interaction between the opinion treatment and
self-conscious emotions. Finally, note that no main effect for emotion is expected.
I.e., pride or embarrassment in and of itself should not cause individualism to
increase or decrease.
The first ANOVA includes the opinion treatment variable (coded 0/1), the exogenous
emotion treatment variable (0/1), and their interaction. The results are displayed in the
first column of Table 3. With respect to the first factor, majority opinion did not appear
to influence participants’ views on average F(1, 82) = .90, p = .35. However, the
interaction term is statistically significant: F(1, 82) = 4.46, p B .05, two-tailed.34 In
other words, the effect of the opinion stimuli changed when emotions were primed.
To clarify these relationships, experimental group means are graphed below in
Fig. 4. Without emotions primed, those in the anti-individualism treatment group
reported slightly more individualistic attitudes than those in the pro-individualism
treatment group, contrary to expectations; however, when pride or embarrassment
was primed, this relationship reversed. Participants appeared to conform to the
perceived majority, as expected.

33
Note that this study tests H6 in the context of conformity to in-group norms. Future research will be
needed to address the emotion moderation hypothesis with respect to polarization in response to out-
group norms.
34
While hypotheses with clear directional claims continue to be tested, the F ratio is akin to a two-tailed
test.

123
Polit Behav (2015) 37:221–251 241

Anti-Individualism Pro-Individualism
.8

.7

Social Welfare Policy

.6

.5

.4

.3
No Emotion Prime Emotion Prime

Fig. 4 Predicted values for opinion and emotion treatment groups (Study 2). Note Predicted values for
interaction between opinion and emotion treatments with controls. In figure, Sex is held to 1 (female) and
Race to 0 (white)

One drawback of the above analysis, however, is that it only takes into account
whether or not participants received an emotion prime. It does not include any
measure of participants’ subjective emotional experiences, including naturally
evoked pride or embarrassment in response to the opinion stimuli when the emotion
primes were not present. A second analysis was conducted, replacing assignment to
an emotion group with the Self-Conscious Emotion Intensity scale.35 See again
Table 3 (second column). Here, again, there is no main effect for the opinion
treatment, but the interaction between opinion and emotion is significant: F(1,
82) = 6.06, p B .05. The pattern of results is similar to that of the previous analysis.
See Fig. 5. No social influence appeared to occur among participants who reported
very low levels of self-conscious emotions. However, this quickly changes as we
advance up the emotion scale, with the gap between the two lines becoming
statistically significant at and above the emotion mid-point (.375).

Study 2 Discussion

In this experiment, the self-conscious emotions pride and embarrassment appeared

to increase individuals’ susceptibility to social influence by peers. No evidence
emerged for political conformity when student participants simply read opinion

35
Self-reported emotion is entered into the ANOVA as a continuous variable, i.e., the analysis is
technically an ANCOVA (with interaction). While this variable theoretically ranges from 0 to 1, the
actual range is 0–.75 because few participants reported experiencing extreme emotion.

123
242 Polit Behav (2015) 37:221–251

Anti-Individualism Pro-Individualim
.8

.6

Social Welfare Policy

.4

.2

0
0 .125 .25 .375 .5 .625 .75
Measured Emotion

Fig. 5 Predicted values by opinion treatment and subjective emotion intensity (Study 2). Note Predicted
values for interaction between opinion treatment and Self-Conscious Emotion Intensity measure with
controls. In figure, Sex is held to 1 (female) and Race to 0 (white). Differences in treatment group
estimates are statistically significant (p B .05) at and above .375

information and among those reporting lower-than-average self-conscious emo-

tional reactions. However, when pride or embarrassment was primed, and among
those experiencing higher-than-average self-conscious emotions, views appeared to
shift in the direction of majority opinion.
This said, one could suggest an important counterargument to the conclusion that
self-conscious emotions increased the likelihood of conformity in this experiment.
Perhaps emotions mattered, but not specifically self-conscious emotions. For
example, maybe the emotional stimuli were effective because they contributed to
emotional arousal (Lazarus 1991, p. 17), increasing participants’ attentiveness to the
opinion stimuli, and, therefore, opinion conformity. We can test this alternative
hypothesis by adding a measure of basic emotional arousal36 and its interaction with
the opinion treatment to the original ANOVA containing the measured self-conscious
emotion variables. See the third column of Table 3. The counterhypothesis is not
supported by the analysis. The Opinion 9 Emotional Arousal interaction term has an
F statistic of nearly 0. The F statistic on the Opinion 9 Self-Conscious Emotion
Intensity interaction declines slightly but easily remains statistically significant: F(1,
81) = 4.51, p B .05. Note that if each of the three basic emotions in the Emotional
Arousal measure is examined separately, a similar pattern is produced.

36
An average of three basic emotions: anxiety, anger, and enthusiasm (see, e.g., Panksepp 1994).

123
Polit Behav (2015) 37:221–251 243

Discussion and Conclusion

Overall, the pattern of results observed reflects the expectations set out in the six
hypotheses. The findings from the first experiment demonstrate the power of in-group
norms to influence and out-group norms to polarize as well as the relevance of
gradations of subjective identity to these processes. The first experiment’s findings also
suggest, but do not demonstrate, a causal role for self-conscious emotions. Results from
the second experiment directly support emotional causation. While these experimental
effects will likely decay over time (e.g., see Chong and Druckman 2013), they also stem
from a single (impersonal) interaction with a group norm. Repeated interactions would
likely reinforce attitudes and lead to stabilization over time.
Alternative explanations for the pattern of findings are worth considering,
however. Could the pattern of results in the two experiments be due to the more
common argument that social influence, generally speaking, is due to information
exchange and/or argument? This is unlikely given that the stimuli deliberately did
not include any substantive information or arguments. What about other more
‘‘rational’’ explanations, such as the possibility that study participants were
conforming for self-interested reasons (Chong 2000) or because they constructed
arguments to explain the majority’s view, persuading themselves in the process
(Mutz 1998)? These theories are also unlikely to explain the experimental results.
Theories such as Chong’s that emphasize self-interested motivations tend to focus
on public behavior, not private opinions, as were the focus of both experimental
studies. With respect to Mutz’s theory, those mechanisms could explain the in-
group influence observed in the first experiment but not the empirical patterns
related to out-group norms or self-conscious emotion.
In sum, the empirical results support the ‘‘social–emotional influence’’ frame-
work proposed herewith. That framework provides an explanation for a phenom-
enon—political conformity—that has long been underexplored and not well
understood as a result. Likewise, it provides a window into between-group
polarization, a topic highly relevant to contemporary U.S. politics. Social identity
allows us to understand which groups will influence a person, and in what way.
Emotion helps us to understand why people fall into line with in-group expectations
so readily despite exposure to diverse perspectives on politics.
In addition, the social–emotional perspective on conformity and polarization
helps to make sense of certain puzzling aspects of public opinion. For example, it
offers an explanation as to why norms, values, and attitudes are affectively charged
(Banaji and Heiphetz 2010; Rokeach 1973) and closely correlated with group
boundaries. The framework also provides insight into why people tend to feel proud
of shared group norms and to feel shame when they disregard them. Finally, the
theoretical framework contributes to the growing literature on emotions and politics.
Heretofore, most empirical emotions and politics scholars have focused on the basic
emotions of fear/anxiety, enthusiasm, and anger (e.g., Brader 2006; Gadarian 2010;
Huddy et al. 2007; Marcus et al. 2000; Valentino et al. 2011), leaving self-conscious
and other more complex emotions to the side. Given the many ways in which group
identity, social relationships, and social status intersect with politics, further
examination of the political relevance of self-conscious emotions would be fruitful.

123
244 Polit Behav (2015) 37:221–251

Acknowledgments A great many people have contributed to this article by sharing their insightful
observations, criticisms, and suggestions. I would like to especially thank my dissertation committee—
Ted Brader, Don Herzog, Don Kinder, and David Winter. I also received helpful comments from Kevin
Arceneaux, John Bullock, Eric Dickson, Erika Franklin Fowler, Eric Groenendyk, Marc Hetherington,
Nathan Kalmoe, Chris Karpowitz, Skip Lupia, George Marcus, Roger Masters, Ngoc Phan, Stephanie
Preston, Lynn Sanders, Laura Stoker, and three anonymous reviewers. I am grateful to The University of
Michigan for generous financial support during my graduate education as well as to the Catholic leaders
and community members in Michigan who made Study 1 possible.

Appendix 1: Religious Identity and Social Conservatism Experiment (Study 1)

Stimuli

PLEASE READ THE TEXT BELOW CAREFULLY. WHEN YOU ARE DONE, ADVANCE TO THE
NEXT PAGE TO ANSWER SOME RELATED QUESTIONS

[Headline insert]
As you may know, the issue of ‘‘family values’’ continues to be discussed in the media. From time-to-
time, public opinions polls are carried out to find out what different types of Americans believe
regarding family values. [Body insert A] According to the survey:
[Body insert B]
What about you? We would like to know your opinion on family values.

Catholics Conservative Catholics Progressive Evangelicals Conservative

Condition Condition Condition

Headline Recent Polls Indicate Recent Polls Indicate Recent Polls Indicate
insert Catholics Are Strong Catholics Are Less Evangelicals Are Strong
Supporters of Family Supportive of Family Supporters of Family
Values Values Values
Body For example, one recent For example, one recent For example, one recent
insert A survey indicates that survey indicates that survey indicates that
American Catholics today American Catholics today American Evangelical (or
continue to strongly seem to question the ‘‘born again’’) Christians
support traditional family importance of traditional today are strong supporters
values family values of traditional family values
Body The majority of Catholics A majority of Catholics say A majority of Evangelicals
insert B who marry stay married one can be a good Catholic say that divorce should be
and never divorce without obeying the avoided, even in the event
Church’s teaching on of an unhappy marriage
divorce
A majority of Catholics A majority of Catholics say A majority of Evangelicals
oppose abortion one can be a good Catholic oppose abortion
without obeying the
Church’s teaching on
abortion
A majority of Catholics A majority of Catholics say A majority of Evangelicals
oppose gay marriage the Church’s opposition to oppose gay marriage
gay marriage is not very
important to them

123
Polit Behav (2015) 37:221–251 245

Measures37

Social Conservatism Scale

Divorce Subscale
Divorce in this country should be more difficult to obtain than it is now.
Divorce is usually the best solution when a couple can’t seem to work out their
marriage problems.
Premarital Sex Subscale
It is wrong for a man and a woman to have sexual relations before marriage.
It’s a good idea for a couple who intend to get married to live together first.
Gay Rights Subscale
Sexual relations between two adults of the same sex is wrong.
Homosexual couples should have the right to marry one another.
Sex and Young People Subscale
Sex education has no place in the nation’s public schools.
Methods of birth control should be available to teenagers who need them.
Abortion Question
There has been discussion about abortion during recent years. Which one of the
opinions below best represents your view? By law, abortion should never be
permitted./The law should permit abortion only in the case of rape, incest, or
when the woman’s life is in danger./The law should permit abortion for reasons
other than rape, incest, or danger to the woman’s life, but only after the need for
the abortion has been clearly established./By law, a woman should always be able
to obtain an abortion as a matter of personal choice./Other

Catholic Identity Scale

How important is being Catholic to you?

How well does the term ‘‘Catholic’’ describe you?

Emotional Reactions to Stimuli

Did the information make you feel [proud/ashamed]?

37
Seven answer categories for Likert items ranged from ‘‘strongly agree’’ to ‘‘strongly disagree.’’
Identity items included five possible responses, from, e.g., ‘‘not very important’’ to ‘‘extremely
important.’’ Emotion questions had five possible responses, from ‘‘not at all’’ to ‘‘extremely.’’

123
246 Polit Behav (2015) 37:221–251

Appendix 2: Self-Conscious Emotions and Economic Individualism Experiment

(Study 2)

Pro-Individualism Stimulus

HOW DO YOU COMPARE?

In the spring of 2005, researchers at UCLA (University of California, Los
Angeles) carried out opinion polls of college students at ten major universities
throughout the United States, including The University of Michigan. UCLA
researchers asked random samples of undergraduate students at each university
about what majors they chose and why, about study habits and extracurricular
activities, about Internet use, about their consumer habits, and, finally, about various
social attitudes and political opinions.
Two of the survey questions focused on attitudes regarding ‘‘economic
individualism.’’ According to results published last year in Public Opinion
Quarterly, most University of Michigan students agree with this principle.
64 % of University of Michigan undergraduates either agreed or strongly agreed
with the statement ‘‘Individuals should strive to be financially self-reliant.’’

Neutral
12%

Disagree/Strongly Agree/Strongly
Disagree Agree
24% 64%

N = 325

61 % of University of Michigan undergraduates either agreed or strongly agreed

with the statement ‘‘One ought to work hard in life.’’

Neutral
16%

Disagree/Strongly Agree/Strongly
Disagree Agree
23% 61%

N=318

19. You were asked to respond to the same two statements on page 2. Compare
your answers to the published data and then check the appropriate response below.

123
Polit Behav (2015) 37:221–251 247

I am in agreement with the majority of University of Michigan undergraduate

students….
____ on both survey items.
____ on one of the survey items.
____ on neither of the survey items.

Pride Stimulus

Take a moment to imagine each of the following scenarios, focusing on how each
situation would make you feel. Then circle the situation that you believe would
make you feel the best.

a. You leave school in April to spend the summer at home. One of your goals is to
improve the way you look—get in shape, buy some new clothes, maybe get a
new haircut, etc. When you return to school in the fall, everyone tells you how
great you look. You go to a party the first weekend back, and two cute guys (or
girls) approach you during the evening and ask you out.
b. You attend a family gathering over winter break with various family members.
One of your relatives asks you how school is going. As it happens, you got
straight As in the fall semester and have secured a really prestigious summer
internship, all of which you tell your relatives. The group gushes about your
accomplishments, and your mom looks especially pleased.
c. You are standing on the curb of a busy street, waiting for the light to turn green
so that you can cross, when you see a little girl wander away from her mother
and dart into the street. You run after her into the traffic, pick her up, and return
her to her mother. A small crowd that has gathered on the sidewalk to watch
breaks into applause.

Embarrassment Stimulus

Take a moment to imagine each of the following scenarios, focusing on how each
situation would make you feel. Then circle the situation that you believe would
make you feel the worst.

a. You are on a first date with someone you really like. You go to dinner, then to a
party. As the evening is coming to an end, both of you are sitting together on a
couch. Your date leans in close to you, and you’re thinking it is finally time for
a kiss. But, instead, your date whispers to you, ‘‘Sorry to tell you this, but, uh,
the zipper on your pants has been down since we left the restaurant.’’
b. It’s a warm spring day, and you are walking through the Diag, which is filled
with students socializing, studying, playing Frisbee, etc. All of a sudden you
trip and, with a loud grunt, fall down. Several books and the bag you had been

123
248 Polit Behav (2015) 37:221–251

carrying scatter all around you. Everyone on the Diag seems to stop what they
are doing to stare at you sprawled out on the pavement.
c. You are attending the wedding ceremony of a family member. The room is
quiet, except for the bride and groom exchanging their vows. All of a sudden
you get a case of the hiccups. Hiccup! Hiccup! A number of people sitting
around you turn to you and say ‘‘shhhhh……’’ You put your hand over your
mouth, but you can’t stop hiccupping.

Dependent Measures

The federal government currently gives money for college to many low-income
high school graduates. Some people believe that these college grants should go only
to those low-income graduates who have taken rigorous courses in high school.
Suppose these people are at one end of a scale, at point 1. Other people feel that
such college grants should go to all low-income high school graduates, regardless of
what courses they have taken. Suppose these people are at the other end, at point 7.
And, of course, some other people have opinions somewhere in between, at points
2–6. Where would you place yourself on this scale?
Currently, the government in Washington provides aid to low-income, single
mothers who have dependent children; this program is typically referred to as
‘‘welfare.’’ Some people feel that the government should require these women to
work in order to receive welfare benefits. Suppose these people are at one end of the
scale, at point 1. Others feel that the government should provide welfare regardless
of work status. Suppose these people are at the other end, at point 7. And, of course,
some other people have opinions somewhere in between, at points 2–6. Where
would you place yourself on this scale?

References

Abrams, D., Wetherell, M., Cochrane, S., Hogg, M. A., & Turner, J. C. (1990). Knowing what to think by
knowing who you are: Self-categorization and the nature of norm formation, conformity and group
polarization. British Journal of Social Psychology, 29, 97–119.
Alwin, D. F., Cohen, R. L., & Newcomb, T. M. (1991). Political attitudes over the lifespan: The
Bennington women after fifty years. Madison, WI: University of Wisconsin Press.
Asch, S. E. (1951). Effects of group pressure upon the modification and distortion of judgments. In H.
Guetzkow (Ed.), Groups, leadership, and men: Research in human relations. Pittsburgh, PA:
Carnegie Press.
Banaji, M. R., & Heiphetz, L. (2010). Attitudes. In S. T. Fiske, D. T. Gilbert, & G. L. Lindzey (Eds.),
Handbook of social psychology (5th ed.). New York, NY: Wiley.
Beck, P. A. (2002). Encouraging political defection: The role of personal discussion networks in partisan
desertions to the opposition party and Perot votes in 1992. Political Behavior, 24(4), 309–337.
Berelson, B. R., Lazarsfeld, P. F., & McPhee, W. N. (1954). Voting: A study of opinion formation in a
presidential campaign. Chicago, IL: University of Chicago Press.
Blanchette, I., & Richards, A. (2010). The influence of affect on higher level cognition: A review of
research on interpretation, judgment, decision making and reasoning. Cognition and Emotion, 24(4),
561–595.

123
Polit Behav (2015) 37:221–251 249

Bolsen, T. (2013). A light bulb goes on: Norms, rhetoric, and actions for the public good. Political
Behavior, 35, 1–20.
Brader, T. (2006). Campaigning for hearts and minds: How emotional appeals in political ads work.
Chicago, IL: University of Chicago Press.
Brown, R. (2000). Social identity theory: Past achievements, current problems and future challenges.
European Journal of Social Psychology, 30, 745–778.
Bullock, J. G., Green, D. P., & Hal, S. E. (2010). Yes, but what’s the mechanism? (Don’t expect an easy
answer). Journal of Personality and Social Psychology, 98(4), 550–558.
Campbell, A., Converse, P. E., Miller, W. E., & Stokes, D. E. (1960). The American voter (Unabridged
ed.). Chicago, IL: University of Chicago Press.
Chong, D. (2000). Rational lives: Norms and values in politics and society. Chicago, IL: University of
Chicago Press.
Chong, D., & Druckman, J. N. (2013). Counterframing effects. The Journal of Politics, 75(1), 1–16.
Clark, R. D., & Maass, A. (1988). The role of social categorization and perceived source credibility in
minority influence. European Journal of Social Psychology, 18, 381–394.
D’Antonio, W. V. (2007). American Catholics today: New realities of their faith and their Church.
Lanham, MD: Rowman & Littlefield.
Davis, J. A., Smith, T. W., & Marsden, P. V. (2008). General social surveys, 1972–2008: Cumulative
codebook. Chicago, IL: National Opinion Research Center.
Deutsch, M., & Gerard, H. B. (1965). A study of normative and informational social influences upon
individual judgment. In H. Proshansky & B. Seidenberg (Eds.), Basic studies in social psychology.
New York, NY: Holt, Rinehart and Winston.
Djupe, P. A., & Gilbert, C. P. (2008). The political influence of churches. Cambridge: Cambridge
University Press.
Downs, A. (1957). An economic theory of democracy. New York, NY: Harper and Row.
Elster, J. (1999). Alchemies of the mind: Rationality and the emotions. Cambridge: Cambridge University
Press.
Feldman, S. (1999). Economic values and inequality. In J. P. Robinson, P. R. Shaver, & L. S. Wrightsman
(Eds.), Measures of political attitudes. San Diego, CA: Academic Press.
Feldman, S., & Johnston, C. (2014). Understanding the determinants of political ideology: Implications of
structural complexity. Political Psychology, 35(3). doi:10.1111/pops.12055.
Festinger, L. (1950). Informal social communication. Psychological Review, 57, 271–282.
Gadarian, S. K. (2010). The politics of threat: How terrorism news shapes foreign policy attitudes. The
Journal of Politics, 72(2), 469–483.
Gerber, A. S., Green, D. P., & Larimer, C. W. (2008). Social pressure and vote turnout: Evidence from a
large-scale field experiment. American Political Science Review, 102(1), 19–31.
Gerber, A. S., Green, D. P., & Larimer, C. W. (2010). An experiment testing the relative effectiveness of
encouraging voter participation by inducing feelings of pride or shame. Political Behavior, 32,
409–422.
Gerber, A. S., & Rogers, T. (2009). Descriptive social norms and motivation to vote: Everybody’s voting
and so should you. The Journal of Politics, 71(1), 178–191.
Haslam, S. A., Ellemers, N., Reicher, S. D., Reynolds, K. J., & Schmitt, M. T. (2010). The social identity
perspective today: An overview of its defining ideas. In T. Postmes & N. R. Branscombe (Eds.),
Rediscovering social identity: Key readings. New York, NY: Psychology Press.
Hogg, M. A., & Reid, S. A. (2006). Social identity, self-categorization, and the communication of group
norms. Communication Theory, 16, 7–30.
Huckfeldt, R., & Sprague, J. (1987). Networks in context: The social flow of political information.
American Political Science Review, 81, 1197–1216.
Huckfeldt, R., & Sprague, J. (1995). Citizens, politics, and social communication: Information and
influence in an election campaign. Cambridge: Cambridge University Press.
Huddy, L. (2001). From social to political identity: A critical examination of social identity theory.
Political Psychology, 22(1), 127–156.
Huddy, L., Feldman, S., & Cassese, E. (2007). On the distinct political effects of anxiety and anger. In W.
R. Neuman, G. E. Marcus, M. MacKuen, & A. N. Crigler (Eds.), The affect effect: Dynamics of
emotion in political thinking and behavior. Chicago, IL: University of Chicago Press.
Huddy, L., & Khatib, N. (2007). American patriotism, national identity, and political involvement.
American Journal of Political Science, 51, 63–77.

123
250 Polit Behav (2015) 37:221–251

Imai, K., Tingley, D., & Yamamoto, T. (2012). Experimental designs for identifying causal mechanisms.
Journal of the Royal Statistical Society, 176(1), 5–51.
Lazarus, R. S. (1991). Emotion and adaptation. Oxford: Oxford University Press.
Lazer, D. (2011). Networks in political science: Back to the future. PS. Political Science & Politics,
44(1), 61–68.
Leege, D. C., Wald, K. D., Krueger, B. S., & Mueller, P. D. (2002). The politics of cultural differences:
Social change and voter mobilization strategies in the post-New Deal period. Princeton, NJ:
Princeton University Press.
Lewis, M. (2000). Self-conscious emotions: Embarrassment, pride, shame, and guilt. In M. Lewis & J.
M. Haviland-Jones (Eds.), Handbook of emotions (2nd ed.). New York, NY: Guilford Press.
Luhtanen, R., & Crocker, J. (1992). A collective self-esteem scale: Self-evaluation of one’s social
identity. Personality and Social Psychology Bulletin, 18, 302–318.
Mackie, D. M., Maitner, A. T., & Smith, E. R. (2009). Intergroup emotions theory. In T. D. Nelson (Ed.),
Handbook of prejudice, stereotyping, and discrimination. New York, NY: Psychology Press.
Marcus, G. E., Neuman, W. R., & MacKuen, M. (2000). Affective intelligence and political judgment.
Chicago, IL: University of Chicago Press.
Marques, J. M., Abrams, D., & Serôdio, R. S. (2001). Being better by being right: Subjective group
dynamics and derogation of in-group deviants when generic norms are undermined. Journal of
Personality and Social Psychology, 81(3), 436–447.
McClurg, S. D. (2006). The electoral relevance of political talk: Examining disagreement and expertise
effects in social networks on political participation. American Journal of Political Science, 50(3),
737–754.
McNair, M. L., & Droppleman, L. F. (1971). Profile of mood states. San Diego, CA: Educational and
Industrial Testing Service.
Milgram, S. (1992). The individual in a social world: Essays and experiments (2nd ed.). New York, NY:
McGraw-Hill Inc.
Miller, R. S. (2007). Is embarrassment a blessing or a curse? In J. L. Tracy, R. W. Robins, & J.
P. Tangney (Eds.), The self-conscious emotions: Theory and research. New York, NY: Guilford
Press.
Morton, R. B., & Williams, K. C. (2010). Experimental political science and the study of causality: From
nature to the lab. Cambridge: Cambridge University Press.
Moskalenko, S., McCauley, C., & Rozin, P. (2006). Group identification under conditions of threat:
College students’ attachment to country, family, ethnicity, religion, and university before and after
September 11, 2001. Political Psychology, 27(1), 77–97.
Mutz, D. C. (1998). Impersonal influence: How perceptions of mass collectives affect political attitudes.
Cambridge: Cambridge University Press.
Nadeau, R., Cloutier, E., & Guay, J.-H. (1993). New evidence about the existence of a bandwagon effect
in the opinion formation process. International Political Science Review, 14(2), 203–213.
Nelson, T. E., Gwiasda, G., & Lyons, J. (2011). Vilification and values. Political Psychology, 32(5),
813–835.
Nickerson, D. W. (2008). Is voting contagious? Evidence from two field experiments. The American
Political Science Review, 102(1), 49–57.
Noelle-Neuman, E. (1993). The spiral of silence: Public opinion—our social skin (2nd ed.). Chicago, IL:
University of Chicago Press.
Panagopoulos, C. (2010). Affect, social pressure and prosocial motivation: Field experimental evidence
of the mobilizing effects of pride, shame and publicizing behavior. Political Behavior, 32, 369–386.
Panksepp, J. (1994). The basics of basic emotions. In P. Ekman & R. J. Davidson (Eds.), The nature of
emotion: Fundamental questions. New York, NY: Oxford University Press.
Pew Research Center. (2008). A portrait of American Catholics on the eve of Pope Benedict’s visit.
Washington, DC: Pew Research Center Publications.
Putnam, R., & Campbell, D. E. (2010). American grace: How religion divides and unites us. New York,
NY: Simon & Schuster.
Rokeach, M. (1973). The nature of human values. New York, NY: The Free Press.
Scheff, T. (1988). Shame and conformity: The deference-emotion system. American Sociological Review,
53, 395–406.
Sears, D. O., & Levy, S. (2003). Childhood and adult political development. In D. O. Sears, L. Huddy, &
R. Jervis (Eds.), Oxford handbook of political psychology. Oxford: Oxford University Press.
Sherif, M. ([1936] 1966). The psychology of social norms. New York, NY: Harper & Row.

123
Polit Behav (2015) 37:221–251 251

Sinclair, B. (2012). The social citizen: Peer networks and political behavior. Chicago, IL: University of
Chicago Press.
Small, D. A., & Lerner, J. S. (2005). Emotional policy: Personal sadness and anger shape judgments about
a welfare case. Political Psychology, 29, 149–168.
Sokhey, A. E., & Djupe, P. A. (2011). Interpersonal networks and democratic politics. PS. Political
Science & Politics, 44(1), 55–59.
Tajfel, H. (1978). Social categorization, social identity, and social comparison. In H. Tajfel (Ed.),
Differentiation between groups: Studies in the social psychology of intergroup relations. London:
Academic Press.
Tajfel, H., & Turner, J. C. (1986). The social identity theory of intergroup behavior. In W. G. Austin & S.
Worchel (Eds.), Psychology of intergroup relations. Chicago, IL: Nelson-Hall Publishers.
Terry, D. J., & Hogg, M. A. (1996). Group norms and the attitude-behavior relationship: A role for group
identification. Personality and Social Psychology Bulletin, 22(8), 776–793.
Terry, D. J., Hogg, M. A., & White, K. M. (1999). The theory of planned behaviour: Self-identity, social
identity, and group norms. British Journal of Social Psychology, 38(3), 225–244.
Theiss-Morse, E. (2009). Who counts as an American? The boundaries of national identity. Cambridge:
Cambridge University Press.
Turner, J. C. (1985). Social categorization and the self-concept: A social cognitive theory of group
behavior. In E. Lawler (Ed.), Advances in group processes: Theory and research. Greenwich, CT:
JAI Press.
Turner, J. C. (1991). Social influence. Bristol, PA: Open University Press.
Turner, J. C., with Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S. (1987). Rediscovering
the social group: A self-categorization theory. Oxford: Blackwell.
Valentino, N. A., Brader, T., Groenendyk, E. W., Gregorowicz, K., & Hutchings, V. (2011). Election
night’s alright for fighting: The role of emotions in political participation. The Journal of Politics,
73(1), 156–170.
Wald, K. D., Owen, D. E., & Hill, S. S., Jr. (1988). Churches as political communities. The American
Political Science Review, 82, 531–548.
Walsh, K. C. (2003). Talking about politics: Informal groups and social identity in American life.
Chicago, IL: University of Chicago Press.
Wood, W., Pool, G. J., Leck, K., & Purvis, D. (1996). Self-definition, defensive processing, and influence:
The normative impact of majority and minority groups. Journal of Personality and Social
Psychology, 71, 1181–1193.
Zajonc, R. B. (1980). Feeling and thinking: Preferences need no inferences. American Psychologist, 35,
151–175.

123