- '-w—-—-a.- m-v—o-u --
3 I 'I, I .IIIIII IIIII t I..."- . f I u:
II III! I ”1‘ A: III I l: l“a- I0
1 . V - . ~..“Iq.;III

   

I I' III.II|I|
1’ I: II IIIII I>II[' MI: 15““ ‘IIN
II" '{III'I’gl ,IIII

    

- ‘l '0 [III
I ‘I'I‘IOIu II ~ I...II dl,‘
II. 1- "1. 1'." l""‘I "-~ IIIIIIIIII
I.'IIII1IX. .. I
' I.I “

3-.
l- ‘I&\

Aq‘ I‘ul‘

|l \fiiy'gI
. .; k
U‘

‘ . ,1
.‘IJII YI.‘ I 5'.“$ r l-v
" ‘If ‘ ' WI" '51,» , fIIImIfi

5"}3‘ . . 4
' .M luv} 1 I y.
Mdg I‘éfj 'N;
:1
I ,NIii-uIIi'J II‘II*T-“v'm.. ’t”.

NIVII I'III'ILiIIf‘
”IL,”

"£32,; I

  
   
  
 

'11

with: I {I u _ {P I'II‘I‘IT'

“I VII
Ir '. . La” II" .
”I : --‘ i
. ' ”I.

"I'MQ.

.1 1: II .I
. “I I.I _ I”
.IIII $1.1 lIII'I‘,"I I 3&4 . I; s
~I‘n' II,nlI'I.IIl $97k! ’I '
(. J'Il‘ . 1“: II".I 30' “1,
r 4'“ UR

'IJLvI
5- 1

  
 

I .‘i‘r I

um a mi?

. . “"1353? .. ’2 311:3} 1qu EPIIIII NAIF
1'I II

    
   

'uI Iv I I W..." H .. ' :1’ I l' > ‘HII'I ‘I. 0.1 I nII M I
“J 'I I-u-II‘ O 'I”.| .. .1' i . 3 U . P ‘ '| .

'II’I:.LIII'I 'IIII ”“4 II",.III o... .‘ ‘IIH'.‘I*.I .Ig‘IIIIIIII‘ II.III I.’I HI' H II MI I ‘lll ‘ I::I II :III
. ‘ ‘I .IIIIIII'I . r”. n ‘1 MM "1' ... I dc"- ' w l '. ,v "H ‘

I l , . )‘ III u If u t '1 u ‘ . II, ' I 1
.w r‘. LKI' ‘3" Ni '. " , " =' .
..I .. | I ' . "III" I IuI II. .

‘IWI 0“ "1mm NHL-M M3; 4; - Again; "’-rsa-.,.

Mill 11H! 11111311111111 '

312 300 34 0131

 

 

 

 

 

 

thiiblb

LIBRARY
Michigan State
University

 

This is to certify that the

dissertation entitled

THE EFFECT OF EXPECTED MEETING AND KNOWLEDGE ABOUT
OTHERS ON ESTIMATES OF ASSUMED BELIEF SIMILARITY
AND CONSENSUS

presented by

DANIEL M. STULTS

has been accepted towards fulfillinent
of the requirements for

Ph-DL degree in Psychology

 

Major professor

Lawrence A. Messe

Date 4-19-85

 

MS U i: an Affirmative Action/Equal Opportunity Institution 0-12771

 

 

MSU

 

 

RETURNING MATERIALS:
Piace in book drop to

 

 

 

LJBRARJES remove this checkout from
4—3“- your record. FINES W111
be charged if book is
returned after the date
stamped beiow.
~. I

 

 

gmfifiifig T

 

THE EFFECT OF EXPECTED MEETING AND KNOWLEDGE ABOUT OTHERS

ON ESTIMATES OF ASSUMED BELIEF SIMILARITY AND CONSENSUS

BY
Daniel M. Stults

A DISSERTATION

Submitted to
.Michigan State University
in partial fullfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Psychology

1985

32 7/ .

:1 -' “1
E >«:,~< r:

0
up.

ABSTRACT
THE EFFECT OF EXPECTED MEETING AND KNOWLEDGE ABOUT OTHERS

ON ESTIMATES OF ASSUMED BELIEF SIMILARITY AND CONSENSUS

BY
Daniel M. Stults

Previous research on assumed similarity and false
consensus has generated conflicting explanations of the
possible mechanisms which underlie peoples' tendency
to overestimate the extent to which others hold
the same opinions as themselves. Some researchers,
e.g., Ross, Greene, House (1977) have speculated that
the phemonemon of opinion projection is strictly due to
cognitive factors. In contrast, others, e.g., Miller and
Marks (1982), have speculated that a need for social
support motivates opinion projection.

The present study set out to test the motivation
versus cognition perspectives by manipulating variables
that were expected to engage motivational components
without significantly activating the cognitive mechanism
of availability. Subjects were placed in the position
of a review board member and asked to read descriptions
of police officers who were under review for either a
merit raise or a reprimand and to make a decision regarding
the officer under review. Following the supervisory
decision, subjects were systematically provided with
information regarding the decision of other subjects who

were purported to be participating in the same session.

In some cases, subjects were told that they would be
meeting with another subject to discuss their decision.
It was hypothesized that estimates of consensus and
assumed belief similarity would vary as a function of
expected meeting and information about others' decisions.
Consistent with previous research, subjects tended
to assume more belief similarity between themselves and
others when they expected a future meeting than when no
meeting was anticipated. Also as hypothesized,
significant amounts of false consensus were generated by
each of the experimental conditions. However, expected
meeting and knowledge about ohters' decisions did not
differentially effect estimates of consensus, nor was
there a high degree of association between measures of
consensus and belief similarity. A number of issues were
discussed in light of these results, including the
speculation that esteem is a potential moderator of
opinion projection. In addition, several potential
follow-up studies that would test the relative
contribution of motivational and cognitive mechanisms

were discussed.

ACKNOWLEDGEMENTS

I am indebted to many people for my progress through
graduate school and through the dissertation. Primary
among these people is Larry Messé who is one of the
nicest and most generous people I know. A mutual friend
once told me he is "almost perfect," and I agree. Norb
Kerr has also been an exceptional friend and advisor,
and was especially instrumental in helping on my first
publication. I would also like to thank Neal Schmitt
who took an interest in me which lead to several fun
projects and helped me feel good about myself as well.
And, Dan Ilgen is to be noted for complementing my graduate
education. Among others, Kelly Brennan was helpful in
the data collection. Dennis Fox was central in inducing
and reducing soul searching throughout this past year.
Mary-Ann Reinhart is to be thanked for good advise which
I did not follow. Also, I would like to thank Mark
Elliott, who seemed to be my only peer through school.
Most importantly, I am indebted to Pam Mishey, who has
contributed more to my personal growth than any other
person.

ii

TABLE OF CONTENTS

List of Tables vi
The Problem: Psychodynamics Versus Psychologic__l
False Consensus - Conceptual Beginnings 4

Operationalizing False Consensus 7

 

Nonmotivational Explanations of False Consensus_l4

 

 

 

 

 

 

 

 

 

 

 

 

Availability 14
Vividness 19
Selective Exposure 23
Fundamental Attribution Error 28
Motivational Explanations of False Consensus____32
Vested Interest 34
Self-Enhancement 36
Self-Justification 40
Social Support 43
Expected Meeting 47
Overview of the Present Study 49
Method 54
Subjects 54
Design 55

 

iii

Materials

 

Measures

 

Procedure

 

Results

 

Manipulation Checks

 

Relationship Between False Consensus

and Assumed Similarity

 

Planned Comparisons

 

Basic and Conceptual Replication of

Miller & Marks (1982)

 

Focal Comparisons

 

Additional Tests of the

Bolstering Hypothesis

 

Effect of Information About

Others' Decisions

 

Confidence About Decision and

False Consensus

 

Anxiety

 

Discussion

 

Replication of Miller and Marks (1982)
Threats to Esteem - A Potential

Underlying Mechanism

 

Stable Self-Serving Perceptions and

Opinion Projection

 

Directions for Future Research

 

Cognitive Manipulations

 

iv

55
56
58
60
60

63
65

65
68

7O

71

72

73

75

76

82

86

87
88

Attribution 9O

 

 

 

 

 

 

 

Adding Confirmatory Evidence 92
Personal Estimates of the Base Rate 93
Motivational Manipulations 94
Summary 95
Appendix A: Further Pilot Studies 98
Appendix B: Experimental Materials 101
List of References 115
Footnotes 123

 

TABLE
1.

LIST OF TABLES

From Ross, Greene, House (1977):
Perceived Consensus: Estimated Commonness

of Own and Alternative Behavioral Choices

Basic Experimental Design for the
Present Study

 

Frequency Count Indicating the Number of
Male and Female Subjects Who Indicated that
their Partner Was More Likely To Be Male,
Female, or Had an Equal Chance of Being
Either Male or Female

 

Correlations Between Assumed Similarity
and False Consensus Broken Down by Cell

Mean Assumed Similarity as a Function of

Expected Meeting and Knowledge About Others_

Mean Consensus Estimates as a Function of

Expected Meeting and Knowledge About Others_

Mean Level of Comfort as a Function of

Expected Meeting and Knowledge About Others_

Mean Level of Comfort with Meeting Others
for Male and Female Subjects as a Function
of Expected Meeting and Knowledge About
Others

 

vi

PAGE

6

51

62

64

67

69

74

75

THE PROBLEM: "PSYCHODYNAMI CS VERSUS PSYCHOLOGIC"

The hypothesized contribution of motivational and
cognitive influences to all levels of behavior has
,stirred debates regarding when these factors work
conjunctively versus disjunctively, and the relative
contribution of each (Miller and Ross, 1975; Ross, 1977;
Zuckerman, 1979; Nisbett and Ross, 1980; Tetlock and
Levi, 1980). Abelson (1983) called this debate a question
of hot (motivational) versus cold (nonmotivational)
cognitions, whereas Nisbett and Ross (1980) have
provided the fitting, though somewhat whimsical, labels
of psychodynamics and psychologic to motivated versus
structural information processing respectively.

This cognition versus motivation debate stems, in part,
from the work of attribution theorists (e.g., Kelley,
1967) who have suggested that people are information
processors who use systematic methods of accessing,
processing and retrieving data. Kelley called people
"intuitive scientists” because he believed they make
systematic assessments of information from the
.environment. For the last several years researchers have
tried to identify the qualities and limitations of
the intuitive scientist. These investigations have

1

2
attempted to examine both logico-deductive process and
mechanistic cognitive process. (cf. Kelley, 1967; Ross,
1977; Taylor and Fiske, 1978; Nisbett and Ross, 1980;
Harvey and Weary, 1984).

Researchers have postulated a number of strategies
that individuals potentially use in the interpretation
of, and interaction with, the environment. Some of these
mechanisms seem rational in nature (e.g., the assessment
of consensus, consistency and distinctiveness
information; the representativeness heuristic). Such
processes suggest a view of humans as a deductive,
reasoning organisms (even if the reasoning is poor and
the information is weak). Other proposed mechanisms
(e.g., top of the head phenomenon, and attitude
accessibility) suggest a more reflex like, mechanistic
processing of information. Both the rational and
mechanistic processes, however, have one thing in common
-- they basically are non-motivational mechanisms (i.e.,
information processing strategies that systematically
weight information in decision making).

Other researchers have taken a different
perspective in their interpretation of human judgment.
Without necessarily denying the role of information
processing mechanisms, these investigators have

postulated the influence of a second kind of component

3
which affects our interpretation of the environment --
motivational variables. The basic tenet of this
motivational perspective is that hedonic relevance affects
individuals' judgements, behaviors, and interpretations of
the environment.

The motivational perspective, which is usually
thought of as need- or drive-based, contends that
individuals' behaviors and judgments reflect the biasing
influence of concerns with physical or psychological
outcomes or personal welfare. A detailed discussion of
some motivational variables follows later in this paper.
However, it should be noted that these motive based
mechanisms include such things as self-justification
(Messé & Sivacek, 1979), self-enhancement (Miller &
Marks, 1982), social support (Sherman, Pressin, Chassin,
Olshavsky, Corty, 1984), and vested interest (Crano,
1983).

At this time, there is not enough empirical evidence
to endorse unequivocally one explanation over
another (Tetlock & Levi, 1982). Nonetheless, the
empirical, and directly verifiable predictions of the
two perspectives have allowed scholarly discussion of
these competing explanations. In fact, some investigators
have already claimed greater support for one perspective
or the other (Miller & Ross, 1975; Bradley, 1978;
Zuckerman, 1979). The present work attempted to shed

further light on the relative contribution of

4
motivational and nonmotivational influences on behavior
by examining phenomenon that has stirred a great deal of debate
between cognitive and motivational theorists -- opinion
projection? In particular, the present work examined the
concepts of assumed similarity and false consensus
within a context that attempted to elucidate factors
that were expected to moderate opinion projection.

The following section reviews false consensus, its

origin and definition.

False Consensus - Conceptual Beginnings

False consensus is the term used by Ross,
Greene, and House (1977) to identify the tendency in
people to view their own beliefs as typical. Such over-
generalization from one's own behavior has been widely
documented (Kelley & Stahelski, 1970; Ross, Greene,
House, 1977; Goethals, Allison, & Frost, 1979; Messé &
Sivacek, 1979; Miller & Marks, 1982; Gilovich, Jennings,
& Jennings, 1983; Krahé, 1983; Marks, 1983; Sherman,
Pressin, Chassin, Olshavansky, & Corty, 1983; Marks,
1984; Sherman, Pressin, & Chassin, 1984). Ross, Greene,
and House (1977) for example, had subjects decide
between two alternative actions, then predict the number
of peers who would choose as they had. Ross et al.

hypothesized that subjects would judge their choices

5
relatively common and appropriate whereas they would
view the unchosen alternative as uncommon, and possibly
inappropriate.

Ross et al. stress that the term relative
is critical in the formulation of the false consensus
phenomenon. They suggest that individuals who choose
extreme actions (i.e., revolutionaries, tight rope
walkers, or clerics who choose a life of celibacy) will
recognize that their choices are uncommon. In such
special cases, perceived uniqueness may be due to
people having direct access to, and a clear
knowledge of, the base rate of these activities. But, for
the most part, people have little or no knowledge of the
base rate of even the most common behaviors. For this
majority of behaviors, Ross et al. suggest estimates of
others' behaviors will be directly linked to our own
behavioral choices.

In their first study, Ross, Greene, and House (1977)
asked 320 college students to respond to one of four
questionnaires. In each case, subjects were presented
with a dichotomous choice for a hypothetical situation
(i.e., sign a release/choose not to sign; choose
individual paper/choose a group paper, etc.; see pp.
281-282). After making their own decision regarding how
they would personally choose, subjects were asked to

estimate the number of their peers who would choose as

Perceived Consensus:

6
Table 1

From Ross, Green, House (1977):
Estimated Commonness of Own and

Alternative Behavioral Choices (Study 1)

 

Rater's own choice
in hypothetically

Estimates of con-

SEDSUS:

estimated
percetage of raters

who would choose

 

 

Story described situation n(%) Option 1 Option 2 F
Supermarket Sign Release 53(663) 75.6 24.4 17.7
Story Not Sign Release 27(44‘) 57.3 42.7
Term Paper Choose individual 64(80%) 67.4 32.6 16.5
Story paper
Choose group paper 16(203) 45.9 54.1
Traffic Pay speeding fine 37(468) 71.8 28.2 12.8
Ticket Contest charge 43(54!) 51.7 48.3
story
Space Vote for cutback 32(40%) 47.9 52.1 4.9
Program Vote against cutback 48(60‘1 39.0 61.0
story
Summary of Choose option 1 186(58%) 65.7 34.3 49.1
Four. Choose option 2 134(42%) 48.5 51.5
stories
they had.
The results are presented in Table 1. For each

scenario, an analysis of variance was performed to test

the difference between subjects' estimates of the

number of peers who would select Option 1. The results

indicate that subjects who themselves chose Option 1

estimated that many more people would choose that option

than did subjects who chose Option 2.

7

Ross et al.'s third and fourth studies show a
similar pattern of results. In Study 3, the researchers
again presented subjects with a hypothetical situation:
How would they choose, if asked to wear a sandwich board
sign around campus (which said, ”Eat at Joe's" or
"Repent")? In each case, about half of the subjects
volunteered to wear the sign. When compared to the
estimates of subjects who refused to wear the sign,
subjects who volunteered typically estimated that twice
as many peers would agree to wear the sign. In Study 4,
subjects were actually asked to wear the sandwich board
sign. The results of this final study were virtually
identical to those of Study 3. The conclusion which Ross
et al. draw from these findings is that subjects do, in
fact, overestimate the extent to which they see their

own choices as common.

Operationalizing False Consensus

There are several different ways false consensus
has been operationalized. The most intuitive
operational definition (yet, a definition which has not
been used in empirical work) suggests that individuals
overestimate consensus in the direction of their own
choice relative to the actual base rate of peoples'
decisions. A second operationalization, the one

suggested by Ross and his colleagues, defines false

8
consensus as the difference between the estimates of
decision preference for individuals who made the same
choice as the subject versus those who chose
differently. Third, some authors (e.g., Sherman,
Presson, Chassin, Corty, Olshavsky, 1983) have labeled
as false consensus the relative differences in frequency
estimates between people who participate in a focal
behavior (i.e., the behavior about which subjects are
asked to make estimates of others' responses) versus
those who do not participate. For instance, Sherman and
his colleagues examined the frequency estimates of
smokers and non-smokers. More generally, according to
Sherman, any difference between estimates of behavior
frequency for individuals who are under different
situational constraints (either by their own choice or
otherwise) can be identified as false consensus. Ross et
a1. and Sherman's definitions are conceptually identical
since both identify false consensus as relative
differences in frequency estimates between two groups
which differ on a single dimension. The differences
between these definitions are detailed below.

According to the first operational definition of false
consensus given above, false consensus is the deviation
from the actual base rate of decision making. Suppose
that a dichotomous decision is split exactly 50/50 among

experimental subjects. False consensus could be measured

9

as the relative deviation from this equality. False
consensus bias then, would be the difference between actual
(equal) and predicted estimates of consensus. This
difference could be measured independently for the two
groups - those who chose Option 1 and those who chose
Option 2. In this case both, the consensus estimates for
those who chose Option 1 and those who chose Option 2
would be tested for deviation from the midpoint.
Again, it should be noted that this first definition
reflects the intuitive concept of false consensus. That
is, the term false consensus implies an inflated
frequency estimate in the direction of one's own choice
or opinion relative to the actual base rate of that
choice. In contrast, the other operational definitions
ignore the base rate.

In fact, current data on false consensus do not
indicate that subjects' estimates of consensus differ
from the actual base rate of decision preference. For
instance, in Table l, the cumulative results across the
four scenarios of Ross et a1.'s (1977) research, show that for
individuals who chose the second option, their estimates
of consensus (49/51) are highly similar to the actual
58/42 split. Similarly, those who chose Option 1 also do
not overwhelmingly advocate the pervasiveness of their own
choice (66/34) relative to the actual split. It is not

clear then, that either estimate really differs from the

10
actual Split in the population decision. In fact, when actual
decision split is used as a covariate, the false consensus
effect (as conceptualized by Ross) was virtually
eliminated.

According to the original Ross et a1. (1977)
(definition 2 page 7) operational definition, false
consensus does not depend on the actual base rate of the
raters' decisions. For example, examination of Table 1
indicates that for the term paper scenario, the actual
split in rater's decisions was 80% - 20%. If false
consensus was a deviation from the base rate, the
majority of subjects in this condition would have had to
estimate consensus in the high 90 percent range to make
consensus statements which were "false." In fact, Ross
et a1.'s subjects who opted to "choose the individual
paper,” made an estimate of consensus (67.4%) which was
substantially below the actual decision split. For this case,
as well as several other instances to be found in Ross
et a1.'s work, subjects actually underestimate
consensus. Nonetheless, according to Ross et a1.'s
operationalization, there was a significant false
consensus effect, since individuals who chose Option 1
predicted that a greater number of others would chose
Option 1 than did subjects who had picked Option 2.

Given that Ross et a1. accept the idea that subjects

can underestimate actual consensus and still demonstrate

of.

11

false consensus, it appears that these researchers are
more interested in the phenomenological issue of false
consensus, than in the accuracy issue. Their measurement
of false consensus capitalizes on the slight bias toward
self by comparing groups of individuals whose consensus
estimates should differ in opposite directions if there
is a bias toward one's own beliefs. For instance, heavy
drinkers may be prone to overestimating the number of
their peers who also drink, whereas members of a
temperance union might underestimate the frequency of
drinkers, i.e., overestimate the number of non-drinkers.

What conclusions can, or should be, drawn from this
discussion of consensus? First, the term false
consensus is, in part, a misnomer, since subjects'
estimates of consensus may actually be lower, not higher
than the actual rate of consensus. Second, in empirical
work false consensus is not measured as a predicted deviation
from the base rate of decision preference. In fact,
Ross et a1. (1977) and Sherman et a1.'s (1983)
operational definitions ignore the actual decision
split. Finally, the false consensus effect, according to
Ross et al., is based on the combined effect of the
overestimate of one group, and underestimate of a second
group.

From a measurement point of view, both Sherman et

a1. (1983) and Ross et a1. (1977) use a single question

12
to assess estimates of consensus. After choosing one
of two alternatives, subjects are asked "What percent of
your peers would choose as you have chosen?" According
to Ross, a main effect for “choice," that is,
differences in estimates of consensus between those who
chose Option 1 and those who chose Option 2, indicates a
false consensus effect. If one were interested in
manipulating factors that affect estimates of consensus,
Ross et a1.'s definition is somewhat problematic. In
order to demonstrate the effect of some variable which
is purported to affect estimates of false consensus
(i.e., enhance the relative difference in frequency
estimates for those who chose Option 1 versus those who
chose Option 2) an interaction between "choice" and
that variable must be demonstrated.

Suppose, for example, that expected meeting (with 2
levels: meet or do not meet) was hypothesized to affect
false consensus. The interaction term, indicating a
false consensus effect at one level of meeting, and no
(or a different) effect of false consensus at a second
level meeting would demonstrate the effect of meeting on
false consensus. This example can be expanded to
another, higher level of complexity. Given a relatively
complex design where one expects consensus to be
depressed in some cells and raised in others, the Ross

et a1. measurement of consensus fails to allow an

13
assessment of the relative effect of the manipulations
between cells, i.e., one can assess the relative effect
of manipulations individually (via the interaction
between "choice" and manipulation A, and "choice and
manipulation B), but it does not allow easy comparisons
between the cells of these different conditions.

Sherman and his colleagues have suggested an
alternative which does permit such a cross condition
comparison of false consensus conditions. Take the case
where two conditions are compared to determine whether
one generated more consensus. Sherman et al. have
suggested that mean differences in the level of
consensus between conditions (with the estimates of both
those who chose Option 1 and those who chose Option 2
reflecting the percent of others who would make the same
choice as they had) serves as an adequate measure of
false consensus. Rather than requiring factorial
analysis to get at interaction effects, as would be
necessary using Ross et al.'s methods, Sherman's method
-of measuring consensus allows for an examination of
manipulations that potentially can affect estimates of
consensus through the use of planned comparisons. It is
Sherman's definition, that false consensus is the relative
degree to which individuals endorse the belief that most
others would tend to make the same decisions as

themselves, which is used in this dissertation to explore

14
the differential impact of manipulations on estimates of
consensus. In the following sections, several potential
explanations of the false consensus phenomenon are.
explored and relevant supporting and opposing empirical

work is discussed.

NONMOTIVATIONAL EXPLANATIONS OF FALSE CONSENSUS

Availability

 

One mechanism that is purported to affect the way
people make estimates of others' beliefs has come to be
known as the availability heuristic, a term coined by
Tversky and Kahneman (1982). In essence, the
availability heuristic refers to a postulated process in
which individuals access information that is stored in
either short- or long-term memory. Based on the
information in memory, estimates of frequency and
probability, they suggested, may depend on subjects'
assessment of availability to, or associative distance
from, information regarding some belief, behavior or
event. A primary postulate of this view is that
repetition enhances attitude availability. For
instance, Tversky and Kahneman noted that people are
better at recalling instances from large samples (i.e.,
multiple observations) than they are at recalling

instances from small samples.

15

Given that repetition enhances attitude
availability, common occurrences are more readily
imagined than are rare occurrences. For instance,
Tversky and Kahneman would suggest something similar to
the following example. Take the situation in which a
senior faculty person is asked to judge the likelihood
of promotion for an assistant faculty member. In
thinking about this issue the senior member may recall
prior instances of promotion for assistant professors.
To evaluate the likelihood of promotion, Tversky and
Kahneman suggest that the senior faculty person would
access the number of prior instances in which promotion
was awarded as well as assess the number of instances in
which promotion was not awarded. Tversky and Kahneman
suggest that if more promotion instances are recalled
(or more tenuously, if it seems easier to recall
instances of promotion), the senior faculty member would
predict a high probability of promotion for the present
candidate.

Three related aspects of this approach are
intuitively unsettling: first, it relies on simple
frequencies; second, it underutilizes the full content of
prior experience; and third, it fails to consider possible
effects of higher-order cognitive processes.

Thus, Tversky and Kahneman suggest that estimates of

promotion will not be based on the senior faculty

16
member's recollection of the quality of previous junior
faculty, nor will they be based on the quality of the
faculty member under review. Rather, in their view, such
estimates are based entirely on the relative numbers of
relevant past experience that favor one outcome over
another.

Kahneman and Tversky's view of the human as
information processor is based on the assumption that
the human mind is limited in its ability to process
information, and therefore, it must use efficient
methods of "data" reduction to think effectively.

They suggest that the use of availability information is
one such labor-saving method. Tversky and Kahneman, in
fact have provided several demonstrations that support
the notion that individuals use the availability
heuristic when making a variety of judgements. In a
series of ten studies, they examined subjects' ability
to estimate frequency and probability in the recall and
construction of target items. They hypothesized that the
judged frequency of classes is biased by the
availability of their instances for both perceived ease
of construction and recall.

In their first study, Tversky and Kahneman provided
subjects with eight problems in which they were given
seven seconds to estimate, from a sequence of nine

letters (e.g., XUZONLCJM) how many words they could

17
produce in a 2 minute period. Subjects estimated the
number of words prior to construction on eight problems
and they constructed words without prior estimation on
eight other problems. The estimation and construction
problems were alternated. Tversky and Kahneman found
that the correlation between estimation of the number of
words which could be produced and actual production
was 0.96, indicating a strong relationship between these
two very different behaviors. They take this relationship
between estimation and construction as an indication
that some abbreviated activity - the availability
heuristic - must exist to allow accurate prediction even
when the short amount of time allowed precluded subjects
from actually attempting constructions.

In the second study of this series, subjects were
asked to recall, rather than construct, instances from
one of two categories (e.g., flowers or Russian
novelists). Again, subjects were given seven seconds to
estimate the number of instances of a given category
they could list in a 2 minute period. The results
indicated that there was a very strong relationship
between estimation and actual recall, g a 0.93. One
explanation for these results is that people use the
number of instances constructed or recalled in a short
period to estimate the number of instances in a longer

(construction/recall) period. Or, as Tversky and Kahneman

18
suggested, subjects may only assess the ease of
availability without using retrieval or construction at
all. I

In an attempt to discuss variables that may
underlie the availability heuristic, Tversky and Kahneman
have suggested that availability may depend on the
strength of associative bonds between cognitions. They
assert that associative bonds are strengthened by
repetition. In essence, Tversky and Kahneman suggest
that the availability heuristic functions as the inverse
of the repetition/associative bond relationship, i.e.,
individuals' assessment of the strength of a bond is
taken as an indicant of exposure frequency. Therefore,
the easier the availability, the greater perceived
strength of an associative bond. But, speed of
availability is the operational definition of
associative bond strength. So how can we distinguish
between the two? The reasoning is tautological, a
measure of one is also the measure of the other.

Some recent research has provided mixed support for
the notion that availability is enhanced through
repetition. Powell & Fazio (1984) asked subjects through
a series of redundant items that were embedded within a
questionnaire to state their opinion, and repeat their
opinion zero, one, three or six times. They predicted

that repeated attitudinal expression would decrease

19
response latency for object-evaluation as measured by
reaction time. In fact, their results indicated a linear
trend showing decreased latency with increases in
attitudinal expression in an initial block of trials. In
a second trial block, however, no effect of repeated
attitudinal expression was demonstrated. Thus, Powell
and Fazio's work provides some support for the
availability hypothesis of Tversky and Kahneman (1982).
Completely consistent with Tversky and Kahneman's view,
repetition was found to enhance availability at least
with rather immediate recall. Relevant to false
consensus, Powell and Fazio's results also suggest that
subjects who have the opportunity to repeat their attitudes
should also make higher estimates of consensus than

subjects who do not repeat their attitudes.

Vividness

In addition to repetition, what is it that makes
some data more accessable than others? Nisbett and
Ross (1981) suggested that accessibility may be
dependent on the "Vividness criterion." Information
that is more likely to attract and hold our attention,
they noted, depends on the extent to which the data are
(a) emotionally interesting, (b) concrete and image
provoking, and (c) proximate in a sensory, temporal or

spatial way. Nisbett and Ross suggest that vivid

20

information is more likely to be stored and remembered
than pallid information. However, the examples these
authors use suggest that "Vividness" may be synonymous
with an additional stimulus or additional information.

Nisbett and Ross suggest that events that happen to us
are more interesting than events that happen to other
people. They suggest further that the emotional interest of
an event is influenced by the degree to which it affects
the participants needs, desires, motives and values. One
benefit of this is that it integrates both "motivational"
and "cognitive" explanations of availability. To say
that availability depends on the "hedonic relevance" (p.
46) of the event clearly calls for an integration of
these two potential influences.

One implication of Nisbett and Ross' ideas is
that vivid data provide more information than
pallid data. For instance, Nisbett and Ross present the
example of Jack: ”Jack skidded on an icy road and
demolished a parked car.” The authors suggest that
casual observers would think Jack was a bit unlucky.
However, Nisbett and Ross suggest that if an observer
was told that the car belonged to a close friend
or relative of theirs, or that it was their own car that was
hit, they would respond differently, e.g., Jack is a
menace and should be removed from the road. Moreover,

the latter scenario would be much more likely to be

21
recalled.

What is it that makes this latter scenario more
likely to be recalled? Perhaps hedonic relevance serves
as an attention-focusing device that triggers a greater
number of associations, and, as a consequence ensures a
greater likelihood of encoding more information. For
instance, a number of researchers (e.g., Hull and Levy, 1979;
Markus, 1977) have demonstrated that there are individual
differences in the information processing of
self-relevant information. Such a view requires
postulating some type of information processing step
that takes this perspective beyond those that rely
entirely on cold cognition for their explanatory power.

Even beyond the problem that Vividness implicitly
contains a motivational component, there is a further
complication in convincingly demonstrating that false
consensus is a manifestation of a purely cognitive
phenomenon. The most direct test of the attitude
accessibility hypothesis is through the use of a
reaction time measure. The faster the reaction time, the
more available the cognition. In terms of applying the
availability heuristic to the false consensus
phenomenon, reaction time is not an ideal measure of
false consensus. Although one could measure the time it
takes to make estimates regarding the number of peers

who might make the same decision as oneself, this

22
procedure would seem to be an imprecise measure. As a
consequence, some indirect method must be used to test
the postulated effect of availability. Specifically, one
must devise situations in which availability would be
manipulated without engaging a motivational component.

The use of reaction time as a measure of attitude
availability suggests a potential method for discerning
the relative influence of motivational and
nonmotivational explanations of false consensus. If one
were to take to the extreme the logic of the
nonmotivational perspective, by definition, those
individuals who demonstrated higher attitude
availability (have faster reaction times for a number of
different measures) would also demonstrate greater
amounts of false consensus. That is to say, those
individuals who had the most efficient processing
apparatus (in terms of hardware), and assuming
motivational differences are random, should demonstrate
the greatest degree of false consensus. Such a result
seems unlikely but would provide support for a cold
cognitive explanation.

In sum, speculations regarding the relative effect
of Vividness information and its purported impact on
attitude availability has not directly address the
false consensus phenomenon. No research has directly

attempted to examine the effect of Vividness on attitude

23
availability, so few conclusions can be drawn. In fact,
as in many other studies, Vividness and the availability
heuristic have been applied to false consensus research

only as a post hoc explanation of the results.

Selective Exposure

Ross and Anderson (1982) suggested that another
phenomenon related to accessibility could be the cause of
the false consensus effect. They reasoned that we know
and associate with people who are similar to us - people
with the same experiences, backgound, values and so
forth. Such people do, they suggested, respond as we would
in a number of varied circumstances. This view, in fact,
is consistent with work on attraction and liking which
suggests that we are attracted to and are friends with
those who have similar attitudes as ourselves (Byrne,
1971). This peer group may then serve as a biased sample
of people from which we may make erroneous judgements
regarding the more general population. In essence,
selective exposure to this group of people may affect
estimates of consensus. Ross and Anderson suggested that
selective exposure itself might affect cognitive
components such as the ability to recall or visualize
occurrences of certain behaviors. In this way, selective

exposure could affect the availability of information. As

24
a consequence, individuals are likely to be misled by
the ease or difficulty they encounter in estimating the
likelihood of certain occurrences.

Sherman, Pressin, Chassin, Corty, and Olshavsky
(1983) systematically examined the selective exposure
hypothesis. In this study, Sherman et al. examined
people's estimates of smoking prevalence as a means of
testing three predictions relevant to the selective
exposure hypothesis. First, they predicted that
estimates of smoking prevalence would be related to the
number of significant others who smoke (i.e., smokers
with whom subjects share their immediate social
environment). Second, false consensus would be stronger
in populations where friendship patterns are homogenous
relative to smoking (i.e., smokers socialize with
smokers; nonsmokers socialize with nonsmokers), and
conversely, the false consensus effect should be weaker
where friendship patterns are heterogenous relative to
smoking (i.e., smokers and nonsmokers intermingle).
Third, Sherman et al..predicted that if selective
exposure underlies the false consensus effect, estimates
of smoking would be relatively high for peer groups, but
lower for nonreference groups. For example, they suggested
that adolescents should show stronger false consensus
when estimating the smoking prevalence of teenagers than

when estimating the smoking prevalence of adults.

25

The selective exposure hypothesis received some
support in this study. Subjects were asked, out of
their 5 closest friends, how many smoked. The number of
close friends who smoked was then correlated with
smokers and nonsmokers estimates of smoking prevalence.
The results indicated a moderate relationship between
number of friends who smoke and estimates of smoking
prevalence 5'5 (.05 to .43). This relationship appeared
to be greater among nonsmokers (average 5 = .27) than
among smokers (average g = .17).

The prediction that subjects would demonstrate
larger false consensus effects when estimating the
smoking prevalence of their peer group than when
estimating for a nonreference group was also supported.
Adolescents made greater estimates of consensus for
their peers than for adults. Finally, the third
prediction was also supported - estimates of consensus
appeared to be greater among homogenous groups than
among heterogenous groups. Thus, there is some support
for the selective exposure hypothesis.

If selective exposure is to stand as a viable
"nonmotivational” explanation of the false consensus
effect the question must be asked, "How nonmotivational
are the phenomena connected to selective exposure --
that is, what causes them?" It is clear from a purely

nonmotivational viewpoint that the assessment of

26
information requires a narrowing of perspective to allow
"sense-making" to occur. However, it has been suggested
that selective exposure is hedonically based. Abelson,
Aronson, McGuire, Newcomb, Rosenberg, & Tannenbaum's
(1968) sourcebook on cognitive consistency addressed the
issue of selective exposure and suggested that exposure
was primarily affected by incentives to acquire personal
gains, social approval or anticipated self-approval.
Recent work (Frey & Rosch, 1984) has provided strong
evidence that selective exposure occurs when a decision
is decision is irreversable. Their results are
consistant with a dissonance theory explanation, and
inconsistent with a cold cognition explanation
because selective exposure, assuming it is a cold
cognition, should be constant, regardless of decision
reversability. In contrast, if selective exposure is
motivated, it could potentially prevent dissonance in
some instances -- those cases where the decision is
irreversable. From a motivational perspective, selective
exposure would be unnecessary in cases where the decision
was reversable.

Mark Snyder's (1984) ”hypothesis confirming" research
is suggestive of similar, motivational explanations of
selective exposure. In his research, Snyder presented
subjects with information from which they tended to

generate selective, hypothesis confirming behaviors.

27

For example, subjects who were told that they
would be meeting an extroverted person were much
more likely to ask questions like, "Have you been to any
good parties lately?" In contrast,vsubjects who anticipated
meeting an introverted person were likely to ask
questions like, "Have you read any good books lately?"
Such selective, hypothesis based questions led to a
confirmation of the subjects' expectations regarding
the degree of their partner's extroversion, irrespective
of the actual level of his or her extroversion. This
confirmatory mode of questioning is consistent with
the cognitive consistency model of selective exposure.

The inferences one can draw from this evidence
are as follows: Selective exposure can affect false
consensus. But, the claim that selective exposure is
"nonmotivational" seems erroneous. It could be argued
that selective exposure is the result of activating
a schema, person prototype, or stereotype. But the
question becomes what "triggers" a schema, and is it
triggered for a mechanical or motivational reason?
The preponderance of evidence suggests that selective
exposure is motivationally based, e.g., there is no
reason, from an information processing standpoint, why
individuals should seek consonant information over
information that is inconsistent with existing

cognitions. If anything, the opposite strategy might be

28
more adaptive. However, motivated selection does not
preclude the possibility that false consensus arises
from a two step process, in which subjects selectively
expose themselves to messages and the subsequent
availability of that information leads to false
consensus. From this perspective, the selective exposure
viewpoint fails to elucidate the relative contribution
of motivational and nonmotivational explanations of the

false consensus effect.

Fundamental Attribution Error

Ross et a1. (1977) suggested that false consensus
might arise from the divergent perspectives of actors
and observers. This hypothesis is drawn directly from
the work of Jones and Nisbett (1972), who speculated
that observers tended to view the causes and
implications of actors' behaviors differently from the
way that the actors themselves perceived their behavior.
Jones and Nisbett suggested that we tend to see others'
behavior in terms of broad, consistent personal
dispositions, whereas we tend to attribute our own
behavior as actors to situational variables. False
consensus, then, may reflect a generalization of one's
sense that his or her decision was situationally
determined. The logical extension of this argument

suggests that manipulating the locus of perceived cause

29
(situational versus dispositional) should affect
estimates of consensus.

Within this framework, Gilovich, Jennings and
Jennings (1983) hypothesized that the false consensus
effect, and the corresponding tendency to make strong
inferences about others who chose differently, would be
mitigated when subjects were asked to explain their
choices in terms of their own personal characteristics
(self-focus condition). Conversely, they also
hypothesized that subjects would show a greater false
consensus effect when asked to explain their choices in
terms of situational factors (situational-focus
condition). A third condition was formed in which no
focus of attention was stimulated (control condition).
Subjects were presented with questionnaires containing
descriptions of four hypothetical situations for which
there were two behavioral alternatives. Subjects were
asked to make their choice, explain why they made their
choice, and then estimate the number of students who
would choose each of the two alternatives. As expected,
subjects under the situational-focus condition exhibited
the greatest amount of false consensus, subjects in the
self-focus condition exhibited the least false
consensus, and subjects in a control condition
demonstrated a moderate amount of consensus.

In a second study Gilovich, Jennings and Jennings

30
(1983) had subjects rate the extent to which their
choices on a series of dilemmas would be determined by
situational or dispositional factors. From a group of
42 such dilemmas, 10 were selected which represented
a broad specturm, from a dilemma which produced choices
that were perceived as completely situationally
determined to a dilemma in which choices were seen as
completely dispositionally determined. This set was then
given to a second group who were also asked to make
consensus estimates regarding their decision on each
dilemma. The results indicated that subjects showed greater
false consensus for dilemmas which previously had been
rated as highly likely to generate situational
responses.

Gilovich et a1.'s results provide striking
confirmation of their hypotheses, and tend to suggest
that a cold cognitive reasoning process may affect
estimates of consensus. What is unclear, however, is
the link between the perception of situational pressures
and the generation of false consensus. Implicit to
Gilovitch et a1.'s model is a rational-deductive
process in which a person reasons ”I acted because of
situational pressures, so others would probably act the
same way." Or, if this is not the process that links
attributions with inflation of consensus estimates,

researchers need to explicate that link between

31
the two. In either case, a thorough model of how
attribution can inflate consensus estimates has not
yet been put foward.

In addition, it can be argued that the fundamental
attribution error could have a motivational basis. For
instance, Greenwald (1980) found that individuals are
more likely to take responsibility for their positive than
for their negative outcomes. Others, such as Berscheid,
Graziano, and Monson, and Dermer (1976) and Pittman and
Pittman (1980) have shown some support for the notion
that people strive to control their environment, and
that in essence, it is need for control that motivates
some attributional process.

Taken as a whole, the research investigating
nonmotivational variables which may affect estimates
of opinion projection (Availability, Vividness,
Selective Exposure, and Fundamental Attribution Error)
provide some interesting and plausible explanations for the
phemonenon. However, a number of considerations serve
to temper one's confidence in the efficacy of strictly
cognitive-based explanations of false consensus.

First, the availability hypothesis has not been
directly examined. It has only been offered as a
post hoc explanation of consensus inflation.

Second, the Vividness criterion, outlined by Nisbett

and Ross, has received no attention in empirical work.

32
Third, the selective exposure hypothesis, although providing
some support for the notion that false consensus is
the result of a cold cognitive process, has been linked
to cognitive consistency and motivational factors. And
finally, the attribution process also seems intimately
tied to the motivation for control and esteem. Thus
far, then, research suggests that a combination of
motivational and nonmotivational components affect
estimates of opinion projection. The next section
reviews evidence which tends to support a motivational

perspective of false consensus.
MOTIVATIONAL EXPLANATIONS FOR FALSE CONSENSUS

Motivational mechanisms, which long have been a
major focus of inquiry in psychology have been hypothesized
to be involved in the false consensus effect. Typically,
the motivational perspective has proposed that false
consensus is related to defensive projection, or
self-serving biases. Such a position suggests that people
hold opinions that enhance, justify or validate their
own behavioral choices. This motivational perspective is
closely allied with the social comparison literature
(Festinger, 1954), which asserts that individuals desire
to hold socially validated opinions. Such a view
suggests that individuals use cognitive bolstering

(Janis & Mann, 1971) or cognitive truth bending, to

33

enhance the perception of self/other consistency. Other
researchers who advocate the motivational perspective
make psychodynamic interpretations of projection which
have been applied to false consensus (cf. Holmes, 1968,
1978). Thus false consensus has been thought of as a
function of personal anxiety. In the same frame, Bramel
(1962, 1963) has suggested that opinion projection may
serve a dissonance reducing function.

Overall, the motivational perspective suggests that
the hedonic relevance of opinion projection has its
functional value in the reduction of stress and anxiety
(Sherwood, 1981; Bennett & Holmes, 1975). The following
section outlines empirical work which tends to support
the motivational explanation of false consensus.

Although each study appears to support different aspects

of the motivational interpretation (i.e., self-justification,
self-enhancement, bolstering social support, etc.), most
interpretations indicate that false consensus is generated
under conditions of personal or social anxiety. However,
some research examining the effect of hedonically-relevant
stimuli suggests that a more general interpretation might be
valid. In this more general motivation based view, false
consensus is seen as being generated by stimuli that

threaten rewards or potential outcomes.

34
Vested Interest

Vested interest is the term used by Sivacek and
Crano (1982) to describe the motivational state of an
individual. If the situation is hedonically relevant,
they suggested, the person has a vested interest in the
outcome of the situation. If a situation has little or
no hedonic relevance, the person is said to have little
vested interest in it. According to the logic
of vested interest, Crano speculated that if vested
interest were high, subjects would be likely to assume
consensus. If vested interest were low, there would be
little motivation to assume consensus.

In a study to examine the relative effect of
hedonic relevance on estimates of consensus, Crano
(1983) presented freshman, SOphmore, and senior college
students with information which indicated that the
administration was planning to add a tuition surcharge
that would affect underclassmen, upperclassmen
or all students. In each condition, subjects made
estimates (on a scale of 1 to 10) about how the plan would
affect them. He then asked subjects to indicate on five
bipolar adjective scales their own attitude toward the
proposed tuition change. Finally, he asked subjects
to predict the percent of peers who would feel the same

about the plan as they did. The results showed that

35

subjects who believed the plan would affect them
reported more negative attitudes regarding the plan,
and clearly assumed more consensus, than did subjects
who had less vested interest in the issue.

The demonstrated link between vested interest and
false consensus appears to reflect a general
wishful thinking phenomenon -- subjects want a desired
outcome, and they hope others would choose as they had.
However, a contrasting, yet plausible explanation
of Crano's results is that subjects who knew that the plan
would affect them were more vigilant, and attended more
to their own decision than did subjects in the non-vested
condition. The heightened attention of vested interest
subjects to their own decision made this cognition
highly available, and as a consequence, the subjects own
decision was used as a heuristic for predicting the
decisions of others. There is, however, no research that
examines whether subjects would have greater
self-focused attention under conditions of vested
interest than under conditions where they were not
invested.

If vested interest does not activate attention to
oneself, then the next important question regarding
false consensus is how general is the phenomenon?

Will any hedonically relevant stimuli facilitate false

consensus? Crano (1983) suggested that when an outcome

36
may be costly, individuals will, according to social
comparison theory (Festinger, 1954), want to validate
the opinions by seeking social support. From this
perspective, false consensus would be a natural
consequence of any decision or action which could be
potentially costly. Accordingly, Crano suggested that
vested interest is very general in its activation of
false consensus, and in fact, a number of different
stimuli have demonstrated a facilitation of consensus
estimates, including self-enhancement, self-justification,

need for social support, and threats to self-esteem.

Self-Enhancement

Results which tend to support a motivational
explanation of opinion projection have been generated in
research that has examined the effect of target's
physical attractiveness on estimates of similarity.
Rating the similarity of specific others, and the
relative degree to which a person projects their
attitudes onto an other has come to be known as ”assumed
similarity." Assumed similarity is like false consensus,
but rather than focusing on individual estimates of the
beliefs of peers in general, assumed similarity
concentrates on whether a person projects their opinions
onto a specific other.

In a study of assumed similarity, Marks, Miller and

37
Maruyama (1972) reasoned that individuals are often
judged by the company they keep. As a consequence, they
suggested, persons expressing a liking for, and a desire
to affiliate with physically attractive persons may
represent an indirect attempt to present oneself as
being similar to socially valued others. In order to
test this hypothesis, subjects were first asked to rate
themselves on five positive, five neutral and five
negative traits. Following self rating, each subject
viewed 12 facial photographs of college-aged females who
varied in physical attractiveness. Then, subjects were
asked to rate each photograph on the same series of
traits on which they had previously rated themselves.

As anticipated, the results indicated that subjects
assumed greater similarity between themselves and
attractive others than between themselves and
unattractive others. In addition, attractive targets
were desired more as friends than were less attractive
targets. The authors suggested two possible explanations
for the results. First, they suggested that there may be
a spurious relationship between ratings of self and
attractive others. Subjects may rate themselves as good
and independently rate "what is beautiful is good.”

A second, motivational explanation suggests that
individuals assume similarity between themselves and

attractive others as a means of enhancing their own

38
image. Such a speculation could be examined empirically
by using self-esteem as a moderator variable, since low
esteem subjects should rate themselves more similar to
neutral or unattractive others while still rating
attractive others high on socially desirable traits.

In a second study (Marks & Miller, 1982), undergraduates
were asked their opinions on eight issues (which varied
in the degree to which they were involving) and were
subsequently asked to estimate the opinions of 12
targets who varied in physical attractiveness.
Consistent with the self-enhancement hypothesis,
subjects projected their own responses onto attractive
targets to a greater extent than they projected onto
unattractive targets. There was no main effect for
involving and uninvolving issues. But, issue
involvement did interact with attractiveness such
that it appeared that subjects distanced themselves
from unattractive others on involving issues.

Marks and Miller's (1982) results are inconsistent
with the views of nonmotivational theorists in two
respects. First, if attitude accessibility influences
judgment and prediction, the amount of assumed
similarity should not differ between attractive and
unattractive targets. Most importantly, they suggest
that their data minimize the possibility of a

spurious relationship between ratings of self and

39

attractive others (see their discussion for a detailed
account). Second, according to Nisbett and
Ross (1980) Vividness of an opinion or an attitude
enhances accessibility. In the present case, one would
predict a main effect for issue involvement. Assumed
similarity should have been greater for involving (i.e.,
more vivid) issues than for uninvolving issues, but this
pattern was not found.

In a quite different study (Krahé, 1983), subjects
were presented with 26 different scenarios describing a
target person in an interpersonal situation. The
scenarios systematically manipulated demographic
similarity between the subject and the target. After
reading each scenario, subjects were asked to rate the
degree of similarity between themselves and the target.
The manipulation of demographic information had little
effect. But, the results indicated that subjects ratings
of assumed similarity was strongly affected by the
purported success or failure of the target. Subjects
saw themselves as very similar to targets who were
successful, but dissimilar to targets who were
Unsuccessful. Her results do not necessarily provide
strong support for a false consensus explanation,
since subjects attempts to associate themselves with
the successful may indicate a "I am like him"

self-serving response, which is different from ”He is

40

like me" which is more typical of false consensus.
Nonetheless, Krahe's findings seem to indicate more
support for a motivational explanation of assumed similarity.
Furthermore, it is unclear how differences in
information processing could create such an effect.

Taken together, these studies give some support to
the notion that individuals assume similarity between
themselves and a target when the target person has what
might be considered to be "good" qualities. Yet, the
potential problem of a spurious relationship between
ratings of self and attractive others has not been
adequately addressed. Thus, it can only be speculated that
the process of assuming similarity between self and
valued others may be motivated by a need for

self-enhancement, but further research is needed.

Self-Justification

Several studies indicate that false consensus may
occur in situations in which subjects feel a need for
self-justification. Dawes, McTavish, and Shaklee (1977)
speculated that attribution about others' behavior in a
mixed-motive game might be a consequence of one's own
behavior. They compared the attributions of actors who
participated in a commons dilemma to the attributions
of observers who did not participate in the dilemma.

They predicted that actors' attributions would

41
be more variable than those of observers, since actors'
attributions would be based on their own behavior. Dawes
et al. speculated that one of two mechanisms might
account for the egocentric inferences made by actors.
First, they suggested that actors may feel a need to
justify their cooperative or competitive choices
(competitors might want to reduce the self-perception of
being exploitive, whereas cooperators might need to
reassure themselves that they were not being naive).
Second, the use of one's own behavior might simply be
used as an inferential guide for predicting the behavior
of others. In essence, they suggested that the
availability of one's own choice might affect estimates
of consensus.

Building on the work of Dawes et a1. (1977) Messé'
and Sivacek (1979) used a one trial prisoner's dilemma
game to examine subjects' use of own choice in the
estimation of other players' behavior. Female college
students were asked to play one trial of a prisoner's
dilemma game and were subsequently asked to predict the
choices of both their partner and a nonpartner. Their
results indicated that (regardless of cooperative or
competitive choice) subjects assumed greater similarity.
between themselves and their partner than between
themselves and a nonpartner. In addition, Messé and

Sivacek found that subjects were more confident when

42
making judgments about a partner than when making
judgments about a nonpartner. Across conditions,
subjects tended to overestimate the extent to which
others would hold the same opinions they held (grand
mean approximately = 65%).

Overall, the results indicated a general false
consensus effect which was moderated by outcome
interdependence. Since subjects were not presented
with any other information which was likely to
distinguish between partners and nonpartners, it is
difficult to find a compelling explanation other than
self-justification for the subjects' consensus
estimates. However, it could be argued that the act of
pairing individuals produces an ingroup, outgroup bias
which encouraged subjects to make "like-me” estimates for
partners, and "unlike-me" estimates for nonpartners.

Both the ingroup bias and the self-justification
interpretations may be more clearly viewed as parts of
a more general social support explanation. Some
investigations of false consensus have contended that
overestimation of consensus may result from a need to
affiliate oneself with a support group. Whether this
is a need for affiliation, or related to a bolstering of
self-esteem is not clear. The following discussion

addresses the social support issue.

43

Social Support

 

Closely allied with research on self-justification,
the work of Sherman and his colleagues have found
results that are consistent with the notion that false
consensus is linked to a need for social support.
Sherman, Presson, Chassin, Corty, Olshavsky (1983) asked
Middle school students, High school students and adults
to estimate the number of boys, girls, men and women who
smoke cigarets. They hypothesized that false consensus
might be used to validate "deviant" behavior. They
suggested that the effect would be greater for young
adolescents, who would need to justify their deviant
behavior, than for members of older groups. In fact, in
a comparison of consensus estimates between smokers and
nonsmokers the results indicated no false consensus
effect for adults, a more apparent effect among high
school students, and a pronounced false consensus effect
among middle school students. In agreement with these
findings, in a comparison of a sample from the midwest
to one in the south, Sherman et a1.'s results indicated
a stronger smoking-false consensus link in the
midwest, where smoking was perceived as more deviant.

One note of caution is necessary before inferences
are drawn from Sherman et a1.'s work. They did not
provide measures of effect size, nor do their tables

indicate the relative cell size of middle versus high

44
school students. Their sample included some 5351
students and 122 adults. Of the total number of
students, 4343 were from a midwestern community.
It is possible that the null findings for adults could
be attributed to sampling error alone. Moreover, the
relative difference in the number of significant false
consensus findings for middle school versus high school
students and midwest sample versus southern sample may
also reflect differences in sampling error.

If, however, their interpretations are correct,
Sherman et al.'s (1983) results can be viewed as
supporting a social support explanation of false
consensus, e.g., subjects may have perceived a false
consensus of their smoking behavior as a means of
bolstering support and providing validation for their
prohibited activity. Similar results have been
demonstrated by Dan Eitzman (unpublished manuscript,
1984) who asked undergraduates to estimate the extent to
which their peers participated in several activities.
Students who were below the drinking age (but drank
regularly) tended to overestimate the number of their
peers who also drank. However, when asked to estimate
the number of peers who did or did not wear socks with
their shoes in the summer time (an activity which is not
deviant), no false consensus effect was found for sock

versus non-sock wearers.

45

A third study generated less equivocal support to a
social support/self-justification explanation of the
false consensus effect. Sherman, Presson, and Chassin
(1984) asked subjects to correctly identify a "real"
suicide note from a set of two or three possible notes.
After subjects picked a note they thought was correct,
they were told (through random assignment) that they had
chosen correctly (success condition), they had chosen
incorrectly (failure condition), or they were not told
about the correctness of their decision (No feedback
condition). Subjects were then asked to estimate the
number of others who would have chosen as they had.

In order to assess the effect of own choice
and case information on estimates of consensus, subjects
were asked to make estimates of consensus after learning
information about how another person picked and the
success or failure of that individual in identifying the
correct suicide note. The consensus estimates of these
two groups (subjects estimated consensus for subjects
who either learned of another person's success or who
learned of another person's failure) was then compared
to subjects estimates for self-pick under conditions of
success or failure.

Sherman et a1. (1984) examined subjects consensus
estimates within the context of a 2 (success/failure) x

2 (self-pick/other-pick) analysis of variance. The

46
results indicated no difference in consensus estimates
between self-and other-pick for subjects in the success
condition. However, there was a significant effect for
self-pick in the failure condition. As predicted,
subjects in the failure condition who had picked the
note themselves tended to estimate that more people
would choose as they had (M = 57.4%) than subjects
who were told that another subject had chosen
incorrectly (M = 42.2%). Sherman et al. in a second,
similar study replicated these results.

The results of the Sherman et al. (1984) studies
suggest a distinctly motivational interpretation.
Consistent with an attributive projection perspective,
subjects who had performed negatively overestimated the
extent to which others would perform as they had. The
failure of subjects in the self-pick/success condition
to demonstrate a higher degree of consensus than
subjects in the other-pick/success condition suggests
that when subjects are secure in their decision, no
bolstering of opinion needs to occur. Sherman et a1.'s
results are also consistent with the vested interest
explanation of Crano -- significantly higher estimates of
consensus occur in conditions where one's outcomes are
potentially negative.

This latter series of studies, which addressed

 

47
motivational components that potentially fascilitate the
opinion projection, provide evidence that at least
in some situations. opinion projection is in part an affect
laden, "hot" cognitive phenomenon, one that is not
readily explained by simple attitude availability or
another "cold" cognitive process. One further piece of
support for a motivational interpretation of opinion
projection is drawn from the literature on assumed
similarity. Research in this area has shown that
expected meeting leads to heightened levels of assumed
similarity -- as is discussed in the next section.

Expected Meeting

Miller and Marks (1982) set out to demonstrate
support for a motivational explanation of assumed
similarity. They noted that previous research has shown
that social interaction facilitates reciprocal liking
(e.g., Arkin & Burger, 1980) among participants, and
that others (e.g., Freedman, Carlsmith, Sears, 1974)
have noted that people who expect to meet and interact
with someone tend to exaggerate the other person's
positive traits and ignore or down play the target's
negative traits. People do this, Freedman et al.
speculated, to convince themselves that their
interaction will be pleasant and harmonious. Based on

this view, Miller and Marks (1982) hypothesized that

48
persons tend to project their opinion onto another
person when they expect to interact with him or her.

The purpose of Miller and Marks' study was to
examine the effect of a specific type of social
interaction with an other upon the degree of similarity
assumed to exist between self and that other. College
undergraduates were examined in dyads in which the other
person was a confederate who posed as a naive subject.
Potential subjects were approached in a lounge area in a
university building and asked if they would be willing
to devote a few minutes to answer some questions.

Each volunteer first answered a fewquestions

regarding campus issues, and was then provided with a
compelling description of a criminal trial which dealt
with a robbery and murder. The subject was asked to
place him/herself in the position of an actual juror and
render a verdict of either guilty or not guilty.

Half of the subjects were led to believe that they
would be meeting another subject (the confederate), who
was seated a few feet away from them, to discuss the
court case and come to a single decision regarding the
guilt of innocence of the defendant. The other subjects
did not expect to interact with the other subject.

In each case, subjects were asked to guess the opinion
of the other subject and how likely he or she was to

vote guilty.

49

The results indicated that subjects assumed a
significantly greater amount of belief similarity between
themselves and an other when a meeting was expected than
when no meeting was expected. Miller and Marks
(1982) conclude that there is a functional value to
projection of one's attitudes. They assert that
exaggerating the similarity between one's own view and
that of an other helps to reassure oneself that an
impending interaction will be harmonious. Miller and
Marks (1982) also suggest that the greater the assumed
similarity between self and other may increase one's
perception of social support for one's position and
serve to bolster one's certainty of opinion

correctness .

OVERVIEW OF THE PRESENT STUDY

Miller and Marks' (1982) paper provides relatively
strong evidence that assumed similarity is driven by
a motivational component. One question of interest is
whether estimates of consensus would also be heightened
by expected meeting. Thus, the present paper takes
another look at the issues of expected meeting and
perceived social support in an attempt to identify more
conclusively circumstances which enhance or depress
false consensus and assumed similarity.

There appear to be a number of similarities between

50
assumed similarity and false consensus. For example,
Marks (1983) has shown that there is a relatively high
degree of relationship between false consensus and
assumed similarity, g = .45. Of primary interest in the
present work both assumed similarity and false consensus
appear to be affected by motivational factors. It is
hypothesized that false consensus, like assumed
similarity, at least in part, is driven by anxiety about
self/other consistency needs and the need for social
support, both of which are hedonically relevant
concerns. For example, it is reasonable to expect that
persons who feel uncertain about what others believe --
e.g., when they expect to encounter an other whose
relevant beliefs are unknown-- will experience anxiety,
and as a consequence will use false consensus as a means
of reducing this aversive state.

The present study attempted to replicate and extend
the work of Miller and Marks (1982) and Sherman et a1.
(1984) by manipulating (1) expected meeting, (2)
information about other subjects' decisions, and (3)
knowledge about which kind of partner (agreeing or
disagreeing) the subject expects to meet. Table 2
presents a summary of the design and should be referred
to for clarification of the hypotheses and planned
comparisons. In the following discussion, the letters

A through F represent the particular conditions as they

 

51
Table 2

Basic Experimental Design for the Present Study

 

 

 

No Knowledge Knowledge that one person agrees
about others and one person disagrees
Meet Meet Meet
Either Agreeing Disagreeing
Person Other Other
+ A++ C + D+ E +-
Expected + ++ + + +
+ ++ + + +
Meeting + ++ + + +
+ ++ + + +
+ B++ F+
No Meeting + ++ +
+ ++ +
Expected + ++ +
+ ++ +

 

 

are labeled in Table 2. They are included to facilitate
the discussion of the hypotheses.

A number of hypotheses can be derived regarding the
the role of expected meeting and knowledge about others.
Hypothesis I: It was hypothesized that the present study
would replicate the work of Miller and Marks (1982) -
when subjects have no knowledge of others' decisions,
expected meeting will enhance assumed similarity.
Moreover, it was hypothesized that the same pattern of
results would be found for the false consensus measure,
in which subjects would be asked what percent of their

peers would choose as they had.

52

Since subjects do not have the opportunity to
selectively expose themselves to information, nor have
information which would stimulate different
attributional processes between conditions (A, B),
nonmotivational theorists would predict no differences
between the two groups. In contrast, motivational theorists
would predict greater consensus in the expected meeting
group. In addition, it was predicted that under the
knowledge conditions (C, F) a similar pattern of results
would appear (predicted comparisons A > B, C > F for

both assumed similarity and false consensus).

Hypothesis II: The critical planned comparisons involve
conditions in which there is an expected meeting and
subjects have knowledge about the other subjects'
decisions (Cells: C, D, E). If false consensus is
facilitated by anxiety concerning self-other
consistency needs, it is hypothesized that subjects who
know that they will meet with an agreeing other will
feel less anxiety about their expected meeting, since
they anticipate consensus, and their need to assume
consensus will be considerably attenuated. Conversely,
subjects who anticipate an interaction with a
disagreeing other would be expected to have increased
anxiety, and thus their need to assume consensus as a
means of reducing that anxiety would be heightened. This

hypothesis is completely consistent with the work of

53
Sherman et al. (1984), who demonstrated that under
conditions of self-failure, subjects overestimated
consensus relative to subjects' estimates of consensus
for others who experienced similar failure. In summary
then, it is hypothesized that subjects who expect to
meet with an agreeing other will exhibit less false
consensus than subjects who expect to meet a disagreeing
other or subjects who do not know which other (agreeing
or disagreeing) they will meet (planned comparisons, C >

D, E > D).

Hypothesis III: Expected meeting with a disagreeing
other will promote false consensus relative to subjects
who have knowledge about others' decisions but do not expect
to meet someone (planned comparison, C > F). This
prediction is relatively straightfoward. In contrast,
there can be less confidence about the extent to which
expected interaction with an antagonistic other will
affect ratings of false consensus over and above the
expected meeting condition in which subjects have
knowledge about others' decisions but are ignorant about
whom they shall actually meet. There is some
uncertainty as to the amount of anxiety the two
conditions will generate. However, a reasonable
assumption is that meeting with a disagreeing other will

generate more anxiety and, thus, generate more false

54
consensus than will the conditions in which subjects
have knowledge about others' decisions, but have no
information regarding which subject they are expected to
meet (planned comparison, E > C).

Comparisons between the "no knowledge" (A, B) and
the "knowledge about others" (C, D, E, F) groups is more
problematic since the latter group receives information
about others. A comparison between cells A-C and B-F
can give an indication of the effect of knowledge on
estimates of consensus and assumed similarity. In
contrast to previous studies that gave subjects
information about the distribution of opinions or
actions for the population and subsequently asked them
to make predictions about individuals (see Tversky and
Kahneman, 1982), the present study gives subjects
specific information about individuals and asks them to

make predictions about the population.

METHOD

Subjects

Sixty-five male and seventy-two female undergraduate
students at Michigan State University participated in
the study for extra credit in an introductory psychology

course 0

55

Design

The design can be conceptualized as a 2(expected
meeting/no expected meeting) x 2(knowledge of others
opinions/no knowledge) factorial that is augmented by
two comparison conditions. In both comparison
conditions, subjects expect to meet with one of two
people to discuss their decisions. In one condition they
are told they will be meeting with someone who agrees
with them, whereas in the other condition they are led
to expect that the person disagrees with them. knowledge
about two other subjects was presented so as to directly
assess the effect of expected meeting with an agreeing other,
a disagreeing other, or an other whose specific beliefs are
unknown. Sex of subject and type of decision scenario
(merit raise or reprimand) were also included in the

design as control variables.

Materials

Decision Task. The decision task with which
subjects were confronted consisted of a supervisory
evaluation of policemen (see Appendix B). An ”Evaluation
of Patrol Officer Performance" sheet described the
criteria for evaluation and emphasized the necessity of
careful consideration in making supervisory decisions.

The materials asked the subject to assume the role of an

56
officer review board member who is evaluating officers
for either (1) a merit raise or (2) a reprimand.

For each of the review decisions, 3 short
description of the officer under review is presented in
an "officer review brief." These briefs describe the
officer in a manner that makes the supervisory decision
ambiguous. For example, pretesting of these materials
indicated that for the merit raise decision, 56% decided
to award, and 44% decided to withhold the raise (N=23).
For the reprimand scenario, 55% of the subjects decided
to issue, and 45 % decided to withhold the reprimand
(N=9).

Measures. The primary dependent variables
presented in a brief questionnaire (see Appendix B) were
measured via the false consensus and assumed similarity
questions typically asked in previous research (Ross et
a1. 1977; Marks, 1983). For example, subjects were asked
what percent of their peers would reach the same
decision as they had -- this constituted the traditional
false consensus measure. In the results, false consensus
is discussed in two ways. First, a global measure of
false consensus was used by comparing frequency
estimates of others choosing Option 1 for those who
chose Option 1 versus those who chose Option 2. This is

similar to the method used by Ross et a1. (1977). In

57
addition, comparisons between cells used the estimates
of frequency of own choice, i.e., the mean consensus
estimates of own choice averaged across those who chose
Option 1 and Option 2 to produce a mean cell estimate of
consensus (cf. Sherman et al., 1984).

For the measure of assumed similarity subjects were
asked to indicate whether or not specific target persons
(best same sexed friend, favorite teacher, person most
disliked, etc.) would make the same decision. The present
set of target persons differed somewhat from those used
by Marks (1983). In his study, 12 targets were
presented, including both same and opposite sexed
persons most liked, and those most disliked, an.
acquaintance, an admired actress or actor, favorite
professor, favorite family member, and a family member
that is not close. Based on Marks (1983), similar
responses were scored 1, and dissimilar responses were
scored 0. The total number of others the subject
indicated would respond similarly served as a measure of
assumed similarity (see Marks, 1983). Previous work
using this type of measure has not reported the
reliability of the scale. However, the reliability of
the present scale was poor (standardized alpha .45).

Subjects were also asked about their sex, their
confidence regarding their decision, and how confident

they thought others would be about their decision.

58
In the expected meeting conditions, subjects were asked
whether they thought they would be meeting someone who
had made the same decision as themselves or someone who
had chosen differently. Also in the expected meeting
conditions, subjects were asked if they thought they
would be meeting with a male, a female, or if they
had an equal chance of meeting a male or a female.
Finally, subjects were asked to indicate how they felt
about meeting with an other (or how they might feel for
those in the no meeting conditions) by indicating their
attitudes on seven bipolar adjectives presented at the
end of the session. Three of these adjectives
(comfortable, relaxed, and anxious) were included to
provide a measure of general comfort with meeting

another subject.
Procedure

Subjects were run in groups of up to six
individuals at a time. They were asked to report to
separate locations so they did not see each other.

On their arrival at the research site, the subjects
were placed in cubicles, where they were introduced to
the general procedures of the study. After being
provided with preliminary information, subjects were

asked to wear headphones (as a precaution to prevent

59
them from hearing the voices of other subjects).
Subjects also were presented with a general information
sheet titled "Supervisory Decision Making," which
reviewed the process in which they were about to
participate. At this point, the subject was asked to
read, sign and date a Psychology Department consent
form. After subjects consented to participate in the
study, they were asked to read one of the
scenarios, described above, and come to a decision
concerning an action they would take if they were in the
position of a supervisory review board member.

Following the decision phase, subjects were'
systematically given information concerning (1) the
decisions of the two subjects in other cubicles, and (2)
information regarding their expected meeting (see
Appendix B for details). Some subjects, were not given
any information about the participants in other cubicles
(cells A, B, see Table 2; these cells replicate the
research of Miller and Marks, 1982). Other subjects
received information that one of the other subjects
agreed with them and one subject disagreed with them,
and they were led to believe that they would meet (a) an
agreeing other (cell D), (b) a disagreeing other (cell
E), (c) a person who either agreed or disagreed with

them (cell C), or (d) they did not expect a meeting, but

60
they knew one person agreed and one person disagreed

with them (cell F).

RESULTS

Manipulation Checks

 

A manipulation check was included to determine if
subjects in the expected meeting conditions (meet
agreeing or disagreeing other) were cognizant of their
partner's decision. Subjects in the expected meeting
condition were asked how they thought their partner
decided. In the condition where subjects expected to
meet an agreeing other, all 23 subjects said they
expected to meet with an agreeing other. Similarly, in
the condition where subjects expected to meet with a
disagreeing other, all 23 subjects endorsed the item
indicating that they anticipated meeting a disagreeing
other. These results indicate that the manipulation of
information about the partner's decision was successful.

It is possible that expecting to meet with a same
sexed other might facilitate false consensus. Although
subjects did not see the person he or she expected to
meet, a check was used to determine if subjects
perceived an equal likelihood of meeting with a male or
female partner. When presented with the question ”Do you

have a guess whether the person you will meet is either

61
male or female?" A majority of the 90 subjects in the
expected meeting conditions endorsed the item indicating
there was an equal chance that their partner was male or
female (n = 54), whereas less than half that number
believed that their partner was male (n = 23) and even
fewer believed that their partner was female (n = 13). A
Chi-Square analysis indicated that the discrepancy of
endorsement was greater than that expected by chance
alone X2(2) = 30.43, p < .01. Table 3 presents
the frequencies with which subjects indicated they
expected to meet a male or female partner as a function
of subject sex. When broken down by condition, there
were no differences between conditions in their
endorsement of this item, 32(6) = 5.04, p > .5.
The results suggest that in general subjects had
few expectations regarding the sex of their partner
-- though males tended to expect a meeting with a male
to a greater extent than females tended to expect a
meeting with a female.

A second measure regarding sex of the partner
asked subjects in the expected meeting conditions if
they had thought about the sex of their partner when
making their decision regarding the proportion of other
undergraduates who would make the same decision they had made.

Seventy-two subjects responded to this question. About

two thirds of the subjects (n = 50) indicated that they

62
did not have a guess about their partner's sex, whereas
less than a third (n = 22) indicated that they thought
about the sex of their partner at this time x2(1) =
10.65, p < .01 (18 subjects did not answer this
question). There were no differences between conditions

2
in the endorsement of this item, X (3) = 2.89, p > .4.

Table 3

Frequency Count Indicating the Number of Male
and Female Subjects Who Indicated that Their Partner Was
More Likely To Be Male, Female, or had an Equal Chance
of Being Either Male or Female

Likely Sex of Partner

 

 

Subject Male Female Equal Chance
Sex + +
+ +
Male + 16 4 23 +
+ +
+ +
+ +
+ +
Female + 7 9 31 +
+ +
+ +
+ +

 

63

Relationship Between Consensus Estimates and Assumed Similarity

A Pearsons correlation was calculated to examine
the relationship between assumed similarity and false
consensus. For the entire sample, there was a moderate
relationship between estimates of consensus and assumed
similarity, 5 (137) = .16, p < .03. However, when
separated into knowledge and no knowledge groups, it
became apparent that knowledge about others affected the
relationship between the two variables. In the No
knowledge condition, the relationship between estimates
of consensus and assumed similarity was moderately
strong, 5 (45) = .37, p < .006, whereas the knowledge
conditions evidenced a weaker correlation, 5 (92) = .07,
p < .3. Thus, the anticipated high degree of association
between assumed similarity and false consensus across
all conditions was not confirmed. Because-of these
unexpected results, a separate correlation was computed
for each cell. The resulting values, which are
presented in Table 4, reveal a strong relationship between
assumed similarity and false consensus in two cells (A,
D).

As indicated previously (p. 6), the measurement
of false consensus can be defined in several
ways. Using the operational definition of false
consensus outlined by Ross et al. (1977), there was a

significant false consensus effect for all six

64
conditions, i.e., comparing the estimates of consensus
for those who chose option 1 versus those who chose
option 2. The main effect for choice across conditions
was substantial, F (1,125) = 82.08, p < .001. For the
relatively more complex task of discerning the effect
of expected meeting and knowledge about others, a
procedure similar to that used by Sherman et a1. (1984)
was used. In their procedure, Sherman and his
colleagues used the combined estimates (across decision)
of consensus and used planned comparisons to test their

hypotheses.

Table 4

Correlations Between Assumed Similarity and
False Consensus Broken Down by Cell

 

 

No Knowledge Knowledge that one person agrees
about others and one person disagrees
Meet Meet Meet
Either Agreeing Disagreeing
Person Other Other
+ A++ 6 + 5+ ’E_:
Expected + ++ + + +
+ .56 ++ -.13 + .08 + .07 +
Meeting + (22) ++ (23) + (23) + (23) +
+ ++ + + +
+ B++ —___—_—F+
No Meeting + ++ +
+ .13 ++ .42 +
Expected + (23) ++ (23) +
+ ++ +

 

 

Note: Correlations greater than .40 are significant at
p < .025. Cell sizes are presented in parentheses.

65

Planned Comparisons

A series of planned comparisons were used to
test the experimental hypotheses. First, it was
hypothesized that, similar to Miller and Marks (1982),
expected meeting would facilitate the tendency to assume
similarity and to make false consensus judgements.
Miller and Marks used a measure of assumed similarity in
their study, and a similar pattern of results was
expected for both the false consensus and the assumed

similarity measures.

Basic and Conceptual Replication of Miller & Marks (1982).
Assumed Similarity. Assumed similarity was measured
by asking respondents to indicate how they thought
specific others, e.g., best friend, favorite teacher,
the person most disliked, etc. would decide if provided
with the same information with which they were provided.
The number of times a subject indicated that a target
person would choose as they had chosen served as the
measure of assumed similarity. The scale of the assumed
similarity measure has a range of 0 to 5. As
hypothesized, a comparison of the expected meeting/No
knowledge group (A) and the no expected meeting/No
knowledge group (B) replicated the assumed similarity
effect demonstrated by Miller and Marks. Table 5
presents the relevant means. expected meeting (A)

facilitated assumed similarity relative to the no

67
expected meeting (B) group t (131) = 1.69, p < .05. The
parallel comparison for subjects who had knowledge about
others and expected a meeting (C) did not differ from
subjects who had knowledge about others but did not
expect a meeting (F) t (131) = .32, p > .75.

Estimates of Consensus. Consensus was measured by
asking subjects what percent of undergraduates would

make the same decision they had made in the supervisory

task if they had been provided with the same information.

Table 5
Mean Assumed Similarity as a Function of Expected
Meeting and Knowledge About Others

No Knowledge Knowledge that one person agrees

about others and one person disagrees

 

 

 

 

Meet Meet Meet
Either Agreeing Disagreeing
Person Other Other
4» A++ c": + 415+ E-+
Expected + ++ + + +
+ 3.82 ++ 3.65 + 3.52 + 3.48 +
Meeting + (.85) ++ (1.03)+ (1.2) + (.85) +
+ ++ + + +
+ B++ F+
No Meeting + ++ +
+ 3.35 ++ 3.74 +
Expected + (.83) ++ (.75) +
+ ++ +
Note: Higher scores indicate greater assumed similarity.

 

Numbers in parentheses are standard deviations.

68

Thus, the scale of the false consensus measure had a potential
range of 0 to 100. According to the hypotheses, subjects
in the expected meeting condition should have reported
greater estimates of consensus than subjects in the no
expected meeting condition. Although the false consensus
measure did show the same pattern of results as the
assumed similarity measure, the difference between
expected meeting/no knowledge (A) did not differ from
the no expected meeting/no knowledge group (B) t (131) =
.84, p > .40. (Table 6 presents the relevant means.)
Similarly, in the conditions in which subjects were told
that one person agreed with them and one person
disagreed with them, there were no differences between
the expected meeting condition (C) and the no expected
meeting condition (F) t (131) = 1.74, p > .09,
two-tailed. In fact, if anything,the observed trend (C <
F) of the false consensus measure was opposite the
hypothesized relationship (C > F).
Focal Comparisons

It was hypothesized that subjects who expected to meet
with an agreeing other would experience less anxiety
and therefore, assume less consensus than a subject who

did not know whom they would meet, or who expected to

69
Table 6

Mean False Consensus Estimates as a Function of Expected
Meeting and Knowledge About Others

 

 

 

No Knowledge Knowledge that one person agrees
about others and one person disagrees
Meet Meet Meet
Either Agreeing Disagreeing
Person Other Other
+ A++ C + D+ E +
Expected + ++ + + +
+ 66.27 ++ 57.83 + 65.44 + 63.30 +
Meeting + (18.14) ++ (20.27)+ (9.76) + (14.97)+
+ ++ + + +
+ B++ F+
No Meeting + ++ +
+ 62.22 ++ 66.22 +
Expected + (16.04) ++ (16.32)+
+ ++ +

 

Note: Higher scores indicate greater false consensus.
Numbers in parentheses are standard deviations.

meet with another subject who did not agree with them on
the supervisor decision (Hypothesis II). The planned
comparisons (C, E > D) examined consensus and assumed
similarity for subjects in the knowledge conditions
where C) they do not know who they will meet, D) they
expect to meet an agreeing other, and E) they expect to
meet a disagreeing other. The results indicated that for
the consensus measure there was no difference between
meeting an unknown other (C) and meeting an agreeing
other (D), t (131) = -1.58, p > .11. One Hypothesis

(III) predicted a difference between meeting an agreeing

70
or a disagreeing other (E > D). However, there was also
no difference between meeting an agreeing (D) or meeting
a disagreeing (E) other, t (131) =.44, p > .65.
The similarity measure, yielded an equivalent
pattern of results; for the C -D comparison,

3 (131)

.48, p > .6, and for the E - D comparison,

t (131)

.16, p > .8. These results indicate that
relative to subjects who expect to meet with an agreeing
other, neither expected meeting with a disagreeing other
nor expected meeting with a person whose decision is
unknown generates more consensus or greater similarity
judgements than expected meeting with an agreeing other.
Also contrary to predictions, expected meeting with
a disagreeing other (E) did not facilitate false consensus
relative to the appropriate no expected meeting group (F)
t (131) = .61, p > .5. Nor was there a difference between
these two groups on the assumed similarity measure.
In fact, expected meeting with a disagreeing other generated
slightly less assumed similarity than the No Meeting

group 5 (131) = .95, p > .3.

Additional Tests of the BolsteringHypothesis
Examination of the two conditions in which subjects

expected to meet but did not know which subjects they

would be meeting (Cells A, C) indicated that subjects

tended to guess that they would be meeting someone who

71
had made the same decision as they had (N = 29), while
fewer subjects guessed that others decided differently
from themselves (N = 16) -- although this difference was
only marginally significant, X2= 3.75, p < .07.
Examining the estimates of consensus for subjects who
expected to meet an agreeing versus a disagreeing other
(combined across conditions A, C) indicated that
subjects who expected to meet an agreeing other (M =
69.07) had a higher estimate of consensus than subjects
who thought that they would meet with a disagreeing
other (M = 49.06), E (l, 41) = 13.53, p = .001. These
findings indicate that among subjects in conditions where
the information regarding the other's decision was not
manipulated, those who expected to meet a disagreeing
other did not attempt to bolster their decision by
assuming a high degree of consensus. This finding is
consistent with the comparison between expected meeting with
an agreeing (D) versus a disagreeing (E) other presented

above.

Effect of Information About Others' Decisions

Planned comparisons were performed on the assumed
similarity measure and the false consensus measure to
examine the effects of knowledge about others'

judgements. It could be argued that information

72
regarding others' decisions (e.g., that one person
agrees and one person disagrees with the subject)
would tend to depress false consensus and assumed
similarity. For the assumed similarity measure, a
comparison of the expected meeting/no knowledge (A)
indicated that subjects did not assume more similarity
in the knolwedge condition (C), t (131) = .60,
p > .50. Similarly, in the no expected meeting
conditions, the no knowledge condition (B) did
not differ from the knowledge condition (F)
t (131) = 1.42, p > .15.

For subjects in the expected meeting condition,
knowledge depressed false consensus (C) relative to the
no knowledge group (A) t (131) = 1.74, p < .042. But,
contrary to predictions subjects in the no expected meeting
knowledge group (F) had a somewhat higher consensus
estimate than subjects in the no knowledge condition (B)
t (131) = *1.75, p < .09. Note that the estimate of two
thirds agreement by the knowledge conditions is what should
be expected by a rational incorporation of the
information that two people hold one opinion and one

person holds a different Opinion.

Confidence About Decision and False Consensus
Marks (1983) previously demonstrated a link
between opinion certainty and false consensus. The

correlation between opinion certainty and false

73
consensus was also examined in the present study.
Immediately following their supervisor decision, subjects
were asked to indicate on a five point scale from very
uncertain to very certain how positive they were that
they had made the best possible decision. Contrary to
Marks' predictions there was not a significant
relationship between confidence and false consensus
r(136) = .13, p > .065. There was no restriction in
range in either the certainty measure or the consensus
measure to account for the difference in correlations
between Marks' (1983) work and the present study.
In contrast, when subjects estimated the degree of
confidence others would have about their decisions,
confidence correlated with their estimates of consensus
with 5's =.39, .47, and .39 for (p's < .01) entire
sample, no knowledge, and knowledge conditions

respectively.

Anxiety

A series of seven bipolar adjectives were presented to
subjects following the false consensus and assumed
similarity questions which asked them to indicate how
they felt about meeting with another subject (in the no
meeting conditions they were asked to indicate how they
would feel if they had expected to meet with someone in
the present study). Of the seven adjectives, three

related to how comfortable the subjects felt about

74
meeting with another subject (these included
comfortable, relaxed, and anxious.) The "anxious" item
was recoded to reflect relaxation or comfort. The three
items were summed to create a composite measure of
comfort (Standardized alpha = .63). The possible scale
(values could range from 3 to 15, where 15 indicated that
the subject felt very comfortable about meeting with
another person. Table 7 presents the relevant means.

Table 7

Mean Level of Comfort as a Function of Expected
Meeting and Knowledge About Others

 

 

 

 

No Knowledge Knowledge that one person agrees
about others and one person disagrees
Meet Meet Meet
Either Agreeing Disagreeing
Person Other Other
+ A++ c + 0+ E +
Expected + ++ + + +
+ 7.18 ++ 7.43 + 6.74 + 7.13 +
Meeting + (3.46)++ (2.64)+ (2.92)+ (1.35)+
+ ++ + + +
+ B++ F+
No Meeting + ++ +
+ 8.22 ++ 6.67 +
Expected + (l.80)++ (1.44)+
+ ++ +

 

 

Note: Higher scores indicate greater comfort with
meeting others. Subjects in the no expected mmeting
conditions were asked to estimate how anxious they would
feel if asked to meet with another subject. Numbers in
parentheses are standard deviations.

75
A 2(subject sex) x 6(condition) X 2(reprimand or
merit raise scenario) ANOVA of the anxiety measure was
also performed. Overall, subjects felt moderately
comfortable, Grand Mean = 7.50. There were no significant
main effects or interactions. Table 8 presents the relevant
means.
Table 8
Mean Level of Comfort with Meeting Others for Male

and Female Subjects as a Function of Expected
Meeting and Knowledge About Others

 

 

 

 

 

No Knowledge Knowledge that one person agrees
about others and one person disagrees
Meet Meet Meet
Either Agreeing Disagreeing
Person, Other Other
Expected + A++ C + 5+ E_T
Meeting + ++ + + +
Male + 5.82 ++ 8.27 + 7.50 + 7.45 +
+ ++ + + 4-
Female + 8.55 ++ 6.67 + 6.15 + 6.83 +
+ ++ + + +
No Expected + B++ F+
Meeting + ++ +
Male + 7.82 ++ 6.73 +
+ ++ 4-
Female + 8.58 ++ 6.42 +
+ ++ +
DISCUSSION

The goal of the present research was threefold.
First, it attempted to extend the work of previous

investigators in the area of assumed similarity who have

76
shown that expected meeting facilitates this type of
belief. It was expected that past findings would be
replicated, and, in addition, expected meeting
with a disagreeing other would further facilitate
assumed similarity even more than expected meeting with
an other whose opinions were unknown. Second, it attempted
to demonstrate differences in estimates of consensus as
a function of expected meeting. In particular, it was
hypothesized that, as with assumed similarity, expecting
to meet with another person to discuss a recent decision
would heighten estimates of consensus relative to
individuals who did not expect to meet someone. And, it
was hypothesized that expecting to meet with a
disagreeing other would increase anxiety or need for
social support, and that these needs would facilitate
false consensus. Three, in general, the present study
attempted to clarify the role of motivational and

nonmotivational components in false consensus.

Replication of Miller and Marks (1982)

The observed differences between expected meeting
and no expected meeting groups on the measure of assumed
similarity, which replicated those of previous studies, are
not readily explained by a nonmotivational interpretation.
Among the possible nonmotivational explanations,
selective exposure is not viable, since the amount of

information regarding others is constant between

77
the two comparison conditions. Nor are there differing
perspectives to cause subjects to make errors in
attribution. Moreover, it seems unlikely that
there are differences in the availability of the
information between expected meeting and no expected
meeting groups.

On the other hand, the observed projection of
similarity onto others may have had some motivational
basis. In Miller and Marks' (1982) study, subjects
projected their verdicts onto peers whom they had not
previously met. The authors suggested that in this
situation where the subject was asked to discuss to
consensus a court case, subjects would exaggerate the
similarity of their viewpoints with an immediately
present other in order to reassure themselves that the
impending social interaction would be smooth,
harmonious, and pleasant. In contrast, the present study
did not allow the kind of directly self serving response
possible in the Miller and Marks study. Subjects were
asked to predict how specific others (e.g., best friend,
favorite family member, etc.) would choose if given the
same information. Thus, the present design allowed some
clarification of one concern voiced by Miller and Marks.
They wondered whether or not the tendency to project
one's opinion arose simply from the required consensus

decision in the forthcoming interaction or whether

78
concern with establishing and maintaining a sense of
social support for and agreement with one's views
motivated projection. The present findings tend to
support the latter contention, since similarity was
assumed for targets with whom the subject did not intend
to immediately interact.

The failure to replicate the expected meeting
effect demonstrated by Miller and Marks (1982) on the
consensus measure was one of the most suprising
results of the present study. It is unclear why subjects
would give higher estimates of assumed similarity under
conditions of expected meeting, yet not estimate higher
levels of false consensus. The assumed similarity
results are consistent with those of Miller and Marks
who found that subjects assumed similarity with a
specific, previously unknown other. But, the present
results are only partially consistent with the findings
of Marks (1983) who found that subjects projected their
own responses onto both their general peer group (i.e.,
demonstrated false consensus) and assumed similarity
with specific acquaintances of the subject.

Although subjects assumed both consensus and similarity
in Marks' study, one qualification should be noted. In
Marks' (1983) study subjects were asked to make a decision
regarding the guilt of a suspect in a realistic trial.

Subjects were asked to respond either guilty or not

 

79
guilty. Subjects who endorsed a guilty verdict,
clearly estimated that a majority of others would
agree with them (i.e., demonstrated a bias in their
estimates of consensus), whereas those who endorsed a not-
guilty verdict predicted that approximately half would
find the suspect guilty and half would find the suspect
non-guilty. In contrast, regardless of verdict, subjects
overestimated the extent to which specific others would
agree with them (i.e., they assumed similarity). Marks'
(1983) results suggest that overestimation of consensus
may only occur in circumstances where subjects' choices
may be costly, regrettable, personally or morally
repugnant. It is possible that the manipulation used
in the present study did not generate such strong
reactions from subjects.

One palusible explanation for these findings is
that the manipulation of meeting and information about
others did not have the anticipated effect of raising
anxiety. It was hypothesized that expected meeting with
others would raise anxiety and create a need to bolster
one's sense of social support. However, an appropriate
manipulation check of anxiety was problematic. If
assuming consensus reduces situational anxiety, then
measuring anxiety following assessment of consensus (as
in the present study) should show no differences in

anxiety between groups. On the other hand, if anxiety is

80
measured before estimates of consensus, this
identification of one's internal state might affect
estimates of consensus. In either case, an appropriate
manipulation check of anxiety was missing from the
present study. And, it is possible that anxiety was not
generated.

I also had predicted that under the knowledge
conditions where subjects did not know whom they would
meet (C, F), the results of both the false consensus and
assumed similarity measures would be similar to those in
the no knowledge conditions. However, this was not the
case for either the false consensus or the assumed
similarity measure. If anything, the results for the
false consensus measure were opposite to what was
predicted. Subjects who did not expect a meeting made
higher estimates of consensus than subjects in the
expected meeting condition. It appears then, that
knowledge about others' decisions and expected meeting
had an unanticipated effect on estimates of consensus.

Consistent with the position of Sherman et al. (1984),
who predicted that subjects would make higher estimates of
consensus under conditions of failure, I predicted that
subjects who were faced with meeting a disagreeing other
would make higher estimates of consensus relative to
subjects who expected to meet with an agreeing or an

unknown other. However, these predictions were not

81
supported. The results may, in part, be attributable to
the unexpected effect of information about others
discussed above (p.64). Yet, the relative differences
between the expected meeting/no knowledge condition
(Cell A) and the expected meeting/knowledge conditions
(Cells D, E) are not great, so it is likely that any
moderating effect of knowledge cannot entirely account
for the failure to conceptually replicate the findings
of Sherman et a1. (1984).

One plausible explanation for the overall similarity
between expected meeting conditions is that contrary to
predictions, information about others constrained
and stablized subjects' estimates of consensus. Rather
than heighten concerns with being correct, information
about others may have simply reinforced subjects' belief
that "most others believe as I do." That is, information
about others probably failed as a manipulation of
subjects' motivational state. Rather than heighten
anxiety, the information may have been used deductively
and then, was not interpreted in motivational terms.

A similar, though more potent manipulation of
opinion projection might be achieved by first telling
subjects that there is an objectively correct or
incorrect solution to the decision they are making, to
provide them with success/failure feedback, and then

telling them they will be meeting someone who either

82
succeeded or failed. It is possible that such a
manipulation would create conditions (e.g., subject
"failed," other "succeeded") which may better elicit a
defensive state where opinion projection would be

likely.

Threats To Esteem - A Potential Underlying Mechanism

The previous discussion suggests that direct
threats to the self should heighten opinion projection.
Using this perspective, some of the previous research
in the area of false consensus may be interpreted within
an esteem framework. Although Sherman et al. (1984)
suggested that subjects assumed consensus to bolster
their sense of social support, their manipulation of
success/failure may have functioned as a manipulation of
self-esteem rather than a manipulation of social
support. In contrast, the present study manipulated the
degree of objective and anticipated social support for
one's opinion. This difference between the two studies,
and the resulting differences in results, suggest that a
more direct threat to self may be necessary for arousing
differences in false consensus. The work by Krahe, who
also gave subjects feedback regarding success or
failure lends itself to a similar interpretation of
opinion projection.

In addition, the findings of other research are

also consistent with the interpretation that false

83
consensus is an esteem related mechanism. For instance,
Sherman, Pressin, Chassin, Corty, Olshavsky (1983)
examined estimates of smoking behavior in middle school
students, high school students and adults. Their results
indicated that middle school students were more likely
to make false consensus judgements regarding smoking
behavior than were their older counterparts. The authors
suggested that their results could be attributed, in
part, to several different factors including selective
exposure and need for social support. However, it may be
the case that there are age differences in the use of
smoking behavior as a bolstering mechanism for
self-esteem; e.g., middle school students might use
smoking behavior as a partial contributor to a sense of
self-importance. Whereas adults use smoking, for this
purpose, to a lesser extent as an indicant of status,
self-importance and maturity. Thus, the potential
explanations for false consensus Sherman et a1. (1983)
proposed may in fact involve variables which covary with
self-esteem needs.

Further, the work by Messé & Sivacek (1979), who
used a confrontational situation, showed that subjects who
were more likely to guess an other acted as they had
in a Prisoner's Dilemma when that person was an
"opponent" than when she was someone in a different

dyad. Their results suggested that subjects seek to

.f

84.

justify their responses, possibly because they felt
culpable for an act which could be perceived as negative.
It is possible that subjects perceived more similatity
with opponents to maintain their esteem, and feel good
about their decision.

Increased belief projection (whether similarity
or consensus) then, may be one type of self-
serving bias which people use to enhance postive
feelings about themselves. Such a view would suggest
that false consensus is one means of actively ehancing
oneself and as such, it is a reflection of motivational
forces -- call it need for self-esteem. Some indirect
and anecdotal evidence exists for such a position. For
instance, work on self- serving bias has shown that
individuals who tend to explain their behavior with the
greatest amount of self-serving bias also tend to score
high on measures of self-esteem (Ickes & Layden, 1978).
More importantly, Abramson and her colleagues (Alloy and
Abramson, 1980) have found that depressed people, who
tend to be low in self-esteem, also tend to be accurate
in their responsibility for failure, whereas others tend
to attribute failure to other factors. Thus, depressed
individuals tend to show little self-serving bias. Is it
possible that depressed subjects would make less
self-enhancing (more realistic) estimates of consensus
than non-depressed, high self-esteem individuals? If so,

this pattern of responses would challenge a cognitive

85
explanation of belief projection which assumes that (given
equivalent levels of information processing ability)
high and low self-esteem individuals should demonstrate
an equivalent amount of false consensus/belief
similarity bias.

However, the relationship between consensus
estimates and esteem is probably much more complex than
that outlined above. Crocker (1985), for example, has
suggested that differences in esteem and actual status
of position may interact to influence stereotyping. For
instance high esteem individuals in high status
positions tended to make fewer stereotyped attributions
than did high esteem individuals in low status positions.
Similarly, these variables may interact to influence the
degree of false consensus demonstrated by subjects under
similar circumstances. Subjects who are high in esteem
needs (and potentially unfulfilled, i.e., holding low
status positions) would be expected to make much greater
estimates of consensus than subjects whose esteem needs
are presently satisfied (i.e., high esteem subjects who

presently hold high esteem positions).

86
Stable Self-Serving Perception and Opinion Projection

As the previous discussion suggests, the robust nature
of self-serving perceptions has been extensively
demonstrated. What is unclear, though, is the extent to
relatively pervasive, the extent to which it can be
manipulated, and (based on the depth, breadth, and
stability of the phenomenon) the relative importance of
opinion projection as a psychological phenomenon.

There is a body of research which indicates that
egocentrism is a pervasive human quality. Myers (1983) has
suggested that people have a tendency to think of
themselves as above average on most if not all socially
valued dimensions. For example, of nearly a million high school
students who were asked to respond to a question
regarding their abilities, 70 percent rated themselves
as above average, and only 2 percent as below average in
leadership ability. Similarly, none of the students rated
themselves below average in the ability to get along
with others.

People tend to demonstrate an egocentric bias on many
other dimensions as well. For instance, they tend to
predict that they will live longer than that projected
by actuarial tables (Larwood, 1978). And, even most
drivers who have been hospitalized for auto accidents
believe that they are better than average drivers

(Svenson, 1981). One striking similarity between these

87
results is the relative stability of the positive
self-perceptions -- even in the face of strong
disconfirming evidence.

Given the strong evidence regarding the
pervasiveness of self-serving biases, the results of the
present study, and those of previous research, one must
ask are manipulations of opinion projection really
significant? This question is asked in a rhetorical, not
in a statistical sense. Is the behavior both extensively
expressed, and sufficiently variable across individuals or
situations to be of interest to psychologists? This is a
question for which no easy answer can be supplied.
However, I would suggest that some interesting questions
could be addressed in this area. Accordingly, I have

discussed some directions for future research below.
DIRECTIONS FOR FUTURE RESEARCH

Given the limitations of the present study, the
question becomes, what should be the direction for future
research? For cognitive factors, one potential starting
point might involve manipulating the particular
components that are thought to influence attitude
accessibility, e.g., repetition, emotional interest,
concreteness, and spatial and temporal proximity. For
motivational factors, attempting to promote states of
arousal or anxiety (as in the present study) in ways

that the arousal is directly measureable, and the

88
manipulation of esteem provides the basic starting
points for examining the link between opinion projection

and motivation more directly.

Cognitive Manipulations

Recent research by Powell and Fazio (1984) has
provided some support for the notion that repeated
attitudinal expression may enhance attitude
accessibility. Their study simply asked subjects to
repeat their attitudinal responses in a lengthy
questionnaire where they were asked their opinion once
or several times. Their attitudes were subsequently
assessed on a reaction time measure. Given the link that
repeated attitudinal expression enhances accessibility,
and the hypothesis that accessibility enhances false
consensus, repeated attitudinal expression should
enhance false consensus. Thus, a simple and appropriate
test of the availability hypothesis would ask subjects
to repeat their attitudinal reaponses and subsequently
ask for estimates of false consensus. A critical control
group could be formed by asking subjects to think about
and repeat the attitudes opposite their own.

A second, hypothesized means of enhancing
accessibility involves increasing the ”emotional
interest" of the object of the attitude. However, it is
not completely clear whether one would call this interest

enhancement a cognitive or a motivational influence --

89
if anything, it is probably a combination of the two.
In any event, if subjects were provided with a task that
had "real" as opposed to pallid consequences, a greater
degree of accessibility and false consensus might be
anticipated. Although some disconfirming evidence for
this hypothesis already exists -- Marks (1982), showed
that issue involvement had no effect on consensus
estimates -- other operational definitions of emotional
interest might produce the effect postulated by cognitive
theorists.

Nisbett and Ross (1980) suggest that availability
is also enhanced by stimuli that are proximal in time or
space. It seems likely that manipulating the time
period between decision and the assessment of opinion
projection could affect the impact of availability. By
implication, if availability affects false consensus,
time of consensus assessment should affect the magnitude
of consensus estimates. Yet, there may be one problem
with this line of reasoning. Manipulating the time of
measure probably will not affect consensus estimates for
issues that have a high degree of contemporary social
relevance (i.e., abortion, see Appendix A) since the
availability of attitudes regarding these issues are
likely to remain stable. Moreover, a similar problem
could be evidenced with the use of novel stimuli which

provide the subject with a unique, interesting

90
experience. Even if the subject is assessed days later,
the novelty may induce an enduring availability effect.
Thus, the general notion that temporal or spatial
proximity can raise availability seems generally
reasonable. But, such an effect may only occur under
certain circumstances which are not particularly
generalizable or particularly psychologically
interesting.

In line with research and speculations about
attitude availability, recent work on direct and
indirect experience may serve as a potential perspective
from which to address the availability/false consensus
issue. Fazio and his colleagues have examined the
effect of direct experience on attitude-behavior
consistency. It seems likely that direct versus
indirect experience would also be seen by cognitive
theorists as a potential means for enhancing

accessibility and examining its impact on false consensus.

Attribution

A further area of study for false consensus
includes a comparison of attitude accessibility to
attributional causes of false consensus. Gilovich,
Jennings and Jennings (1983) demonstrated that subjects
who were led to believe that their decisions were
related to dispositional factors made lower estimates of

consensus than subjects who were led to believe that

91

their decisions were due to situational factors. In the
area of objective self-awareness (OSA), researchers have
noted that subjects who are objectively self-aware are
more likely to make dispositional attributions than
subjects who are not self-aware. Given this finding, it
is reasonable to speculate that subjects who are made
objectively self-aware will make lower estimates of
consensus than subjects who are not made self-aware.
But, there is indirect evidence to support the opposite
prediction. While objective self-awareness is purported
to facilitate dispositional attributions, it is also
purported to enhance concentration on one's own
decision. In essence, self-awareness should enhance
rehearsal of one's own opinion, and hence make one's own
opinion highly salient. Thus, under conditions of
objective self-awareness, subjects should hold highly
available, dispositionally based beliefs. If false
consensus is the cold cognitive phenomenon that Ross and
his colleagues have suggested, one would hypothesize
that high attitude availability under conditions of high
self-awareness would lead to an overestimation of
consensus. Whereas, an attributional perspective would
suggest that subjects who were high in self-awareness
would not overestimate consensus.

Given Wicklund's (1972) discussion of objective

self-awareness, a delineation of false consensus as a

92
motivational or nonmotivational phenomenon is still
problematic. Wicklund suggests that self-focused
attention leads to self-evaluation, which, in turn
creates either self-exaltation or self-criticism. In
either case, OSA may engage an esteem component which
confounds its expected effect on availability. If, using
a reaction time measure, no differences in availability
were demonstrated between high and low objective self-
awareness conditions, inferences regarding the relative
contribution of esteem and attributional focus could be

made.

Adding Confirmatory Evidence - When Does It Enhance

Consensus Estimates?

Yet another question concerns whether or not
individuals will demonstrate a greater bias toward self
when there is additional confirmatory evidence. I would
suggest that, in fact, the bias would not be appreciably
enhanced. For instance, the results of the present study
indicated that regardless of condition (amount of
information present) subjects estimated consensus at a
level of approximately 65 percent. Even with the use of
materials where the actual decision split was dramatic
(80/20) consensus estimates remained at approximately
the same level. It may be quite reasonable to argue that
false consensus is a fairly stable phenomenon which

requires a strong manipulation, like threat to self-

93

esteem to enhance or depress estimates of consensus.
The Use of Personal Estimates of Base Rate

One result of the present study showed that
consensus estimates of subjects who were not told whom
they were to meet (agreeing or disagreeing other) made
dramatically different estimates of consensus when their
estimates were broken down according to whom they guessed
they would be meeting. Subjects who had made low
estimates of consensus (M=49.06) tended to believe that
they would be meeting with someone who disagreed with
them. On the other hand, subjects who made high
estimates of consensus tended to believe that they would
be meeting with someone who made the same decision as
they had (M=69.07). This pattern of results suggest that
subjects' estimates regarding particular cases may be
based on their own subjective estimate of the base rate.

If, as indicated in the present study, subjects do
make predictions based on their own estimate of the base
rate, it is possible, if not probable, that much of the
work done by Kahneman and Tversky has misinterpreted a
critical element in their investigation of the
prediction process. They suggested that individuals'
sense of base rate did not influence estimates in
prediction problems where some artificial base rate was

given. However, it may very well be the case that

94
subjects do use their own subjective estimates in the

prediction of frequency and probability.

Motivational Manipulations

As discussed previously, there may be a number of
potential motivational causes of false consensus. First,
consensus might be heightened by a private need for
social support. For example, false consensus might
serve the purpose of providing a private affirmation of
correctness. Second, consensus might be affected by a
public justification of behavior. Subjects may assume
consensus as a means of publically enhancing themselves
or publically mitigating their culpability in a
potentially negative situation. Or, both of these may fall
under a general need for esteem maintainance. Still
more generally, false consensus might be a function of
hedonic relevance, as demonstrated by Crano (1983).

The thrust of future research examining the.
motivational perspective will need to develop adequate
manipulation checks for perceived social support,
anxiety, esteem and hedonic relevance. For instance,
it is unclear the extent to which false consensus may be
used as a general means of reducing state anxiety. If a
general rise in arousal is interpreted as social
anxiety, other methods of inducing arousal could be used
to test the motivational perspective of false consensus.

Rather than just speculating that arousal has taken

95
place, such arousal might be assessed using a direct
method of physiological measurement - e.g., heart rate,
galvonic skin response, etc. Such measurements have been
used in previous research on dissonance. For instance,
they have demonstrated the hypothesized differences in
anxiety between dissonance and non-dissonance conditions
(Croyle & Cooper, 1983).

As suggested previously, very little work has been
done that directly examines the relationship between esteem
and false consensus. Yet, there appears to be a strong
similarity between the manipulations of self-enhancement,
self-justification and esteem, and their effect on consensus.
One direction for future research is to address esteem as
a general factor affecting false consensus.

Most importantly, advocates of the motivational
perspective will have to convincingly control the
extent to which cold cognitive factors influence opinion
projection if their explanation is to be taken seriously.
In order to do this, researchers will need to use
measures of attitude availability, or other cognitive
paradigms, to demonstrate equivalent levels of

availability between motivated conditions.

SUMMARY
The present study has, perhaps, generated
more questions than answers. Among these questions,

I asked how future research might provide better tests of

96
the availability explanation of false consensus. For
example, it was suggested that repeated attitudinal
expression might increase estimates of consensus.
And, it was suggested that, if vivid information is
more cognitively available, it too should enhance
estimates of consensus. However, recent pilot studies
fail to support either ot these predictions. (see
Appendix A).

The observed stability of the consensus estimates
across the experimental conditions raised questions
regarding when evidence, confirmatory or disconfirmatory,
will affect consensus estimates. It is clear from the
present results that individuals did not inflate their
estimates of consensus when they expected to meet with
someone who held a different opinion than themselves.
While other research using different materials may be
able to generate differences in consensus using a
manipulation similar to the one used in the present study,
the present results tend to support the contention that
the self-serving bias of opinion projection is a
relatively stable phenomenon.

I The relative effect of motivational mechanisms on
false consensus was also discussed. Among the
potential factors which might affect estimates of
consensus, little has been done to identify the

relative communality or differences between the

97

varied perspectives. It was suggested that self-
enhancement, self-justification and need for social
support all might fall under the general notion of
esteem maintenance. Yet, one other perspective remains
unreconciled with an esteem based explanation.
Crano's (1983) vested interest research demonstrates
that consensus can be enhanced by making threats to
individuals' (non esteem related) rewards. What
similarities or differences exist between the

esteem threatening and concrete financial threats

of Crano's paradigm? Or, what is it about people
such that they make similar attributions following
these threats? These questions remain to be answered

by future research.

APPENDI CI ES

APPENDIX A: Additional Pilot Studies

98

Two additional pilot studies were performed to
explore further the false consensus effect. Abtracts
of these studies are presented below.

Study 1

It was hypothesized that the lack of differences in
consensus estimates in the dissertation was due to task
related issues. Specifically, it was possible that
subjects did not see the task as interesting or
important. In this pilot study, I used the issues of
affirmative action and abortion to examine the effect of
expected meeting with an agreeing or a disagreeing
other.

The study took the form of a simple
2(Abortion/Affirmative Action) by 2(Meet Agree/Meet
Disagree) by 2(Order of Issue) repeated measures factorial
design. The subjects were 41 male and female
undergraduates. After reading a short, neutral
descrition of the issue, subjects were asked to indicate
whether or not they supported the issue and wished the
present laws to remain the same, or if they wished the
laws would be changed. Subjects were then asked what
percent of other undergraduates would decide as they
had. I

The results indicated no differences between

expecting to meet with an agreeing (M=66.25) or a

disagreeing other (M=63.58), {(1,37) =.48, ns. And,

there were no differences between consensus estimates for

99

order of presentation or task. These results indicated
that expecting to meet with an agreeing versus a
disagreeing other, even on important issues did not
affect estimates of consensus. Such a finding led me
to reconsider cognitive explanations of the false
consensus phenomenon.
Study 2

I speculated that if false consensus is driven by
attitude accessibility, then concentrating on one's own
decision should increase accessibility and as a
consequence inflate consensus estimates; and,
distraction from one's own position should reduce
accessibility and deflate consensus estimates. Female
undergraduates (N=102) were asked to make the same
supervisory decisions used in the dissertation.
Following their decision, some subjects were asked to
write arguments supporting their decision and repeat
their decision either 1 or 4 times on several other
pieces of paper which were to be used for "administrative
purposes." Other subjects were asked to write out
arguments which supported the decision which was
opposite their own, and to repeat the opposite decision 1
or 4 times. After listing arguments and repeating
decisions, subjects were assessed on the false consensus
measure which asked what percent of undergraduates would

make the same decision they had. Thus, the design was a

100
2(Repeat Own/Other Decision) by 2(Times Repeated 1 or 4)
with an external control group which only read the
materials and made the consensus estimate.

The results indicated no differences in consensus
estimates between subjects who repeated their own
decision and those who repeated the opposite decision.
Consensus estimates ranged from 63 percent to 69 percent
with a average standard deviation of 18.5. Nor were
there any differences in the total number of arguments
by subjects who listed their own supporting arguments
versus those who listed arguments for the position
opposite their own. No conclusions were drawn from

these results.

APPENDIX B: Experimental Materials

101

EXPERIMENTER INSTRUCTIONS

1) Have a list of the names of the person(s) you are to
meet for the experimental session.

2) Meet each subject individually at one of the
specified locations.
A) Ask if they are waiting for the "Supervisory
Decision Making" study, ask for their names.
B) If they are waiting for our study, introduce
yourself, and lead them to room 201.

3) Ask the subject to have a seat. Explain the
procedure using the exact language presented below.

We are interested in how people make supervisory
decisions. Right now, we would like you to read these
two pages to get an idea of what you will be doing
today. To participate in the study, you need to read
over the consent form, sign and date it. I have to seat
other people who will also be participating, so I will
have to be away for a minute. In order to cut down on
distractions we would like you to wear these headphones.
When you have finished reading the project description,
slip the signed consent form under the door and I will
be with you as soon as I can. O.K.? Do you have any
Questions? O.K., I will be back in a minute.

O.K., Are you ready to begin? I am going to give you a
series of supervisory decisions to make, and you will be
asked to make one decision at a time. In each case, I
will give you a description of a police officer who is
up for either promotion, reprimand or a merit raise. It
will be your job as a supervisor to decide how the
officer will be treated. After you have carefully read
the description, indicate at the bottom of the page how
you would decide on this particular case. Once you have
decided, slip your response under the door, and I will
get back to you in a minute.

102
Instructions for Condition A:

Expected Meeting/No knowledge

I see that you have decided to .

 

In a few moments we would like to videotape a
discussion between you and another subject about the
officer review brief you just saw. We are interested in
how people come to agreement on a decision, so we will
ask you and the other person to try to reach a single
decision on the case you just responded to.

But, before you discuss this review brief with the
other person, we would like you to answer a few
questions and make a few more supervisory decisions.
Please read over the questions carefully and answer them
to the best of your ability. Again, when you are done
answering the questions, slip the questionnaire under
the door.

(PRESENT FIRST SET OF QUESTIONS)

Wait for the subject to complete the questoionnaire.

You will not be meeting anyone to discuss the remaining
supervisory decisions. Now, lets go onto the next
officer review briefs. This packet contains two review
briefs. read each one carefully, make a decision, and
slip it under the door when you are done.

MUST MEMORIZE-put in words that are comfortable for you.

Now that you are all done with the decision making
process we were going to have you meet with another
person for a videotaped discussion. Unfortunately, we
have been having some problems with our video equipment.
Still, your participation has been very helpful and your
data will still be of use to us. Thank you for
participating.

103

Instructions for Condition A (continued)

Debriefing

In each case, provide the subject with a
debriefing form. Never allow a subject to leave without
going over the form with him or her.
Sign the subject's credit form

Thank them again for participating and dismiss each one
individually.

104
Instructions for Condition B:

No expected meeting/No knowledge

I see that you have decided to .

In a few moments we would like you to respond to some
other supervisory decisions. But, before you make more
supervisory decisions we would like you to answer a few
questions. Please read over the questions carefully and
answer them to the best of your ability. Again, when you
are done answering the questions, slip the questionnaire
under the door.

(PRESENT FIRST SET or QUESTIONS)

Wait for the subject to complete the questoionnaire.
Now, lets go onto the next officer review briefs. This
packet contains two review briefs. read each one

carefully, make a decision, and slip it under the door
when you are done.

105
Instructions for Condition C:

Expected Meeting with Either Agree/disagree

I see that you have decided to .

In a few moments we would like to videotape a
discussion between you and another subject about the
officer review brief you just saw. We are interested in
how people come to agreement on a decision, so we will
ask you and the other person to try to reach a single
decision on the case you just responded to. Most people
who expect to meet someone tO discuss an issue are
curious about how the others decided. Of the other
people who are participating in this in this study right
now, one person made the same decision as you, and one
person decided differently. I will flip a coin to
determine which person you will be meeting to discuss
the case.

But, before you discuss this review brief with the
other person, we would like you to answer a few
questions and make a few more supervisory decisions.
Please read over the questions carefully and answer them
to the best Of your ability. Again, when you are done
answering the questions, slip the questionnaire under
the door.

(PRESENT FIRST SET OF QUESTIONS)
Wait for the subject to complete the questoionnaire.

You will not be meeting anyone to discuss the remaining
supervisory decisions. Now, lets go onto the next
Officer review briefs. This packet contains two review
briefs. read each one carefully, make a decision, and
slip it under the door when you are done.

MUST MEMORIZE-put in words that are comfortable for you.

Now that you are all done with the decision making
process we were going to have you meet with another
person for a videotaped discussion. Unfortunately, we
have been having some problems with our video equipment.
Still, your participation has been very helpful and your
data will still be of use to us. Thank you for
participating.

106
Instructions for conditions D and E:

Expected Meeting with Agreeing/Disagreeing Other

I see that you have decided to .

In a few moments we would like to videotape a
discussion between you and another subject about the
officer review brief you just saw. We are interested in
how people come to agreement on a decision, so we will
ask you and the other person to try to reach a single
decision on the case you just responded to. Most people
who expect to meet someone to discuss an issue are
curious about how the others decided. Of the other
people who are participating in this in this Study right
now, one person made the same decision as you, and one
person decided differently. I fliped a coin to determine
which person you will be meeting to discuss the case.
The flip came out so that the person you will be
meeting:

D) MADE THE SAME DECISION AS YOU.
E) DECIDED DIFFERENTLY THAN YOU.

But, before you discuss this review brief with the
other person, we would like you to answer a few
questions and make a few more supervisory decisions.
Please read over the questions carefully and answer them
to the best of your ability. Again, when you are done
answering the questions, slip the questionnaire under
the door.

(PRESENT FIRST SET OF QUESTIONS)

Wait for the subject to complete the questoionnaire.

You will not be meeting anyone to discuss the remaining
supervisory decisions. Now, lets go onto the next
officer review briefs. This packet contains two review
briefs. read each one carefully, make a decision, and
slip it under the door when you are done.

107
Instructions for Condition F:

NO Expected Meeting/One agrees-One disagrees

I see that you have decided to .

Some people are curious how Others decided on the
supervisory decision you just made. So, let me tell you
that of the other subjects who are participating in the
study right now, one person made the same decision as
you, and one person decided differently.

In a few moments we would like you to respond to
some other supervisory decisions. But, before you make
more supervisory decisions we would like you to answer a
few questions. Please read over the questions carefully
and answer them to the best of your ability. Again, when
you are done answering the questions, slip the
questionnaire under the door.

(PRESENT FIRST SET OF QUESTIONS)

Wait for the subject to complete the questoionnaire.
Now, lets go onto the next officer review briefs. This
packet contains two review briefs. read each one

carefully, make a decision, and slip it under the door
when you are done.

108
Officer Review Brief
Officer: Thomas Dayton
Years Employed: 12
Training Class Rank, Rank/Class size: 230/655

Officer Dayton successfully secures crime scenes
and protects evidence despite occasional harrassment
from citizens and the media. The officer is able to pull
out the most important details of a witness' statements.
Officer Dayton also uses the proper methods for
approaching dangerous suspects. While on patrol, he
makes sure to get to know the businessmen on his beat.
Although the officer sometimes finds it difficult to
cope with young people. Officer Dayton generally is seen
as being courteous and friendly with various citizen
groups and communicates in an unbiased manner. In
addition, Officer Dayton's reports are short, neat amd
written in good grammar.

In the last six months, the Officer has been sent
home twice for being intoxicated at report. He is
presently under treatment with the staff psychologist
and attends alcoholics anonymous. There are no
documented incidents of alcohol interfering with the
performance Of his duty. Officer Dayton is sometimes
late for meetings, and he has been known to be
antagonistic toward his supervisors. Yet, his
performance is generally solid.

Note. Merit pay increases should be applied to officers
who demonstrate above average performance. They should
not be handed out as a matter of course. Please give all
your decisions serious consideration.
As a supervisor would you

1) Award the Merit pay increase

2) Withhold the Merit increase
How confident are you that you made the best decision
regarding the merit pay evaluation?

Not at all confident : : : : Very Confident
l 2 3 4

 

109
Officer Review Brief
Officer: Mark Martindale
Years Employed: 3

Training Class Rank, Rank/Class size: 453/480

Officer Martindale appears to have a good rapport
with the public. He accepts criticism well and is very
interested in his work. Most of Officer Martindale's
reports are well written. He is good on the radio,
saying only what is necessary. The Officer attempts to
handle additional case assignments when his own are
completed, and he often works on cases while off duty.
Officer Martindale is calm and deliberate in his
actions, but he seems to think he knows more than he
does.

In the last month, Officer Martindale forgot to
advise a suspect of his constitutional rights when
making an arrest. This led to the dismissal of one
case. The officer is not always observant on patrol --
he sometimes missis things the best officers notice.
Also, the officer sometimes issues traffic citations in
an unjustified and uneven manner. Otherwise, he appears
to be dependable and have reasonable job knowledge.

Note. Patrol Officers come under reprimand when their
service has been flawed -- demonstrating unprofessional
conduct, breaking departmental regulations or acting in
a manner at risk to the public. Please give all your
decisions serious consideration.
As a supervisor, would you

1) Issue the reprimand

2) Withold the reprimand
How confident are you that you made the best decision
regarding the reprimand decision?

Not at all Confident : : : : Very Confident
1 2 3 4 5

110
Reprimand Evaluation Sheet
In your Opinion, if given the same information as you
and other subjects have been given, what percent of
undergraduates would make the same decision you made?
Percent

In the next two questions, make sure that the sum of
your responses total 100%

What percent of your fellow students would feel
confident about the decision they would make?
Percent Confident

What percent of your fellow students would NOT feel
confident about their decision?

Percent Not at all Confident
Think about each of the persons listed below. If they
were in the supervisory position, indicate how you think
they would have decided regarding the officer under

review. (Check one for each person)

Issue Withhold
Reprimand Reprimand

1) Your best same sexed friend

2) Your favorite teacher Of all time
3) The person you dislike the most
4) An officer not under review

5) Your favorite family member

111

Reprimand - continued

Do you think the person with whom you will be meeting
made the same or a different decision regarding the
officer reprimand?

(Circle one)

A) Made the Same Decision 8) Made a different decision
In your Opinion, how Confident is the person you will be
meeting about the decision they have made regarding the

reprimand evaluation?

Not at all confident : : : : Confident
l 2 3 4 5

 

Do you have a guess whether the person you will meet is
either male or female?

A) Male
B) Female
C) Equal chance of being male or Female

If you did have a guess about the sex of the person you
will be meeting to discuss the reprimand evaluation, did
you think about whether the person is male or female
when you made your decision regarding what percent or
other undergraduates made the same decision as you ?

YES NO

112
Merit Raise Evaluation Sheet
In your Opinion, if given the same information as you
and other subjects have been given, what percent of

undergraduates would make the same decision you made?

Percent

In the next two questions, make sure that the sum of your
responses total 100%
What percent of your fellow students would feel
confident about the decision they would make?

Percent Confident
What percent of your fellow students would NOT feel
confident about their decision?

Percent Not at all Confident
Think about each Of the persons listed below. If they
were in the supervisory position, indicate how you think
they would have decided regarding the officer under
review. (Check one for each person)

Withold
Award Raise Raise

1) Your best same sexed friend
2) Your favorite teacher of all time
3) The person you dislike the most

4) An officer not under review

5) Your favorite family member

113
Merit Raise - continued

Do you think the person with whom you will be meeting
made the same or a different decision regarding the
merit pay increase?

(Circle one)
A) Made the Same Decision

B) Made a different decision

In your opinion, how Confident is the person you will be
meeting about the decision they have made regarding the
merit pay evaluation?

Not at all confident : : : : Confident
1 2 3 4 5

Do you have a guess whether the person you will meet is
either male or female?

A) Male
B) Female
C) Equal chance of being male or Female

If you did have a guess about the sex of the person you
will be meeting to discuss the merit raise evaluation,
did you think about whether the person is male or female
when you made your decision regarding what percent or
other undergraduates made the same decision as you ?

YES NO

114
Introductory Psychology Student Questionnaire

Many students in introductory psychology classes
who participate in research for extra credit are asked
to meet with other students to discuss issues or solve
group problems. In the questions below, please indicate
how you feel about meeting others in the present
experiment.

Friendly : :
l 2 3 4 5

Unfriendly

Very relaxed : . Not at all relaxed
1 2 3 4 5

: Uncomfortable
l 2 3 4 5

Very comfortable

Curious : : : : Uninterested
l 2 3 4 5

Very anxious : : Not at all anxious
1 2 3 4 5

 

Talkative : : : : Not talkative
l 2 3 4 5

Bored : : : : Excited
1 2 3 4 5

115
LIST OF REFERENCES
Abelson, R. (1963) Computer simulation of "hot" cognition.
In S. S. Tompkins & S. Messick (Eds.) Computer
Simulation of Personality. New York: Wiley.
Abelson, R., Aronson, E., McGuire, W., Newcombe, T.,

Rosenberg, M., & Tannenbaum, P. (Eds.) The Cognitive

 

Consistency Theories: A Sourcebook. Chicago: McNally,
1968.

Alloy, L. 8., & Abramson, L. Y. (1980) The cognitive component
of human helplessness and depression: A critical anlysis.
In J. Gaber and M. E. P. Seligman (Eds.), Mgmgg
Helplessness: Theory and Applications. New York:
Academic Press.

Arkin, R., a Burger, J. (1980) Effects of unit relation
tendencies on interpersonal attraction. Social
Psycholonguarterly, 43, 380-391.

Bennett, D. H., & Holmes, D. S. (1975) Influence of denial
(situation redefinition) and projection on anxiety

associated with threat to self-esteem. Journal of

 

Personality and Social Psychology, 32, 915-921.
Berschied, E., Graziano, W., Monson, T., & Dermer, M. (1976)
(1976) Outcome dependency: Attention, attribution
and attraction. Journal of Personality and Social

Psychology, 34, 978-989.

116

Bradley, G. W. (1978) Self-serving bias in the
attribution process: A reexamination of the fact or
fiction question. Journal of Personality and Social
Psychology, 33, 56-71.

Bramel, D. (1962) A dissonance theory approach to
defensive projection. Journal of Abnormal and Social
Psychology, 34, 121-129.

Bramel, D. (1963) Selection of a target for defensive
projection. Journal Of Abnormal and Social Psychology,
33, 318-324.

Byrne, D. (1971) The attraction paradigm. New York:
Academic Press, 1971.

Crocker, J. (1985) Personal communication.

Crano, W. D. (1983) Assumed Consensus of Attitudes: The
effect of vested interest. Personality and Social
Psychology Bulletin, 3, 597-608.

Croyle, R., Cooper, J. (1983) Dissonance arousal:
Physiological evidence. Journal of Personality
and Social Psychology, 43, 782-791.

Dawes, R. M., McTavish, J., & Shaklee, H. (1977)

Behavior, communication and assumptions about other
people's behavior in a commons dilemma situation.
Journal of Personality and Social Psychology, 3, 1-11.

Eitzman, D. (1984) Consensus bias with regard to illegal

activities. Unpublished Manuscript, Michigan State

University.

117
Festinger, L. (1954) A theory of social comparison

processes. Human Relations, 1, 117-140.

 

Freeman, H., Carlsmith, M., & Sears (1974) Social Psychology,

 

(2nd Ed.). Englewood Cliffs: Prentice Hall.

Frey, D., & Rosch, M. Information seeking after decisions:
The roles Of novelty of information and decision
reversability. Personality and Social Psychology
Bulletin, 33, 91-98.

Gilovich, T., Jennings, D. L., & Jennings, S. (1983)

Causal analysis and estimates of consensus: Undermining
the false consensus effect. Journal of Personality and
Social Psychology, g3, 550-559.

Goethals, G. R., Allison, S. J., & Frost, M. (1979)
Perceptions of the magnitude and diversity of social
support. Journal of Experimental Social Psychology, 33,
570-581.

Greenwald, A. G. (1980) The totalitarian ego. American
Psychologist, 33,603-618.

Harvey, J. H., & Weary, G. (1984) Current Issues in Attribution
and Research. Annual Review Of Psychology, 33, 427-459.

Holmes, D. S. (1968) Dimensions of projection. Psychological
Bulletin, 33, 248-268.

Holmes, D. S. (1978) Projection as a defensive mechanism.

Psychological Bulletin, 33, 677-688.

118
Hull, J. G., & Levy, A. S. (1979) The organizational
functions of the self: An alternative to the Duval and Wicklun
model of self-awareness. Journal of Personality and Social
Psychology, 31, 756-768.
Ickes, W., & Layden, M. A. (1978) Attributional styles. In
J. H. Harvey, W. Ickes, & R. F. Kidd (Eds.)
New Directions in Attribution Research (Vol. 2).
Hillsdale, N.J.: Lawrence Earlbaum.
Janis, 1., & Mann, L. (1977) Decision Making. New
York: Free Press.
Jones, E. E., & Nisbett, R. (1972) The actor and the
observer: Divergent perceptions Of the causes of

behavior. In L. Berkowitz (Ed.), Advances in

 

Experimental Social Psyohology (Vol. 2). New York:
Academic Press.
Kelley, H. H. (1967) Attribution theory in social psychology.
Nebraska Symposium on Motivation, 33, 192-238.

Kelley, H. H., & Stahelski, A. J. (1970) The social interaction
basis Of cooperators' and competitors beliefs about others,
Journal of Personality and Social Psychology, 33, 66-91.

Krahé, B. (1983) Self-serving biases in perceived

similarity and causal attributions Of other people's
performance. Social Psychology Quarterly, 33, 318-329.

Larwood, L. (1978) Swine flu: A field study of

self-serving biases. Journal of Applied Social

Psychology, 33, 283-289.

119

Marks, G. (1984) Thinking one's abilities are unique and one's

opinions are common. Personality and Social Psychology
Bulletin, 33, 203-208.

Marks, G. (1983) The relationship between projection Of
opinion similarity and certainty of opinion correctness.
Unpublished Doctoral Dissertation, University of
Southern California.

Marks, 6., & Miller, N. (1982) Target attractiveness as
a mediator of assumed attitude similarity.

Personality and Social Psychology Bulletin, 3, 728-735.

Marks, 6., Miller, N., & Maruyama, G. (1981) Effects of
targets' physical attractivness on assumptions of
similarity. Journal of Personality and Social
Psychology, 33, 198-206.

Markus, H. (1977) Self-Schemata and processing
information about the self. Journal of Personality and
Social Psychology, 33, 63-78.

Messé, L. A. & Sivacek, J. (1979) Prediction of others
responses in a mixed-motive game: Self-justification or
false consensus? Journal of Personality and Social
Psychology, 3, 602- 607.

Miller, D. T., & Ross, M. (1975) Self-serving
biases in the attribution Of causality: Fact or

fiction? Psychological Bulletin, 33, 213-225.

120

Miller, N. & Marks, G. (1982) Assumed similarity
between self and other Effect of expectation of future
interaction with that other. Social Psychology
Quarterly, 33, 100-105.

Myers, D. G. (1983) Social Psychology. New York: McGraw
Hill Book Company.

Pittman, T, & Pittman, N. L. (1980) Deprivation of control
and the attribution process. Journal of Personality
and Social Psychology, 33, 377-389.

Powell, M. C., & Fazio, R. H. (1984) Attitude
accessibility as a function of repeated attitudinal
expression. Personality and Social Psychology Bulletin,
33, 139-148.

Ross, L. (1977) The intuitive psychologist and
his shortcomings: Distortions in the attribution
process. In L. Berkowitz (Ed.), Advances in
Experimental Social Psychology (Vol. 10). New York:
Academic Press.

Ross, L., S Anderson, C. Shortcomings in the
attribution process: On the origins and
maintenance of erroneous social assessments.

In Kahneman, D., Slovic, P., & Tversky, A. (Eds.),
Judgement Under Uncertainty: Heuristics and Biases.

Cambridge: Cambridge University Press

121
Ross, L., Greene, D., & House, P. (1977) The
"False Consensus Effect": An egocentric bias in social
perception and attribution processes. Journal of
Experimental Social Psychology, 33, 279-301.
Sherman, S. J., Presson, C. C., Chassin, L., Corty,
E., & Olshavsky, R. (1983) The false consensus effect
in estimates of smoking prevalence. Personality and
Social Psychology Bulletin, 3, 197-208.
Sherman, S. J., Presson, C. C., & Cahssin, L.
(1984) Mechanisms underlying false consensus: The
special role of threats to the self. Personality and
Social Psychology Bulletin, 33, 127-138.
Sherwood, G. G. (1981) Self-serving biases in
person perception: A reexamination of projection as a
mechanism of defense. Psychological Bulletin, 33,
445-459.
Sivacek, J. & Crano, W. D. (1982) Vested Interest as a
moderator of attitude-behavior consistency. Journal
of Experimental Social Psychology, 33, 210-221.
Snyder, M. (1984) When belief creates reality.

In L. Berkowitz (Ed.), Advances in Experimental Social
Psychology (Vol. 18). New York: Academic Press.
Svenson, O. (1981) Are we all less risky and more skillful
than our fellow drivers? Acta Psychologica, 31,

143-148.

122
Taylor, S. E., & Fiske, S. E. (1978) Salience,
attention, and attribution: Top of the head phenomena.
In L. Berkowitz (Ed.), Advances in Experimental Social
Psychology (Vol. 11). New York: Academic Press.
Tetlock, P. E., & Levi, A. (1982) Attribution
Bias: On the inconclusiveness of the
cognition-motivation debate. Journal of Experimental
Social Psychology, 33, 68-88.
Tversky, A., & Kahneman, D. (1982) Availability:
A heuristic for judging frequency and probability.
In Kahneman, D., Slovic, P., & Tversky, A. (Eds.),
Judgement Under Uncertainty: Heuristics and Biases.
Cambridge: Cambridge University Press.
Wicklund, R. A. (1975) Objective Self Awareness.
In L. Berkowitz (Ed.), Advances in Experimental Social
Psychology (Vol. 8). New York: Academic Press.
Zuckerman, M. (1979) Attribution of success and failure
revisited, or: The motivational bias is alive and well
in attribution theory. Journal of Personality and

Social Psychology, 33, 245-287.

1 2 3
FOOTNOTES

1

Opinion projection is defined in this paper as the
general tendency for people to overestimate the extent
to which others hold the same Opinion as themselves.
This applys equally to the projection of one's Opinion
onto a specific other (typically called assumed belief
similarity) as well as projection onto others in general
(typically called false consensus).