Non-experimental research:

observational, archival, case-study research

9.63 Fall 2005

 Randomized experiments

● Experimenter randomly assigns Ss to conditions.

● Results can only be explained by
– chance
– the experimental manipulation
Limitations of randomized experiments

● Randomization can be
– impractical
– unethical
– expensive
● Examples:
Limitations of randomized experiments

● Randomization can be
– impractical
– unethical
– expensive
● Examples:
– Does marijuana use cause depression?
– How does growing up in a single-parent home affect
personality?
 A choice

Well-controlled Grapple with the

experiments human condition

– Visual psychophysics – Personality

– Motor control – Social psychology

Are non-experimental approaches scientific?

● Randomized experiments are rare in

– Geology
– Paleontology
– Astronomy
– Field biology
 Outline

● Observational studies
– Everyday behavior
– Fringe groups: cognitive dissonance
– “The Love Lab”
● Archival studies
– Prejudice and the Weakest Link

● Case studies
 Observational studies

● Discoveries can be made just by watching

people's behavior (no interventions or
experimental manipulations!)
Everyday behavior: slips of the tongue

● Speech errors provide insight into speech production

unanimity → unamity (syllable deletion)

easily enough → easy enoughly (suffix move)
tend to turn out → turn to tend out (word exchange)
my sister went to the Grand Canyon →
the Grand Canyon went to my sister (phrase exchange)

(Fromkin)

● Errors like these suggest that sentences are built from

units at many scales (syllables, words, phrases)
Everyday behavior: slips of the tongue

● Speech errors provide insight into semantics

(Johnston)
Everyday behavior: slips of the tongue

● Speech errors provide evidence of unconscious

motivations ?

● President of the Lower House of the Austrian

parliament:
– “Gentlemen, I take notice that a full quorum of members is
present and herewith declare the sitting closed.

(Freud)

 Everyday behavior: Personal Space

● Taking the T
 Everyday behavior: Personal Space

● Taking the T

1
2

 Everyday behavior: Personal Space

● Taking the T

1
3 4
2

 Everyday behavior: Personal Space

● Taking the T

1
3 4
2

● Individual seats “allow” people to sit closer to

each other
 Edward T Hall

● Proxemics: “the study of man's transactions as he

perceives and uses intimate, personal, social and
public space “
 Seating positions

Intimate, Competitive
Cooperative
 High status seats

● People in high status seats

– Talk more (even if seats are randomly assigned!)
– Are more likely to become jury foremen.
 Territorial behavior in parking lots

● "A study of more than 400 drivers at an Atlanta-

area mall parking lot found that motorists defend
their spots instinctively" (AP, May 13, 1997)
● Drivers took 7 seconds longer on average if
someone was waiting
● When honked, drivers took an extra 10 seconds
longer to leave.
● Men were faster to leave for a high status car.
Women were not.
 2003 Literature Prize

John Trinkaus, of the Zicklin School of Business, New York

City, for meticulously collecting data and publishing more
than 80 detailed academic reports about things that annoyed
him (such as:
– What percentage of young people wear baseball caps with
the peak facing to the rear rather than to the front;
– What percentage of pedestrians wear sport shoes that are
white rather than some other color;
– What percentage of swimmers swim laps in the shallow

end of a pool rather than the deep end;

– What percentage of automobile drivers almost, but not

completely, come to a stop at one particular stop-sign;

 Everyday behavior: Trinkaus

● 1994: Wearing Baseball-Type Caps: An Informal Look

Observed 407 people wearing baseball-type caps with

the peak in back in the downtown area and on 2 college
campuses (1 in an inner borough and 1 in an outer
borough) of a large city. About 40% of Ss in the
downtown area and at the inner-borough college wore
the cap with the peak to the rear, while about 10% of the
outer-borough college Ss had the peak to the rear.
 Everyday behavior: Trinkaus

● 1982: Stop Sign Compliance: An Informal Look

● 1983: Stop-Light Compliance -- Another Look
● 1993: Stop Sign Compliance: A Follow-Up Look
● 1997: Stop Sign Compliance: A Final Look
 Everyday behavior

● There are revealing observations that any of us

could have made if only we'd thought to look.
● Presumably there are many others left!
 Outline

● Observational studies
– Everyday behavior
– Fringe groups: cognitive dissonance
– “The Love Lab”
● Archival studies
– Prejudice and the Weakest Link

● Case studies
 Participant-Observer research

● Researchers join the group they want to study

(sometimes hiding their true motive)
● Example: “When prophecy fails”
– Festinger,
Riecken and Schachter study Marion
Keech and the seekers, who predict that the world will
be destroyed in a flood on December 21, 1954
 Cognitive Dissonance (Festinger)

● Cognitive dissonance:
– Holding
dissonant (inconsistent) beliefs is

uncomfortable.

– Prediction: people act to reduce dissonance whenever

it is experienced
Example: The Fox and the Grapes

Image removed due to copyright reasons.

 Cognitive Dissonance (Festinger)

● Experiment:
– Ss were given a boring task.
– Ss
were asked to tell an incoming subject that the task
was really quite interesting.
– Half were offered $1 for the lie – the rest were given
$20
– Question: when asked to express how they really felt
about the task, which group rated it as more
enjoyable?
 Belief disconfirmation (Festinger)

● When
– A person is deeply commited to a belief
– The belief is disconfirmed
– The believer has social support

the believer often increases his efforts to convince

others that he is correct
 Participant-Observer study

● Marian Keech and the seekers

– Predicted a flood on December 21
– Expected to be picked up by flying saucers
● Press release on December 22
– “...
the cataclysm was stayed by the hand of the God
of Earth”

(Festinger, Riecken, Schachter: When Prophecy Fails)

 Keech and the Seekers

● Only 2 out of 11 abandoned their belief

● After December 21, the group
– Soughtthe media coverage they had previously
avoided
– Welcomed outsiders to the group

– Released tapes that had previously been considered

secret
 Other apocalyptic predictions

● 1844: Millerites
● 1891: Mormons
● 1878, 1910, 1914: Jehovah's Witnesses
 Outline

● Observational studies
– Everyday behavior
– Fringe groups: cognitive dissonance
– “The Love Lab”
● Archival studies
– Prejudice and the Weakest Link

● Case studies
 Observational work in the lab

● Bring people into the lab and scrutinize their

behavior

● No manipulation of independent variables!

 Gottman's “Love Lab”

● What makes a relationship work?

● Method:
– film couples interacting in the lab

– Sensorsmeasure heart rate, how much

each person is sweating

Image removed due to copyright reasons.

 Coding interactions

● SPAFF (Specific Affect Coding system)

– 1. Disgust
– 2. Contempt
– 11. Whining
– 13. Stonewalling
● Coders assign one number per person per second
– Husband: 7, 7, 2, 2, 2, 2, 13, 14, ...
– Wife : 6, 6, 6, 6, 6, 8, 12, 12, ...
 Analysis

● Goals:
– 1) Explain and understand the coded data
– 2) Predict whether the relationship is likely to last
– 3) Identify interventions that may help the

relationship

 Early qualitative approaches

● What destroys a relationship ?

– The “Four horseman of the apocalypse”
● Criticism
● Contempt
● Defensiveness
● Stonewalling
 Mathematical Model (Murray)

Wife’s past Husband’s

Wife at time
= Constant
+ behaviour + influence on
t+1 wife

Wt+1 = a + r1 rW
1 W
t t+ IHW (Ht)

Husband’s Wife’s influence

Husband at
time = Constant
+ past + on husband
behaviour
t+1

H t+1 = b + r2 H
r2t Ht + IWH (Wt)

Emotional Inertia: r1 and r2 constants, |r| < 1

 Conflict-Avoiding Marriage

Conflict-Avoiding husband Conflict-Avoiding wife

IWH(W) IHW(H)
.3
-.05 .15
W .01 H
negative positive negative positive

Both have little effect

on each other in the
negative range
 (Murray)

Basic Marriage Types

Characteristics (stable and unstable types):

Volatile (S) – romantic, passionate, have heated arguments

with cycles of fights and sex
Validating (S) – calmer, intimate, value companionate marriage,
shared experience rather than individuality
Avoiders (S) – avoid confrontation and conflict, interact only in
positive range of their emotions
Hostile (U) – (mixed) conflict-avoiding wife, validating husband
Hostile-Detached (U) – (mixed) volatile wife, validating husband
 Relationship (H,W) Phase Space

25

20

15

H unhappy 10
H happy
W happy 5
W happy
Wife Score

-5

-10

H unhappy -15 H happy

W unhappy -20 W unhappy
-25
-20 -10 0 10 20
Husband Score

Figure by MIT OCW.

 Predicting marriage success

● Based on a fifteen minute conversation, Gottman

can predict with 90 % accuracy whether a couple
will still be together in 15 years.
● Therapy: changing interaction functions
 Outline

● Observational studies
– Everyday behavior
– Fringe groups: cognitive dissonance
– “The Love Lab”
● Archival studies
– Prejudice and the Weakest Link

● Case studies
 Archival studies

● Analyse existing datasets

(eg census data, voting records)
● These data have sometimes been collected at
great cost, and on a scale that far exceeds what a
single researcher could manage.
 Discrimination

An employer is looking to fill several jobs. Given

a man (M) and a woman (W) with identical
resumes, suppose that he always chooses M.

● Taste-based discrimination:
– The employer prefers not to interact with women, and

will incur a financial cost to avoid such interactions

● Information-based discrimination:
– The employer has no animus towards women, but
suspects that W is actually less able than M.
 Studying discrimination

● Measuring discrimination is hard.

● Distinguishing between taste-based and
information-based discrimination is hard.
● Levitt (2004) attempts to do both using data from
“The Weakest Link”
 The Weakest Link

● Rules:
– Contestants
take turns to answer questions – correct
answers add to the prize pool
– One player is voted off after every round

– At
the end of the show, the two remaining contestants
compete head to head for the prize pool.

● Incentives:
– Early: vote for weak players
– Later: vote for strong players
 Predictions:

Suppose that Bob discriminates against women:

● Taste-based:
– Bob votes disproportionately for women in both early
and late rounds
● Information-based:
– Bob votes disproportionately for women, but only in
the early rounds
 Results

No evidence of discrimination against women or

blacks.

Some evidence of
● Taste-based discrimination against older players
● Information-based discrimination against Hispanics

(NB: Levitt controls for within-game performance )

 Outline

● Observational studies
– Everyday behavior
– Fringe groups: cognitive dissonance
– “The Love Lab”
● Archival studies
– Prejudice and the Weakest Link

● Case studies
 Case studies

● Examine individual instances of some

phenomenon.
– eg HM

Image removed due to copyright reasons.

 Project Prakash (Sinha et al)

● Study people who were

born blind, but have
had their sight restored
by cataract surgery
Image removed due to copyright reasons.
 Conclusions

● Experiment design is not always like choosing a

recipe from a cookbook. Try to find creative
ways to study the questions you care about.
● There is interesting work to be done using
– data that are easily collected (seating patterns in
subways)
– publicly available data (The Weakest Link)

 Four Bodily distances (Hall)

1) intimate (0 to 18 inches)
2) personal-casual (1.5 to 4 feet)
3) social-consultive (4 to 10 feet)
4) public (10 feet and beyond)

Personal space varies across cultures.

Europeans tend to stand closer than Americans.
 Everyday behavior: Trinkaus

● 1993: Compliance With the Item Limit of the Food

Supermarket Express Checkout Lane: An Informal Look

75 15-min observations of customers' behavior at a food

supermarket showed that only about 15% of shoppers
observed the item limit of the express lane.... Results
indicate a tendency not to play by the rules in the
absence of meaningful real or imagined constraints.