\ I ......40 O.'70 v .0 .0111. . ...... o v r . . . . . . .. . . .C.I.0 0....1.....¢uuc...‘.... .... O ......:Oh.0..‘l'H.\lh\.c‘It.. .IJI:I‘1.Q'.5‘.E
. . . . ... c . ..... _. .. .

     

              

ocvov... - . .

.. . .. ... . n . ... . ...-.11... ...: ......11.4.0. I...
. . . a . . _. O a. . .0 .. .o . . .11..vva'...r.40 d in...
o o . . . . . . ... .. . . . .... I . .01.. o «a (500.0

... .. . . . .. . _ . _ . ... . . . . 1.. . .. .

. . u v . .. ... . . . . . . . _. . . .. .....o'v...
. . a . a . . .. . . a... u. o .... v... .
. . ...o . . o to o . .. o . o . . . .f. o .1
.. . ... . . o . . . . . ... ... . . ....0. ...H._...:I.c.“
_ . . . . . . .. . u . n . o . . . .1.. . . . .. ... .. .
. . . . . . o . . 0 _ . . ......1. I
. . o n. . I... o . . ‘
. . . . . . . . . . . .o... .. . ...

. . . u . . o 01. . ..
. o . ....o . ... . .... ... .1. .11.. Thcounc... v.9 .h
. _ _ . . . . . _ . ... .. ..L‘. .\. .
. . . . . . . .. .. . ......._................ .. n

. . . . ...
o . . . . . . . . a . . . . . v . .. .wa.o .. . o ... J ....G n
. . ... .. . . . . ... . ...... .... . .1....
. . ... ... . . . . I. .I’ o.

    

o a
...I'Ovv
..o..

   

. .. .. . ... .c . . . . .. . .... . . . .. .. . .... ... o. ......411'. .. .-

. I
. . . . . . . . .. v. o. .
I . . . ._ . ... . . o . ..........4...........
.. . . . . _ . _ .. .. . ...! .
. . a ._ . . ... . — ... . an. ....fil....
. . . .. ....Ip.

. * n ..
. . . .. .1 .. o ... .. ........;...u.._.1.o.a.. ...-0.0.111.1.0r0{1~ M/{WLHr
. . .1 .10.... (—1.4...

 

 

  
  
 

.....v. ...fx. ......33 . ..... ...

   
   

- _ . .. . . . . .. .. .... u p. . ..., . .......r.. ......1...fo.... ......W. .... u. ...”... .1 ¢.f.w.... .
. .. . ._ . .... .. .... 4.. . . ...:...... .t..o1/.’I.r!a”/,’ ......
. .. ... . .. ........._......._..c... ... . ...... a... ..
. U. . .. . . ... . Q... .... . .... .. . ...1..a.,.....)/sr'o/1VT

          

wuooym—ganwwnnnv ...—MW
0
T.
’4‘.
'. C
1:.
.-.
-_,
.'-.'
-I
.t.
d
‘-O

‘MENT

 

ITUDE CHANGE

  
  

of MA.

.

egr
TATE UNWER

BB

. . . o. . ...o;

a . o . t...

ION LITTLE

MAR;
1921

. ... . ......

13‘.ng -

     
 

      

. o..r......r.ub..r'b. a...
.v'..tW..11/p.o-..-. .

 

the D

i‘.’ 1

 

M

. .... .v .0 4'. 4
.........w.. 0...... “...-.... If.“

at.

     

To ROLE'PLAYING muuc

......

.

ATT

for
HIGAN '
‘ UGLAS

 

m.

r

SIGN OF CRITICAL JU.

 
 
  

PPROACH

 

SUSP

   
 

. J.DOO ' ...
. . _ o . .. 01 .1...’J.. «I. .J. .mlarl.00.l
. .. .. .r.. a..l...:.l....'o.ll.l..

. . . .4 ...:1. .o/fo ...
. . . .. . ...
. . . d...)vJ_-o.l!1..o.. . ’w

0 OD'KAO 0“-

p.

.9...” ... .gawar-pay-g "mg-“W-”
‘
O
.
'.
‘
.
I"
I
O
f
'
‘
O
h .0

c
a
0
q
I

'9‘!

.
. . . O
1 .

  

        

      

  
 

. . . . . . .
o. u . u .
0. . . . .9... o. . .
. . . . . . ..s. . .... ..L... ... .. .m..a..uL.\.1...u.:._.4.. ......- ...\...: ..n... o v.. ......~w...r:.f. ”.7. ..o.o~m..
. - ..n\. .11. . . . . .. a s . AI . .. I O. I q . . y .
o .. r.-. o . ...ov.. . . . . u. q . 1 29" .» ..M. o 0.33.. ... ....V. a... 3.2.1.. . 2.0.1.... . 0.3... o. .. ...“..v. . .... b. vs....p.. ...t “.4. ... o. .... ‘ . *flrv m.up~.~.-.o..). ... 8.00- ... . h.- .stONHM.-.

II I lion. .....0. l...<?.1.:.1.._... .... ......l..... ’1......a..1....... . . :......0 ..:..... ..‘n. ...H.|... ......d 0...... 49.:

. .... ... ..u» 3 73...”..1..._!...........J1... ...“... I. on... «d. .W. .. t . o .1. .u ...: .. .. .- .. ... ...“..F... 53.3.03... ...-...... .... . .....C; .. .....m... . . .. ...... (.10. . o .. 0hc.a...—...
I u . o . v. . .v .. .. . I!- . fl . 91.: . . . 1‘ .a . a o . .

...0: .. .....1... ....... ......t: ......n...... .... .... .. ....w ..Qv... :12? .1.... . l ... w ......1 . ..............).u. .. «.1.... .v ._ ... (.-.?
. ...... t... ..V. ... Y..?:...; .. ...v.. .7:.... )1. on .

 

II‘! 1! . i

‘MWWWWxx um I

3
LIB R A R Y
Michigan State
University

1293 00656 7766

 

i

ABSTRACT

SUSPENSION OF CRITICAL JUDGMENT: AN
APPROACH TO ROLE-PLAYING INDUCED
ATTITUDE CHANGE

BY

Douglas Marion Little

This research tested a derivation from a new
formulation of the role-playing--attitude-change relation-
ship. Termed the suspension of critical judgment approach,
this new proposition points to the critical role of defen-
sive reactions in moderating the attitudinal consequences
of counterattitudinal advocacy. In attempting to demon-
strate the utility of this new position, the present
research focused on the general hypothesis that role-
playing is particularly effective in inducing attitude
change to the extent to which the act of portraying an
attitude position not one's own leads to a disengagement
or circumvention of the defensive reactions frequently
accompanying the reception of counterattitudinal material.

It was specifically hypothesized that both passive
exposure to a counterattitudinal communication and active
involvement in the expression of a counterattitudinal

communication produce a significant amount of attitude

Douglas Marion Little

change in the direction of the counterattitudinal position.
More importantly, it was hypothesized that active involve-
ment in the expression of a counterattitudinal communication
leads to a greater amount of attitude change than passive
exposure to the counterattitudinal communication when

there is no mention of the persuasive intent of the
counterattitudinal communication.

In the 2 x 4 factorial design used to test the
above hypotheses, all experimental subjects were asked to
read a 1,000-word essay arguing against an all-volunteer
army. One member of the pair of subjects present at each
session was asked to use the material to improvise a
counterattitudinal speech while the other member was
asked to outline the material. In an orthogonal manipu-
lation, one-fourth of the subjects were warned about the
persuasive intent of the essay before reading it, one-
fourth were warned after reading it, one-fourth were
warned before the dependent variable measures were dis-
tributed, and one-fourth of the subjects were not specifi-
cally warned about the persuasive intent of the essay. A
control group of subjects who merely filled out a question-
naire about national issues supplemented the design.

The first hypothesis concerning the effectiveness
of both active and passive methods of exposure to counter-
attitudinal material was confirmed, although not as

strongly as eXpected. The interaction predicted by the

Douglas Marion Little

second hypothesis was not confirmed. Not only was there no
tendency for the mention of persuasive intent manipulation
to reduce the superiority of active counterattitudinal
advocacy over passive exposure, there was a slight ten-
dency (.08) for passive exposure to lead to more attitude
change than exposure involving active participation.

A check on the manipulation of persuasive intent
revealed that it did not have the intended effects. It
was suggested that, since both an active and a passive
participant were present at each experimental session,
the role assignment procedures may have interacted with
the mention of persuasive intent manipulation to create
differential feelings of relief. Presumably, the feelings
of relief produced differential attitude change. It was
recommended that future role-playing research avoid designs
in which two subjects are present at an experimental
session where there is only one counterattitudinal per-
formance to be given. The relevance of certain dependent
variable measures for the attention, improvisation, and

satisfaction hypotheses was also discussed.

SUSPENSION OF CRITICAL JUDGMENT: AN
APPROACH TO ROLE-PLAYING INDUCED

ATTITUDE CHANGE

BY

Douglas Marion Little

A THESIS

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

MASTER OF ARTS

Department of Psychology

1971

 

,* /v"z7’
:7! /Z<”:*"

to the other members of the Little family—-
my mother, father, brother, grandmother,
and grandfather

ii

ACKNOWLEDGMENTS

I would like to thank the members of my committee
for providing a standard of excellence that will serve me
well throughout my career. Special thanks are due to my
chairman, Dr. Eugene Jacobson, for his patience and con-
tinued reassurances; to Dr. William Crano for his insights
about research design and aid in data analysis; and to
Dr. Lawrence Messé for his challenging questions about the
nature of defensive, belief-maintaining processes and his
advice about life in the academic world.

My close friends Bob Homant, Bob Ruh, and Karen
Sanderson also contributed invaluable advice and support
at critical points during the course of this research. I
can not thank them enough.

I am also grateful to the Cooperation/Conflict
Resolution Group of Michigan State University for provid-
ing access to their subject pool. Department Chairman,
Dr. Lawrence O'Kelly, and Mr. Roger Halley were extremely
helpful in granting and administering the funds used to

encourage students to participate in this research.

iii

TABLE OF CONTENTS

Chapter Page
I 0 INTRODUCTION 0 O O I O O O O O O O l
The Problem . . . . . . . . . l
Explorations of the Role-Playing-
Attitude Change Relationship . . . . 3
The Dissonance Hypothesis. . . . . 4
The Improvisation Hypothesis. . . . 23
The Biased Scanning Hypothesis . . . 29
The Attention Hypothesis . . . . . 32
The Satisfaction Hypothesis . . . . 33
The Effort Hypothesis . . . . . . 36
Defensive Reactions and the Role-Play—
ing--Attitude-Change Relationship . . 43
Evidence Concerning the Role of
Defensive Reactions . . . . . . 45
The Suspension of Critical Judgment
Approach. . . . . . . . . . 54
Hypotheses and Rationale . . . . . . 62
II 0 METHOD 0 O O O O O I O O O O O O 70
Overview . . . . . . . . . . . 70
Subjects . . . . . . . . . 71
Attitude Topic and Essay Material . . . 72
Procedure . . . . . . . . . . . 73
Manipulation of Role-Playing
variable 0 O I O O O O O O O 7 4
Manipulation of Persuasive Intent . . 76
Control Condition . . . . . . . 76
Dependent Measures . . . . . . . . 77
Attitudes Toward National Issues . . 77
Evaluation of the Improvised Talk . . 77

iv

Chapter

Perception of Essay and Its Source

Recall . . . . .

III. RESULTS . . . . . . .

Check on Premanipulation Equivalence of

Experimental Subjects.

Assessment of Experimental Manipulations

Type of Exposure . .

Mention of Persuasive Intent

Test of Hypotheses . .

Hypothesis 1 . . .
Hypothesis 2 . .

Supplementary Analyses .

The Attention Hypothesis

The Improvisation Hypothesis
The Satisfaction Hypothesis

IV. DISCUSSION . . . . . .
REFERENCES . . . . . . . .
APPENDICES

Appendix

A. Counterattitudinal Communication.

B. Postexperimental Questionnaire

Page

78
78

79
79
81

81
81

86

86
88

90
90
90
93
96

109

114

120

LIST OF TABLES

Table i Page

1. F Statistics and Associated Probability
Values from Analyses of Variance of
the Nonmanipulated Attitude Issues . . . 80

2. Means and Analysis of Variance of Subjects'
Self-Perceptions of Their Thoughts About
Persuasive Intent. . . . . . . . . 83

3. Means and Analysis of Variance of Subject's
Ratings of Fairness of the Essay. . . . 85

4. Means of Attitude Toward an All-Volunteer
Army for Experimental and Control
Groups 0 O O O O O O O O O O O 8 8

5. Analysis of Variance of Attitude Toward an
All-Volunteer Army Scores for the
Experimental Subjects . . . . . . . 89

6. Means and Analysis of Variance for Recall
Scores for Experimental Subjects. . . . 91

7. Correlation Between Attitude Toward an All-
Volunteer Army and Length and Rated
Overall Persuasiveness of Performance
for Active Participation Subjects . . . 93

8. Correlation Between Attitude Toward an All-
Volunteer Army and Self-Rating of
Performance for Active Participation
Subjects. . . . . . . . . . . . 95

vi

CHAPTER I

INTRODUCTION

The Problem

 

It seems an understatement to suggest that the
area of counterattitudinal advocacy is controversial.
Attempts to account for the attitude change effects
accompanying counterattitudinal advocacy have provided
little closure, in spite of nearly 20 years of research.
To be sure, a relatively large number of mediating
mechanisms have been proposed and investigated. However,
about the only conclusion the student of persuasion can
have much confidence in is that, under an apparently
broad range of conditions, role—playing procedures lead
to a significant amount of attitude change in the
direction of the publicly portrayed position. Extant
explanations of the role-playing effect remain largely
problematic.

A reappraisal of the findings and conclusions from
various studies in the role-playing area provides a basis
for yet another explanatory prOposition, but one which may

serve to clarify and integrate the theory and research in

 

9’

..‘v

Au.-

.3

pm

I "i

this area of counterattitudinal advocacy. Generally, this
new position, to be termed the "suspension of.critical
judgment" approach, states that the attitude change
effects associated with the public expression of an
attitude-discrepant point of view are primarily a

function of the manner in which defensive reactions such
as counterargumentation and source and communication
derogation are moderated by role portrayal.

The major purpose of the present endeavor is an
elaboration and testing of a new formulation designed to
account for role-playing induced attitude change. The
discussion begins with a selective review of evidence from
investigations of the influence of counterattitudinal
advocacy on private attitudes and of the theoretical
explanations proposed to account for this evidence. An
effort is made to support the rather nonspecific claim
made above that current explanations of the role-playing
effect are inadequate or unsatisfactory. Next, data and
speculation are presented which suggest that psychological
resistance and defensive reactions are critical factors
in role-playing--attitude-change dynamics. This review
provides the background for a more detailed and thorough
discussion of the new formulation being presented in this
thesis. Finally, hypotheses based on one of the impli-

cations of this formulation are presented.

 

(I)
(I!

'i'

n!

...

‘4

I]!

-r-

Explorations of the Role-Playing--
Attitude-Change RelatiOnSHip

 

For the present discussion the focus of interest
is on those behaviors which may be characterized as a
person's involvement in presenting an attitude-discrepant
position as though it were the person's own position.
Several reviewers have ably recorded the rather large
body of research literature which suggests that under a
wide range of conditions this active involvement in the
processing of counterattitudinal material leads to a
marked amount of attitude change in the direction of the
public expression (Collins, 1970; Elms, 1967; Insko, 1967;
McGuire, 1966, 1969(a); Sears & Abeles, 1969).

Even more interesting and informative than the
findings of the general efficacy of counterattitudinal
advocacy are the data which indicate that exposure to
information by means of active participation in the
expression of counterattitudinal material leads to a
greater amount of attitude change in the direction of
the publicly portrayed position than does a more passive
form of exposure to the counterattitudinal material (Cul-
bertson, 1957; Harvey & Beverly, 1961; Janis & King, 1954;
Janis & Mann, 1965; King & Janis, 1956; Scott, 1957). The
enhanced effectiveness of role-playing procedures has also
been reported in terms of the temporal persistence of the
attitude-change effects (Elms, 1966; Mann & Janis, 1968;

Watts, 1967).

The following review is directed toward an
appraisal of the various propositions which have been
put forth to account for the attitude-change effects
referred to above. The major explanations to be covered
include dissonance, improvisation, and biased scanning,
although the satisfaction, attention, and effort hypothe-
ses will also be considered. In introducing another con-
cept or mechanism there is a certain need for justifi-
cation, and it is to this need that the present review
is primarily directed. Hopefully, this discussion will
clarify the inadequacies and shortcomings of current
theoretical rationales for the so-called role-playing
effect and, at the same time, begin to suggest the need
for and the elemental aspects of a new explanatory propo—
sition.

The dissonance hypothesis. Dissonance theory can

 

easily be seen as the most dominant force in the area of
counterattitudinal advocacy. The classic version of the

theory as expressed by Festinger (1957) in A Theory of

 

Cognitive Dissonance contains rather uncomplicated pre-
dictions about the cognitive consequences of counter-
attitudinal behavior. Basically, the theory indicates
that (1) if a person has engaged in attitude-discrepant
behavior and (2) if the person feels or perceives that
the behavior does not follow from or is inconsistent

With.his attitudes or beliefs with respect to the matter,

then the person will experience cognitive dissonance.
This state of cognitive dissonance is presumed to be a
psychologically aversive one which the person will be
motivated to reduce or e1iminate--the more so the higher
the level of dissonance. One of the several ways in
which this dissonance can be resolved is by an alteration
of the attitude in question so that the attitude becomes
consistent with the counterattitudinal act.

Thus, Festinger would explain the attitude change
resulting from counterattitudinal advocacy as due to dis-
sonance reduction. An additional feature of dissonance
theory concerns the nature of the pressures used to induce
the counterattitudinal behavior. It was postulated that
the greater the justification (whether monetary reward,
threat of physical harm, or otherwise) for the discrepant
act, the less dissonance produced and consequently, the
less the need for attitude change. Furthermore, the
level of inducement just sufficient to elicit the incon-
sistent act should yield the maximum amount of dissonance.
That is to say, in a role-playing situation an inverse
relationship should exist between justification and atti-
tude change. To take reward for an example, it is pre-
dicted that the more a person is paid for saying something
he does not believe, the less he should change his attitude

in the direction of the public performance.

The first dissonance derived investigation of
counterattitudinal advocacy was the, now classic, study
by Festinger and Carlsmith (1959) which examined the
influence of the reward variable on the cognitive
reactions to forced compliance. In that experiment
subjects were engaged for approximately an hour in very
repetitive and monotonous tasks and were then led to
believe the experiment was over. At this point, experi-
mental subjects were induced (by the promise of either
$1 to $20) to help out the experimenter in his alleged
study of the effects of expectations on performance. The
subjects were asked to misrepresent the nature of the
study they had just taken part in to a fellow student by
indicating that the dull experiment had actually been
interesting, enjoyable, lots of fun, intriguing, and
exciting. Control subjects took part in the boring tasks
but were not asked to deceive another student. Subse-
quently, in a study divorced from the one in which forced
compliance had been obtained, the subjects expressed their
feelings about the various aspects of the previous experi-
ment. The results were consistent with the dissonance
theory predictions. Subjects paid $1 found the experiment
significantly more interesting and enjoyable than either
control subjects or subjects paid $20. The latter two
groups were not significantly different on the mentioned

attitude dimension.

Additional, somewhat dramatic, evidence in support
of the dissonance theory predictions was collected by
Cohen (1962). It seems that in 1959, the annual spring
"riot" at Yale had been met by an unexpected and massive
retaliation by the New Haven police department. Since
this annual venture had been traditionally a more or less
harmless, aimless affair, the student sentiment was over-
whelmingly against the police action. Shortly after the
riot, subjects were approached by a person who presented
himself as a fellow student working for the Institute for
Human Relations. The subjects were asked, for varying
degrees of reward ($.50, $1, $5, and $10), to write an
essay in favor of the action taken by the New Haven police.
Under the guise of getting relevant arguments on both sides
of the issue, subjects were asked to write "the strongest,
most forceful, most creative and thoughtful essay you can,
unequivocally against your own position and in favor of
the police side of the riots (p. 75)."

An analysis of the attitude data revealed, as
predicted, a significant linear trend; most change was
demonstrated by the subjects in the $.50 condition and
least change was shown in the $10 condition. The $10,
$5, and control groups were not significantly different
in terms of attitude toward the New Haven police action,
while the $.50 condition differed significantly from all

the other conditions. The $1 subjects differed

significantly from subjects in all other conditions but
those in the $5 condition. It can be seen that in both
studies the more a subject was paid for taking a counter-
attitudinal position, the less he changed his attitude
in the direction of the publicly proclaimed position, and
the dissonance theory predictions were confirmed.

The studies by Festinger and Carlsmith (1959) and
Cohen (1962) have generated a great deal of critical
analysis and subsequent research. Basically, the criti-
cisms of the classical dissonance theory exPlanation of the

role-playing effect can be placed into two main categories.

 

First of all, there are a set of objections which suggest
that the data which apparently support the dissonance pre-
dictions can be more appropriately and accurately inter-
preted in terms of some alternative proposition. In the
second category are a number of suggestions which, in
general, can be taken to say that the original dissonance
theory statements were over extended and that the range of
behavioral situations to which the theory can address
itself are much more circumscribed than initially antici-
pated.

A number of observers, most notably Chapanis and
Chapanis (1964), Rosenberg (1965) and Janis (Janis & Gil-
more, 1965; Elms & Janis, 1965), have warned that the
dissonance manipulations may have created internal states

other than or in addition to the ones intended thus

permitting alternative interpretations of the data. All
have suggested that the monetary reward manipulations may
have created, in some form or another, interfering psy-
chological reactions. Chapanis and Chapanis (1964)
specifically suggested that the high reward in the
Festinger and Carlsmith (1959) study was so incredible
or implausible that it produced wariness and suspicion
in the subjects. This resistance was held to be more
prevalent in the high reward condition and, as a result,
responsible for the inverse relationship between reward
and attitude change.

Janis and Gilmore (1965) in reinterpreting the
Festinger and Carlsmith (1959) study, spoke of "some
degree of suspicious wariness about being exploited by
the experimenter or some degree of guilt about being
'bought' to lie to a fellow student [p. 18]" and sug-
gested that this wariness or guilt could account for the
observed attitude change effects. Janis and Gilmore
(1965) assumed that the sponsor in the Festinger and
Carlsmith (1959) study was negatively perceived and that
this led the subjects in the $20 reward condition to
experience more "suspiciousness, guilt, or other negative
feelings [p. 26]" than the subjects in the $1 reward
condition. Presumably the relationship between reward
and attitude change would be direct under favorable

sponsorship.

10

In their own study, Janis and Gilmore (1965)
manipulated reward ($1 and $20), sponsorship (positive-
public welfare and negative-commercial), and condition
of role-playing (mere commitment and active participation).
While the reward—sponsorship interaction failed to materi-
alize in the Janis and Gilmore study (1965), the sponsor-
ship manipulation led to attitude change effects which
were embarrassing to dissonance theory. Those role-
playing subjects in the favorable sponsorship condition
demonstrated more attitude change in the direction of
their role performance than those in the unfavorable
sponsorship condition. Elms and Janis (1965) investigated
the same conceptual variables as did Janis and Gilmore
(1965). They found a significant main effect for sponsor—
ship and a near significant interaction between the
sponsorship and reward variables for role-playing sub-
jects. The differences for non-role-playing subjects
were nonsignificant and tended to indicate more change
under high reward conditions. In sum, the data of these
two studies provide suggestive evidence that sponsorship
conditions can moderate the effects of incentive and
support an alternate interpretation of the mentioned
dissonance theory studies.

While Chapanis and Chapanis (1964) and Janis and
Gilmore (1965) focused their criticisms of dissonance

theory on the Festinger and Carlsmith (1959) study,

11

Rosenberg (1965) directed his attention primarily to the
study by Cohen (1962). As in the other critiques
reviewed above, Rosenberg also suggested that the
intended dissonance manipulation may have had unintended
effects leading to a false confirmation of the theory's
predictions. In a reexamination of Cohen's procedures,
Rosenberg (1965) detailed a number of factors which could
have led to suspicion and hostility. This suspicion and
hostility was assumed to have led to "evaluation appre-
hension" and/or negative affect toward the experimenter--
the more so the higher the level of reward. Thus, Rosen-
berg suggested that as the level of reward increased in
Cohen's study there was a greater tendency on the part of
the subjects to resist showing any influence which could
be linked to exposure to the counterattitudinal material
and, as a result, that Cohen found an inverse relationship
between reward for role-performance and attitude change.
Note that Rosenberg (1965) is referring to a
conscious effort on the part of the subjects to cover up
actual influence by the persuasive materia1--a purposeful
distortion of attitude to achieve a desired evaluation
or to get back at the experimenter. Rosenberg hypothe—
sized that effects of negative affect toward the experi-
menter or of evaluation apprehension could be eliminated
if the counterattitudinal performance phase of the experi-

ment were markedly separated from the attitude measurement

12

phase. With the two phases separated, a positive relation-
ship between reward and attitude change was predicted. It
is not entirely clear whether Chapanis and Chapanis (1964)
were hypothesizing a mechanism like the one proposed by
Rosenberg (1965). Janis and Gilmore (1965), however,
seemed to have been suggesting that the effect of the
suspicion and hostility had a direct effect and actually
impeded attitude change. Rosenberg (1965), on the other
hand, felt that if the appropriate measurement conditions
could be established, then the actually changed attitudes
would be revealed.

To test his criticisms, Rosenberg (1965) conducted
an altered replication of the Cohen (1962) study. To
accomplish the necessary separation, subjects who reported
to the experiment were told that they would be kept wait-
ing for 15 or 20 minutes and that if they wanted to they
could report to "another little experiment that some
graduate student in education is doing [p. 33]." Subjects
who reported to the graduate student in education were
asked (for $.50, $1, or $5) to write an essay about why
the Ohio State football team should not be allowed to
participate in the Rose Bowl. Upon returning to the
experiment to which they had initially reported, the
subjects responded to a number of attitude items includ-
ing a critical item about the Rose Bowl as part of a sort

of local Gallup poll on student attitudes concerning

13

university issues. A control group was included in which
subjects were merely asked to respond to the student poll.

An analysis of the data confirmed the predicted
positive relationship between reward and attitude change.
While the $.50 and $1 conditions were not significantly
different in attitude toward the Rose Bowl policy, they
were each significantly more favorable toward the
advocated Rose Bowl policy than the control group and
significantly less favorable than the $5 condition.
Generally, Rosenberg interpreted the results as a vali-
dation of his concept of evaluation apprehension and his
criticisms of the dissonance theory research.

One conclusion that can be drawn from the above
discussion is that the data apparently supportive of
dissonance theory predictions can be adequately explained
in terms of negative reactions to the dissonance manipu-
lations--reactions such as hostility and resentment and
suspicion. Unfortunately, in neither the Festinger and
Carlsmith (1959) study nor the Cohen (1962) study were
there any direct measures of the subjects' reactions to
the manipulations. There were no measures of what the
manipulations meant to the subjects and thus there was
no direct evidence to support the claims of the critics.
In addition, except for a standard, rather general,
manipulation check item employed by Janis and Gilmore

(1965), the critics have failed to produce direct evidence

14

of the interfering responses which are supposed to be able
to account for the attitude change effects. In spite of
the observed lack of direct evidence, the interfering
response criticism of the dissonance theory explanation
of the role-playing effect seems quite logical and com-
pelling. Moreover, explicit evidence does exist which
suggests that the interfering response analysis is a
cogent one. Somewhat surprisingly, this evidence comes
from two studies which have generally been taken as sup-
port for dissonance theory predictions.

In studies by Kelman (1953) and Cohen, Brehm, and
Fleming (1958) high justification conditions led to less
attitude change than low justification conditions, as
predicted by dissonance theory. However in both studies
the subjects in high justification conditions expressed
more negative reactions to the manipulations than did
subjects in low justification conditions. In the Cohen
§£_31. (1958) study, it was reported that subjects given
many reasons to justify their attitude—discrepant behavior
were less "self-motivated" and showed more "interfering
- responses" (these measures were not described) than sub-
jects given only a request to engage in the counteratti-
tudinal behavior. In the Kelman (1953) study, those sub-

jects offered the book Huckleberry Finn, and an opportunity

 

to get out of class to watch the movie, "Huckleberry Finn,"

for writing essays against a type of comic book which they

15

favored, demonstrated significantly more interfering
responses than did subjects given less justification.
Interfering responses were measured by a postexperimental
questionnaire which asked the seventh-grade students such
things as whether they attempted to or actually did think
of arguments contrary to the ones they had written.

Although certain aspects of these studies (subject
self-selection in the Kelman (1953) study and only border-
line significance after eliminating over half of the
original subjects in the Cohen §E_21. (1958) study) force
moderation in the conclusions that can be drawn from them,
the kind of data provided is strongly supportive of the
interfering responses criticism. This evidence takes on
increased importance when it is considered that data on
the subjects' reactions to the manipulations are exceed-
ingly rare in dissonance theory research (Chapanis &
Chapanis, 1964).

Other criticisms of the early dissonance theory
fonmulations have arisen from an attempt to reconcile the
findings of Rosenberg (1965), Janis and Gilmore (1965),
and Elms and Janis (1965) with dissonance theory derived
studies such as those by Festinger and Carlsmith (1959)
and Cohen (1962). Basically, the argument has been
reiterated, although in several different forms, that
the preliminary dissonance analysis of the role-playing
effect was too general and over-extended; that a more

complex and restricted set of propositions is necessary.

16

It is important to recall at this point that the
purpose of the present discussion is an appraisal of the
adequacy and utility of the dissonance theory explanation
of counterattitudinal advocacy. The purpose is not to
present an overall evaluation of all aspects of dissonance
theory, but merely to examine its applicability to the
restricted set of behaviors which is the focus of this
study. Thus there is no attempt to cover all modifications
and subsequent reformulations of the theory. The intent
in the remainder of this section is to demonstrate that
there is no consensus as to the domain of counter-
attitudinal behaviors to which dissonance theory is
most directly pertinent.

Carlsmith, Collins, and Helmreich (1966) have
introduced the possibility that cognitive dissonance has
several distinct forms and that the theoretical pre-
dictions which are appropriate for one type of dissonance
may not be appropriate for another. The study by Carl-
smith gt_al. (1966) was an attempt to reconcile the dis-
crepant findings mentioned above by suggesting that
dissonance theory predictions will be supported if the
counterattitudinal behavior involves a face-to-face con-
frontation, while so-called incentive theory predictions
will result in situations such as those involving anony-
mous essay writing. Using essentially the same pro-

cedures as used by Festinger and Carlsmith (1959), it

17

was discovered that the amount of money offered to adopt
a counterattitudinal position had different effects under
the two role-playing conditions. Under face-to-face
conditions, the more the reward given, the less the
attitude change; while role-playing by writing an anony-
mous essay led to more attitude change the more reward
offered. Although the data supported the hypothesized
relationships, the reformulation is found wanting when
applied to the essay writing subjects in the Cohen (1962)
study where the dissonance-theory predicted inverse
relationship between reward and attitude change was
obtained.

In a similar vein, Aronson (1966) cited the Carl-
smith gt_al. (1966) study as support for his claim that
the limiting conditions of dissonance theory are defined
by the commitment variable. Under high commitment con-
ditions, presumably those in which a person's behavior
is public (can be definitely associated with him, and is
directed to a person whom he believes is unaware of the
inauthenticity of his act), Aronson contends that dis-
sonance predictions will be appropriate. Under low
commitment conditions, incentive or reinforcement theory
predictions should obtain. While Helmreich and Collins
(1968) found evidence to support this analysis, they

review several other studies in the "Studies in Forced

18

Compliance" series which failed to find a dissonance
effect under high commitment conditions.

Linder, Cooper, and Jones (1967) discussed yet
another aspect of commitment--decision freedom. It was
suggested that dissonance predictions about the effect
of reward on attitude change could most appropriately be
made when the subject took the incentives into account
in making his choice about whether or not to engage in
the counterattitudinal behavior. On the other hand,
reward and attitude change were expected to be positively
related when the subject had already committed himself
prior to hearing about the incentives. Confirmation of
their hypothesis about the role of choice was obtained in
two different replications; one based on the Cohen (1962)
study and the other based on the Rosenberg (1965) study.
In both studies there was a dissonance effect in free
decision conditions and an incentive or reinforcement
effect under no-choice conditions. That both effects
were obtained without a separation of role-playing and
attitude measurement phases of the study was assumed to
discredit Rosenberg's criticisms of the dissonance
research.

Unfortunately, as was the case with the public-
private distinction made by Carlsmith et_al. (1966),
this rather convincing attempt to order the counter-

attitudinal advocacy literature is confronted with some

19

exceptions to the rule. Crano and Messé (1970) obtained

a dissonance effect under low freedom of choice conditions
similar to those in the Rosenberg study (1965). Further-
more, Janis and Gilmore (1965) found that for role-playing
subjects with a high degree of choice such as that in the
Cohen (1962) study, more attitude change occurred under
favorable sponsorship conditions than under unfavorable
sponsorship conditions. To summarize, in the Crano and
Messé study where incentive effects would be predicted by
the Linder, Cooper, and Jones (1967) analysis, both dis-
sonance and incentive effects were obtained, and in the
Janis and Gilmore study where a dissonance effect should
have occurred it did not.

For final consideration, another attempted reformu-
lation of the limiting conditions of the dissonance effect
will be discussed. Rosenberg (1966) has suggested that
dissonance theory is most applicable to those instances
of counterattitudinal behavior of a simple and limited
nature (not going to a movie that one wants to see, for
example). On the other hand, when a person's behavior
involves a "complex and extended performance, one that
leads to the development and elaboration of a new set of
cognitions [p. 144]," dissonance theory is not considered
to be particularly germain. The convenient terms
"counterattitudinal action" and "counterattitudinal

advocacy," respectively, were introduced to distinguish

20

between the two types of behavior. The category of
behaviors involving complex and extended activity was
further subdivided into those in which the task set was
one of "duplicity" and those in which the task was one
of "self-examination." It was assumed that a positive
relationship between magnitude of reward and attitude
would occur when performance was undertaken with a self-
examination set. When the task set was one involving
deception, higher levels of reward could lead to lower
levels of attitude change. Under the latter circum-
stances the data would lead to a spurious confirmation
of a dissonance prediction.

Unfortunately, Rosenberg (1966) did little to
clarify exactly how the change effects were mediated with
a duplicity set and merely spoke of insulation from the
change implications of the newly elaborated arguments.
Linder gt_21, have objected that Rosenberg's criteria
fail to explain properly the results of their two experi—
ments since in their studies no deception set was involved
and yet both inverse and direct relationships between
reward and attitude change were observed. However, an
examination of the procedures employed by Linder §E_§l.
(1967) does not provide unequivocal evidence that the
subjects could not have assumed a deception or duplicity
task set. Perhaps the problem lies in the fact that

Rosenberg (1966) has only presented a preliminary

21

classification system in which other relevant dimensions
must be provided by future research and critical analysis.
Rosenberg did acknowledge that other task sets and other
aspects of the staging of counterattitudinal advocacy
could be important in determining the extent to which

the elaborated position is internalized.

In general, the foregoing material does not pre-
sent a very favorable picture of dissonance theory's
ability to explain the role-playing effect. Several
studies (Carlsmith gt_al., 1966; Crano & Messé, 1970;

Elms & Janis, 1965; Linder gt_al., 1967; Rosenberg, 1965,
for example) have found a positive relationship between
incentive for counterattitudinal advocacy and attitude
change, and clearly indicate that dissonance theory is
incapable of accounting for all of the results of counter-
attitudinal performance. That is, some instances of
counterattitudinal behavior are outside the domain of
dissonance theory. Furthermore even if dissonance theory
is relevant to some counterattitudinal behaviors, the
conditions under which this is so are ambiguous--although
a strong case has been made for the centrality of the
commitment and choice variables.

Even more serious is the criticism that dissonance
theory is not pertinent to counterattitudinal advocacy at
all. Janis (Janis & Gilmore, 1965) and others have

suggested that mechanisms other than dissonance can

22

explain role-playing induced attitude change. Similarly,
Rosenberg (1965, 1966) has suggested that dissonance
theory has no great relevance for complex behaviors in
which a person elaborates a set of attitude-discrepant
arguments.

Admittedly the data and arguments from the
material above could be organized in ways less critical
of dissonance theory. A point to be made, however, is
that, since instances of counterattitudinal behavior
represent a very complex and nonhomogeneous set of
behaviors, a single rather simple mechanism such as
dissonance cannot be expected to account adequately for
the attitudinal consequences of all such situations.

Until recently dissonance theory has had a rather
inordinate influence on the investigation of counter-
attitudinal phenomena. Attention has been diverted from
an exploration of other aspects of attitude-discrepant
behavior. In part, the somewhat lengthy discussion
which has preceded was an attempt to counteract this
imbalance by asserting the legitimacy of and need for
other approaches to the phenomena in question.

Following is a brief review of some of the other,
less researched explanations of the role-playing effect.
The same general criticisms that were made concerning
dissonance theory are also applicable to each of these

other proposed mechanisms and as such will not be repeated

23

for each additional explanatory proposition. None can
explain all the data and none have adequately specified
their limits of applicability.

Another major source of perspectives on the role-
playing issue in addition to dissonance theory has been

the Yale Communication Research Program. In Communication

 

and Persuasion, Hovland, Janis, and Kelley (1953) outlined

 

a number of hypotheses to account for the attitude change
effects of "active participation." No doubt the improvi-
sation hypothesis and its derivative, the biased scanning
hypothesis, are the explanations most associated with the
Yale group. Some form of one of these two propositions
(often referred to as the incentive.theory position) must
surely be considered the strongest contender to the dis-
sonance occupied throne. After a brief consideration of
the improvisation and biased scanning hypotheses, expla-
nations in terms of satisfaction, attention, and effort
will be covered.

The improvisation hypothesis. The improvisation

 

hypothesis asserts that role-playing will produce attitude
change in direct proportion to the extent to which the
role player develops good quality arguments, examples,

and illustrations. The more a person elaborates on the
assigned position, the more his private opinion should
become consistent with the overt counterattitudinal act.

Thus, role-playing conditions leading to better quality

24

and quantity of performance should produce more attitude
change (Janis & King, 1954).

Both the improvisation hypothesis and the satis-
faction hypothesis (to be discussed later) arose from an
attempt to explain the data of the pioneering study by
Janis and King (1954). In that study, three attitude
topics were used and three subjects were scheduled for
each experimental session. Each subject in turn assumed
the role of active participant and delivered an informal
talk based on a prepared outline, while the two passive
participants merely read the prepared outline and listened
to the informal talk. On two of the three issues, active
participants changed their attitudes to a greater extent
than did passive controls and the conclusion was drawn
that "overt verbalization induced by role-playing tends
to augment the effectiveness of a persuasive communi-
cation [p. 218]."

Supplementary observations on the third issue
which failed to show a superiority of active participation
over passive exposure provided some suggestive leads
about the mechanisms underlying role-playing induced
attitude change. Compared to active role-players in the
other conditions, active role-players on this third
communication stayed closer to the prepared outline and
were less likely to develope new arguments or illustrating

material. This led Janis and King (1954) to hypothesize

25

that "the gain in role-playing may occur primarily because
the active participant tends to be impressed by his own
cogent, arguments, clarifying illustrations, and con-
vincing appeals which he is stimulated to think up in
order to do a good job of 'selling' the idea to others

[p. 218]." Evidence was also gathered which seemed to
indicate that role-players on this third issue were also
less satisfied with their performance. This will be

taken up later in the discussion of the satisfaction
hypothesis.

Support for this improvisation hypothesis was
provided by King and Janis (1956) who manipulated improvi-
sation and found significantly more attitude change in
the experimental condition requiring subjects to elaborate
on an outline of arguments than in the condition in which
subjects were asked to publicly read a speech verbatim.
Also consistent with the hypothesis was the finding by
Kelman (1953) that the differences between high and low
response restriction conditions, in terms of performance
ratings of overall quality and number of arguments,
paralleled condition differences in attitude change.

In contrast to the evidence just reviewed, the
vast majority of studies examining this issue have not
confirmed the improvisation hypothesis. In fact, two
of the main studies from the so-called "incentive" theory

orientation have failed to demonstrate that differences

26

in quality of performance are responsible for differences
in attitude change. Janis and Gilmore (1965) were sur-
prisingly uncommunicative about the finding that while
quality of counterattitudinal performance was significantly
affected by both "reward" and "sponsorship" variables,
attitude change was only significantly different between
sponsorship conditions. Elms and Janis (1965) failed to
find any significant differences in mean number of good
quality arguments produced among the role-playing con-
ditions.

An examination of essay characteristics in the
Rosenberg (1965) study further complicates the picture.
With respect to number of words per essay, it was found
that subjects in the $1 condition wrote significantly
longer essays than the $.50 group but were not significantly
different from the $5 subjects. In terms of attitude
change, the $.50 and $1 subjects were not significantly
different but both differed significantly from the $5
group. Ratings of basic "persuasiveness" do provide
suggestive evidence that role-playing effects are mediated
by quality of role-playing performance. When subjects
in the $.50 and $1 condition were combined and then
divided into high and low "persuasiveness" halves, there
was a significant tendency for those who wrote more highly
persuasive essays to demonstrate more attitude change.

However, when each condition ($.50, $1, $5) was examined

27

separately, the significant relationship between essay
quality and attitude change was not obtained. Other
analyses were reported which question the importance of
quality of performance in the change process. When low
reward subjects with low persuasiveness scores were
eliminated from the analysis so that the remaining sub-
jects had a slightly higher mean persuasiveness rating
than high reward subjects, high reward subjects still
showed a significantly greater amount of attitude change
than low reward subjects.

Many other studies have failed to support the
improvisation hypothesis: Carlsmith, Collins, & Helmreich
(1966); Crano & Messé (1970); Festinger & Carlsmith (1959);
Helmreich & Collins (1968); Holmes & Strickland (1970);
Linder, Cooper, & Jones (1967); Nel, Helmreich, & Aronson
(1969); Zimbardo (1965).

What can be inferred from the fact that there is
so little empirical support for the notion that quality
of improvisation is positively correlated with the degree
of attitude change? One conclusion, assumed by many
researchers, is that the improvisation hypothesis is
incorrect. But, this research question is so riddled
with conceptual and theoretical problems that it may yet
be too early to close the issue.

One consideration is that the emphasis may have

been misplaced; that the important factor is not objective ,

28

ratings of quality of performance but rather the role
player's self-evaluation of performance. As Steiner and
Darroch (1969) have pointed out, the major theoretical
positions on role-playing actually suggest that "subjects',
appraisals of their own performance are more appropriate
measures of quality [p. 312]." One implication of the
Steiner and Darroch discussion is that the improvisation
and satisfaction hypotheses have a degree of theoretical
interdependence.

Another problem with attempts to relate objective
measures of role performance to attitude change is that
usually the subject's ability is not taken into account.
Now, it seems likely that quality ratings of role per-
formance would be highly correlated with measures of
verbal intelligence. But if quality of role performance
is for the most part a function of verbal intelligence,
and verbal intelligence is complexly and nonmonotonically
related to attitude change (McGuire, 1968), then ratings
of role performance would not be expected to covary
directly with attitude change. More simply, it is
being suggested that verbal intelligence may be obscuring
the relationship between quality of improvisation and
attitude change. Perhaps the relationship between
quality of role performance and attitude change would
appear if the influence of intelligence was removed sta-

tistically using verbal intelligence as a covariate.

29

A much more basic issue involves the conceptu—
alization of improvisation. An observation by Brehm and
Cohen in 1962 is still quite apt today; they observed
”that the concept of improvisation is not yet well
defined either conceptually or empirically [p. 252]."
Extant formulations about the improvisation hypothesis
have not clearly specified the critical aspects of the
role-performance, although three basic categories of
improvisation have been specified: reformulation--a
restatement of the communication in one's own words;
thinking of new examples or illustrations, and thinking
of new arguments. While King and Janis (1956) belittled
the importance of reformulation and instead stressed
the inventive aSpects of improvisation, they presented
no data to support their argument and the whole issue
had gone virtually unmentioned since then. Clearly,
the adequate measurement of the improvisation variable
will, to an extent, depend upon the resolution of this
definitional quandry. The general conclusion to be
drawn about the improvisation hypothesis is that while
the evidence is not very supportive, there are enough
unanswered questions and unresolved issues to merit
its continued investigation.

The biased scanning hypothesis. In later work by

 

Janis (Janis & Gilmore, 1965; Elms & Janis, 1965) a two-

factor explanation known as the "biased scanning"

30

hypothesis was introduced. (For a rudimentary form of
this hypothesis see King & Janis, 1956, p. 184.) A
rather troublesome aspect of this proposition is that
it has been explicitly stated in two different forms
without a discussion of the points of correspondence and
noncorrespondence between the two different formulations.
In Janis and Gilmore (1965) biased scanning referred to
(l) thinking of good positive arguments and (2) suppressing
thoughts about negative arguments. Elms and Janis (1965)
defined biased scanning as
. . . (l) fulfilling the demands of the role-playing
task by recalling and inventing arguments that are
capable of functioning as positive incentives for
accepting a new attitude position, and (2) appraising
the recalled and improvised arguments with a psycho-
logical set that fosters open-minded cognitive explor-
ation of theirjpotential incentive value, rather than
a negativistic set of the type engendered by the

arousal of feelings of hostility, resentment, or
suspicion [p. 59].

 

It is not entirely clear what has been added by the second
definition. While the correspondence between the two can
be made, the second appears to be more general and subtle
differences do seem to exist.

A major problem, then, with the biased scanning
hypothesis is that it has been so ambiguously stated
that procedures for evaluating its validity and utility
are unclear. This problem was underscored in the Elms
and Janis (1965) study where the quality of essay writing
was assumed to reflect or be an index of the psychological

set fostering open-minded cognitive exploration--the

31

second factor of their two-factor definition of biased
scanning. Thus, when it was observed that there were no
significant differences in quality of performance in con-
trast to significant differences in attitude change among
role-players, Elms and Janis were led to conclude that
there was no evidence to support their notion that the
attitude change "was mediated by a corresponding increase
in biased scanning [p. 60]." However, it could be
objected that quality of performance measures are inade-
quate or inappropriate measures of an open-minded cognitive
set (Greenwald, 1969, has recently made this point).
Although Elms and Janis acknowledged that quality of per-
formance was only an indirect measure of the role-player's
cognitive set, they provided only superficial direction as
to how the hypothesis might be more adequately evaluated.
Thus, it can be seen that the biased scanning
hypothesis suffers from a lack of clarity and definitional
rigor. At present there is no direct evidence against
which its claims can be evaluated. Before the hypothesis
can be expected to contribute much to an understanding of
the mechanisms mediating the role-p1aying--attitude change
relationship, it must first receive more adequate con-
ceptual treatment. As an additional point, it is noted
that the improvisation hypothesis is included as a
central aspect of the biased scanning hypothesis--a
matter which only further compounds the problems of the

biased scanning hypothesis.

32

The attention hypothesis. Another of the expla-

 

nations discussed by Hovland, Janis, and Kelley (1953)

was the attention hypothesis. According to this propo-

sition, role-playing is particularly effective because,
. . . the ego-involving task of verbalizing a com-
munication to others probably induces greater
attention to the content, which may increase the
chances that one will think about it and be influ-
enced by it [p. 230].

While theoretically plausible, the attention
hypothesis has not been empirically supported. Janis
and King (1954) reasoned that if role-playing is effec-
tive because it leads to greater attention to the content
of a communication, then any other technique which evokes
increased attention should have a similar effect. In
their study, a supplementary control condition was added
so that, in addition to an active and a passive partici-
pant, another subject was asked to follow along with the
prepared outline used by the active participant and take
down the main arguments he presented.

Janis and King (1954) reported that, although the
supplementary control group of subjects took fairly com-
plete notes (which seemed to indicate a high degree of
attention), this note-taking control group displayed
about the same amount of attitude change as the passive
control group, and significantly less than the role-

playing group. As a result, it was concluded that vari-

ations in attention level probably were not crucial in

33

accounting for the results of the study. In a follow-up
to their first study, King and Janis (1956) provided
further evidence with respect to the attention hypothesis.
From a slightly different perspective, they reasoned that
if active participation leads to greater attention then
active participants should, on the basis of the heightened
learning efficiency, demonstrate greater recall than
passive participants. Consistent with the conclusion

of the first study, it was found that all experimental
conditions (improvisation, read, and passive control)
obtained approximately equal recall scores. Other
researchers (e.g., Zimbardo, 1965) have failed to find
differences in recall which parallel differences in atti—
tude change and, thus it appears that the role-playing
effect is not mediated by manipulated differences in
attention.

The satisfaction hypothesis. Another explanation,

 

the satisfaction hypothesis, contends that a role-player
will begin to internalize the publicly presented position
to the extent to which he feels he did a good job in por-
traying the role. These feelings of a job well done may
refer either to the way the performance was given (the

structure of the role-playing) or to the particular argu-
ments given (the content of the role-playing). Moreover,

this self-satisfaction may derive either directly from a

34

person's perception and evaluation of his own performance
or indirectly from the feedback given the person by other
role-observers.

As mentioned earlier, this hypothesis developed
from the Janis and King (1954) study. In that study, the
role-playing effect was obtained for only two of the three
conditions. Active participants on the issue which failed
to yield the predicted effect were significantly less
satisfied with their performance, per se, than were active
participants in the other conditions.

In the King and Janis study (1956) the satisfaction
hypothesis was further explored. Variations in role
requirements produced intended differences in satisfaction
with performance but the active participation condition
which maximized satisfaction did not induce the most
attitude change. A supplementary variation in the
"improvisation" condition consisted of varying the role-
performance feedback given to role-players. The manipu-
lation successfully induced different levels of satis-
faction with performance but did not affect the amount
of attitude change manifested. On the basis of the results
of the second study it was concluded that the amount of
attitude change produced by role-playing is not related
to the amount of satisfaction with performance.

Others have failed to find satisfaction with per-

formance differences corresponding to attitude change

35

differences. For instance, Helmreich and Collins (1968)
found evidence consistent with the findings of the King
and Janis study. Although there were significant dif-
ferences in attitude change between various conditions,
there were no significant differences in subjects' self-
ratings of how clear or sincere or persuasive they had
been.

Results in this area are somewhat contradictory
though, as other research (e.g., Scott, 1957; Wallace,
1966) seems to suggest that satisfaction g§n_have an
affect on the amount of attitude change produced by role-
playing. The two studies to be considered here can be
cited as support for the satisfaction hypothesis but must
rely on the assumption that positive feedback from
judges evaluating role-performance affected satisfaction
which, in turn, induced attitude change.

Scott (1957) arranged class debates in which each
debater was expressing a counterattitudinal position.
After the debate half of the subjects were arbitrarily
assigned as winners and half as losers, although subjects
were led to believe that the results reflected the
actual class opinion of who had given the best speech.
Results indicated that the winners changed their attitudes
significantly more (in the direction of the side debated)
than the losers (who showed a slight tendency to intensify

their original opinions).

36

Wallace (1966) also had subjects engage in a
debate. Naive subjects debated with an experimental
accomplice in the presence of two "judges." Some sub-
jects were rated better than the average college student
in terms of the content of their speech but only average
in terms of the way the speech was presented. Other sub-
jects were rated better than the average college student
in the way the speech was presented, but only average in
terms of the content presented. Yet another set of sub-
jects was given average ratings on both content and per-
formance. Debate opponents were given neutral ratings on
both dimensions. Data analysis revealed that subjects
rated above average on manner in which the speech was
given changed significantly more than the other two
groups.

The satisfaction hypothesis has yet to receive
systematic experimental treatment and future research
should be conducted to clarify the nature and the effect
of the self-satisfaction variable. At present, the sup-
port for the hypothesis is equivocal.

The effort hypothesis. The final explanatory

 

proposition to be considered is the effort hypothesis.
Zimbardo (1965) presented a derivation from dissonance
theory centered on the role of the effort involved in
expressing a counterattitudinal position. Basically,

it was his contention that the more effort exerted in

37

expressing a view inconsistent with one's own private
opinion, the more dissonance created and hence the more
attitude change in the direction of the role-played
position. Differences between active role-players and
passively exposed participants are thus assumed to be
due to the fact that the improvising subjects exerted
more effort rather than to "the 'cognitive-intellectual'
aspects of improvisation having to do with its content
or quality per se [p. 106]."

In a 2 X 2 factorial design, half of Zimbardo's
subjects improvised a speech from an outline of arguments,
while the other half read a prepared speech which con-
tained the same arguments along with appropriate examples.
Half of the subjects in each role-playing condition per-
formed under low effort conditions (slight delayed audi-
tory feedback of their performance) and half performed
under high effort conditions (a longer and highly dis-
tracting delayed auditory feedback). Perceived physical
effort was significantly greater for high than for low
effort participants but did not differ between subjects
who improvised and those who read the prepared speech.
Results concerning the net percentage of subjects chang-
ing their attitudes in the advocated direction paralleled
the effort ratings. More high effort subjects changed
toward the publicly expressed position than low effort

subjects; and there were no differences between improvising

38

and reading subjects. (Differences in magnitude of
attitude change did not reach conventional levels of
significance.) Zimbardo concluded that his dissonance
theory based prediction concerning effort was supported.

Even though the argument has some plausibility,
Zimbardo has not made a strong enough case that it is
effort that is creating the dissonance or that the
effort involved is substantial enough to lead to dis—
sonance and subsequently to attitude change. It would
seem that effort expenditure would lead to the arousal
of dissonance only if the effort expended were incom-
mensurate with that which was anticipated to be necessary
and appropriate for the particular goal or task. That
is to say, it is not effort by itself, but effort relative
to value of the goal which would determine the presence
and extent of dissonance created.

Furthermore, even if dissonance is created by the
effortfulness of the expression of the counterattitudinal
position, it is not clear why a change would be expected
on any but task relevant attitudes such as those about
the pleasantness or interestingness of the experimental
task or the value of the research. After all, the effort
is not an intrinsic part of the attitude issue, but it is
an intrinsic part of the particular experiment and the
specific task. (WOuld we expect the subject to also be

more favorably inclined toward such things as:

39

psychology, the day of the week the experimental session
was scheduled on, and experimenters with the color of

hair and eyes like the one administering the particular
effort manipulation?) What is being suggested is that the
effort-justification application of dissonance theory to
the area of counterattitudinal advocacy has not been done
critically enough. The effort justification argument is
far from obvious.

If the dissonance-effort explanation is rejected,
what ga§_explain the observed patterns between effort and
attitude change? As Zimbardo (1965) observed, there are
several alternative interpretations of his data. One
strong possibility is that the effort manipulation
affected the extent to which subjects were able to
think about negative or undesirable aspects of the
message or counterattitudinal act.

In a subsequent study, Zimbardo and Ebbesen (1970)
attempted to dismiss this "distraction from manipulative
intent" interpretation, as well as an interpretation based
on novelty. Subjects read prepared speeches under con-
ditions of white noise, delayed auditory feedback (DAF),
and normal auditory feedback. It was implied that the
white noise was as novel and distracting as DAF, but that
it required less effort than DAF. When it was observed
that the white noise condition led to less perceived

effort and less attitude change than the DAF condition,

40

it was concluded that the distraction and novelty inter-
pretations has been eliminated and the effort interpretation
supported. Unfortunately, Zimbardo and Ebbesen (1970) pre-
sented no evidence that white noise was equally "distract-
ing" as DAF and thus much of the force of their argument

is lost. In contrast to their conclusions, it is main-
tained that both studies (Zimbardo, 1965; Zimbardo &
Ebbesen, 1970) are consistent with the hypothesis that
conditions which inhibit defensive reactions such as
counterarguing, without drastically interfering with the
reception of communication content, will facilitate atti-
tude change. This latter hypothesis is a basic element

of the susPension of critical judgment approach and is

more fully discussed in a subsequent section.

There is one further aspect of the Zimbardo (1965)
study which should receive some attention because it
highlights a major methodological problem for research
attempting to pin down the mechanisms mediating role-
playing induced attitude change. The specific issue
concerns Zimbardo's rejection of the improvisation
hypothesis, partially on the basis of a failure to find
attitude change differences between "improvisation" and
"read" conditions. The crucial point to note is that
Zimbardo's conclusion must be tempered by the consider-
ation of the fact that the improvisation-read manipulation

was confounded with content differences. That is, the

41

absence of a difference in attitude change between improvi-
sation and read conditions may have been due to differences
in exposure to different communication content. In fact,
"read" subjects perceived their speech to be more logical
than did "improvisation" subjects and this may have
obscured the effects of improvisation.

That content differences can effect the relative
superiority of active role-playing techniques relative
to more passive-exposure techniques was clearly illus-
trated by Watts (1967). The Watts (1967) study is a
significant one for role-playing research and merits
further discussion since it most clearly points out the
hazards involved in drawing conclusions about role-playing
effectiveness when exposure and attention to the arguments
of a communication position are not equated among experi-
mental conditions.

Watts was interested in the relative persistence
of opinion change induced by "active" (improvising and
writing an essay) as compared with "passive" (reading and
underlining topic sentences) participation. In order to
avoid regression artifacts, Watts wanted to equate the
immediate mean opinion change for the two types of par-
ticipation. On the basis of several pilot studies, each
of the three persuasive messages to be used in the study
was altered until subjects who merely read a particular

essay demonstrated the same amount of opinion as those

42

subjects who (without being presented with new information)
wrote an essay on the topic. Results of the main study
confirmed the success of the pilot studies; immediate
attitude change for the active and passive participants
was essentially equal.

One observation to be drawn is that the superiority
of "active" over "passive" participation can be moderated
by varying what content subjects within each condition are
exposed to. Clearly, if differences in effectiveness
between active and passive participation can be inten-
tionally influenced by varying information exposure, it
is reasonable to assume that differences in effectiveness
can be unintentionally produced. That is, for example,
an experimenter may inadvertently manipulate the degree
of "superiority" of role-playing by his particular choice
of a standard essay to which all passive participants
are to be exposed. A central aspect of this issue is
that subjects in the Watts study differed simultaneously
on two major dimensions. First of all, active participants
thought of arguments and presented them while passive par-
ticipants did not and, secondly, active participants con-
sidered a different set of arguments than did passive par-
ticipants. The obvious drawback to this is that if two
groups of subjects differ simultaneously on two variables--
"type of participation" and "amount of information con-
sidered"--then unequivocal statements about which variable

leads to differences in attitude change are not possible.

43

Thus, if active and passive forms of participation
are to be compared with respect to their effectiveness in
inducing attitude change, it is desirable that all other
things including exposure to information be as equal as
possible. Of course, it is not possible to equate com-
pletely the informational content considered between
subjects, even for those within the same condition. And
then too, the interest in equating information exposure
will depend upon the research question of interest. But
the confounding of the type of participation and infor-
mation exposure variables is a crucial matter in research
comparing the effectiveness of role-playing with other
techniques of persuasion, and appropriate measures should
be taken to reduce differences in exposure as much as pos-
sible. In the present research, an effort was made to
equate exposure to arguments on the experimental issue
by having both an active and passive participant present
at each session. Both active and passive participants
read the same persuasive essay and the passive participant
then listened to arguments and illustrations presented by
the active participant.

Defensive Reactions and the Role-

Elaying--Attitude-Changg
ReIatIonSHip

 

 

 

The confrontation between the two major expla-
.nations, dissonance and improvisation, has not been as

Ifruitful or informative as it might have been hoped. In

44

general, the issue of the mechanisms underlying the role-
playing--attitude-change relationship is replete with
unresolved questions and ambiguities. Unfortunately, it
seems that the intensity of the dispute has tended to
limit the type of questions asked and the interpretations
made of the patterns in the data.

That is, not enough attention has been directed
to the analysis of the research.data with a view that
there are mechanisms other than the ones in dispute that
could more parsimoniously order the phenomena in question.
An examination of research in the role-playing area does
suggest that all possibilities have not been exhausted.
One particularly compelling explanation stresses the role
of psychological resistance in the attitude change process.
The following section will review evidence from extant
studies which suggests that defensive reactions (such as
communication derogation and counterargumentation, for
example) moderate the extent to which counterattitudinal
advocacy leads to attitude change. Basically the studies
can be divided into two categories. One set of studies
suggests that those role-playing conditions which produce
the lowest level of interfering responses, lead to the
greatest amount of attitude change (Cohen, Brehm, & Flem-
ing, 1958; Collins & Helmreich, 1970; Kelman, 1953; Mann,
1967). Another set of studies, while demonstrating the

same relationship, suggests even more. Studies by Elms

45

(1967), Greenwald (1969, 1970), and Janis and Mann (1965)
suggest that role-playing is effective because it actually
inhibits or prevents defensive reactions. The latter set
of studies implies more than mere covariation and suggests
something of the causal nature of the relationship between
role-playing and attitude change.

Evidence concerning the role of defensive reactions.
The Kelman (1953) study, previously discussed, was one of
the earliest studies on counterattitudinal advocacy which
presented evidence that the extent to which role-playing
leads to attitude change depends upon the manner in which
the role-playing procedures influence processes of psy-
chological resistance. It will be recalled that in that
study, low restriction subjects (those offered a low
probability of receiving a desirable reward) demonstrated
more attitude change (p = .07) in the direction of the
role-played position than did high restriction subjects
(those offered a high probability of receiving a desirable
reward). Of most importance for the present discussion
however, was the finding that attitude change was
inversely related to the presence of interfering reactions.
Low restriction subjects who had demonstrated more atti-
tude change also reported fewer interfering reactions
such as thoughts about counterarguments.

Another previously discussed study found a similar

pattern of results. Cohen et a1. (1958) induced male

46

undergraduates at Yale to write essays advocating the
unpopular policy of coeducation at Yale. Post hoc
analyses which eliminated subjects with extreme initial
scores revealed that subjects given many reasons to
justify their counterattitudinal behavior were less
"self-motivated" and showed more "interfering responses"
than did subjects given only a request to engage in the
counterattitudinal behavior and tended to change (p < .07)
their attitudes less.

Other evidence of the mentioned covariation between
defensive reactions and attitude change can be found in
the study by Collins and Helmreich (1970). That study
was one of several by various researchers which has
attempted to establish the conditions under which the
relationship between reward for counterattitudinal
behavior and attitude change is inverse and those under
which it is direct. Collins and Helmreich (1970) reasoned
that role-playing conditions focusing on the "consequences”
of the counterattitudinal act would be most likely to
arouse dissonance and support dissonance theory pre-
dictions, while conditions focusing on the "process" of
counterattitudinal advocacy would create a situation
where incentive predictions would be confirmed.

For a quinine tasting task, "process" subjects
were asked to think about and describe aspects of the
solutions which were actually pleasant and exotic--pre-

sumably emphasizing the truthfulness of their performance.

47

"Consequence" subjects were instructed to create essays
about the solutions that would help persuade others that
the solutions were not bitter tasting. Monetary reward
was also included as a manipulated variable; subjects
were offered either $.50 or $2.50 for their essays.
Contrary to the predictions, the only significant

result was a main effect for type of instructions.
Process instructions led to more attitude change than
consequence instructions. The important additional find-
ing was that consequence subjects tended to dissociate
themselves from the essay to a markedly greater degree
than did process instructions. Again, the evidence
suggests that the more the role-playing procedures
activate defensive reactions, the less attitude change
occurs.

A major conclusion of a study by Mann (1967) was
that "if the requirements or conditions of improvisation
arouse responses of resentment and guilt, then inter-
ference with positive attitude change may be expected
[p. 347]." The basis of this statement was a finding
that, for an emotional form of role-playing designed
to use shame to facilitate change, high-verbalization
female subjects responded with more resentment and less
attitude change than low-verbalization female subjects.
In fact there was a boomerang effect, high-verbalization

female subjects became even more polarized in their

48

initial position. (These effects were not observed for
male subjects under shame-emotional role-playing and,
furthermore there was no similar effect of resentment
with fear-emotional role-playing or cognitive role-playing
for either male or female subjects.)

Each of the four preceding studies has shown that
role-playing conditions which minimize defensive reactions
maximize attitude change. The four studies which follow
go beyond this and suggest that under certain circum-
stances role-playing induces attitude change because it
leads to an inhibition of defensive reactions.

The first study to be described in this second set
was by Janis and Mann (1965) and employed emotional role-
playing. This form of role-playing uses props and staging
devices to encourage subjects to become emotionally
involved in a realistic life-like situation. Janis and
Mann induced their subjects, moderate to heavy smokers,
to assume the role of a patient who was returning for a
third visit to her doctor because of a bad cough that was
not responding to treatment. During the course of several
scenes which were acted out, the patient was informed that
she had a form of malignant cancer, and that, while surgery
was essential, there was only a moderate chance for a
successful outcome. In a set of dialogues with the experi-
menter who was assuming the role of the doctor, subjects

discussed several matters including their thoughts and

49

feelings concerning the bad news and their knowledge about
the relationship between smoking and cancer. Control sub-
jects merely listened to one of the more dramatic and
emotional tape-recorded sessions of an actual subject.
Data analysis revealed that experimental subjects

changed their attitudes in a role-consistent direction to
a significantly greater extent than did control subjects
and, on a follow-up interview two weeks later, reported a
significantly greater decrease in cigarette consumption.
The authors' discussion singled out fear arousal as a
mediating factor in the obtained attitude and behavior
change. The most noteworthy aspect of the study in terms
of the present discussion was a comment by Janis and Mann
(1965) about the effectiveness of role-playing vis-a-vis
defensive processes. It was observed that

. . . the technique of emotional role-playing may

prove to be an exceptionally successful means of

arousing potentially adaptive fear reactions, break-

ing through the defensive facade that normally pre-

vents many people from taking account of their

personal vulnerability to objective sources of

danger [p. 90].
Clearly, the study provides dramatic, although suggestive,
evidence that role-playing has a tendency to disengage
defensive processes.

More recent evidence that the disengagement of

defensive mechanisms is a central aspect of the role-

playing effect is provided by Elms' (1966) study on the

influence of fantasy ability on attitude change produced

50

through role-playing. Empathic fantasy ability, defined
as a person's ability to vividly imagine that he holds

an attitude on a topic different from the one he actually
has, was linked to the increased effectiveness of active
as compared with passive forms of role-playing. For male
subjects playing the role of a non-smoker, the correlation
between empathic fantasy ability and attitude change on a
delayed postmeasure was .79. The corresponding cor-
relation for male subjects, who were exposed to the

same experimental essay as the role-players and who
witnessed the active participant's role performance, was
only .18. The difference between the two correlations was
highly significant. Other measures of fantasy ability
were correlated with attitude change but active and passive
participants did not differ significantly in terms of the
magnitude of correlation.

It was Elms' belief that empathic fantasy ability
would most likely be activated by the instructions and
expectations of a role-playing task with its "as if" char-
acteristics. Moreover, it was hypothesized that the
trait acts so that, in some manner or another, one's own
psychological defenses are circumvented, thus permitting
cognitive contact with the new attitude.

The finding of a high correlation of empathic
fantasy ability with attitude change for active role-

players together with the fact that both active and

51

passive role-players were exposed to virtually the same
information suggests that the crucial factor in role-playing
induced attitude change is not what information is pre-
sented but rather the manner in which the information is
received and processed. In line with the formulation
presented herein, the situational characteristics cor-
responding to the demands of a role-playing task and the
personality characteristics corresponding to empathic
fantasy ability can be conceptualized as having a sort
of functional equivalence with respect to their influence
on attitude change. The results found by Elms concerning
empathic fantasy ability strongly reinforce the notion
that role-playing is uniquely effective because it leads
to a lowering of the psychological resistance usually
associated with the counterattitudinal position.

Further and more direct evidence that role—playing
is effective because it leads to an inhibition of implicit
interfering responses was provided by Greenwald (1969).

In that study, subjects were led to believe that they
would write an essay on the issue of general-specialized
education; some were told that they would write in favor
of specialized education and others were told that they
would write in favor of general education.

In preparation for the expository writing exer-
cise, subjects were asked to evaluate the validity of a

set of statements relevant to the general-specialized

52

education issue. Half of the statements were in favor of
specialized education. Subjects expecting to advocate a
position consistent with their private opinions, evaluated
the arguments that supported their private opinion as
significantly more valid than the arguments opposing their
own private opinions. On the other hand, subjects expect-
ing to advocate a position inconsistent with their private
opinions reported that the arguments consistent with their
private opinions were virtually as valid as the arguments
inconsistent with their private opinions.

On the basis of an analysis of subjects' evalu-
ations of the statements, Greenwald (1969) concluded that
. . . counterattitudinal role-playing assignments
induce a disposition toward unbiased evaluation of
controversial information, in sharp contrast with
the biased (opinion-consistent) disposition of sub-
jects who expected to advocate their own opinions

[p. 383].
While subjects did not overtly portray a counterattitudinal
position, it was assumed that the opinion change following
counterattitudinal advocacy results from the unbiased
judgmental disposition activated by the role-playing
requirements.

In summary, counterattitudinal role-playing may be

uniquely effective because it succeeds in getting

the subject to give impartial evaluation to infor-

mation opposing his own opinion-~something he would

do rarely, if at all, under other circumstances

[p. 387].

As a limitation on the conclusions of the study,

Greenwald (1969) suggested that the role-player's impartial

53

disposition may be of no consequence for material about
which a person has already formed negative cognitive
responses. This assumption was confirmed in a sub-
sequent study (Greenwald, 1970). Quite simply, it was
demonstrated that the role-playing effect does not occur
when a person has an opportunity to evaluate and react
to (or consider and counterargue against) the counter-
attitudinal, role-structuring, material before he is
assigned the task of representing this counterattitudinal
position. However, when a person examines the counter-
attitudinal material aftg£_he has been assigned the
position to be publicly portrayed, the role-playing
effect does occur. That is, under the latter conditions,
persons actively involved in presenting an attitudinal
position discrepant from their own demonstrate more
attitude change than persons more passively exposed to
the counterattitudinal material.

In summary, Greenwald (1970) confirmed the finding
of his previous study that role assignment leads to an
impartial evaluation with respect to new information
relevant to the role to be played. More importantly,
since subjects did actually engage in essay writing,
Greenwald was able to present more direct evidence that
the unbiased disposition induced by role assignment is
a factor in the persuasive efficacy of role-playing pro—

cedures.

54

The suspension of critical jpdgment approach. In
spite of the meager amount of attention devoted to the
resistance aspects of the persuasion process, it can be
seen that a line of research is accumulating that suggests
that defensive reactions play a critical role in determin-
ing the extent of attitude change produced by counter-
attitudinal advocacy. Basically, this research suggests
that role-playing will be particularly effective in
changing attitudes to the extent to which the act of
portraying an attitude position not one's own leads to
a disepgagement or circumvention of the defensive

reactions often accompanying the reception of counter-

 

attitudinal material.

The first set of studies just reviewed indicates
that role-playing is more effective in changing attitudes
yhgp_there are less interfering responses. The second
set of studies suggests that role-playing is effective
because it reduces the incidence of interfering responses.
The nature of these interfering responses can be assumed
to include such things as judgmental distortions, counter-
argumentation, and communication and communicator dero-
gation, although their specification and confirmation
awaits further research. Similarly, the manner in which
these reactions are affected by the counterattitudinal

behavioral sequence remains to be established.

55

It is a matter of utmost importance, however, to
note that role-playing procedures have been shown under
certain conditions to lead to no appreciable change in
attitudes or to no more change than a more passive form
of exposure to counterattitudinal material. One of the
earliest studies in this research area, the one by Janis
and King (1954), only found role-playing superior to
passive exposure on two of the three attitude issues
studied. Conditions in numerous other experiments are
also consistent with the above claim. For example, the
$5 and the $10 reward subjects in the Cohen (1962) study
were not significantly different in attitude from an
unexposed control group. Other similar findings include
those for: the $20 reward subjects in the Festinger and
Carlsmith (1959) study; the subjects paid $.50 in the
essay writing condition and those paid $1.50 and $5 in
the face-to-face role-playing condition of the Carlsmith
gg_gl, (1966) study; and the debate losers in the studies
by Scott (1957, 1959). In the role-playing conditions
cited above and in experimental conditions in several
other unmentioned studies, role-playing subjects did not
demonstrate more attitude change than control subjects.

This suggests that role-playing procedures can
have a full continuum of effects--ranging from change
facilitating to change inhibiting. One means of con-
ceptualizing this effect would be to view the role-playing

procedures as moderating a person's evaluative disposition.

56

Perhaps it would be useful to postulate that each
person has some sort of screening mechanism for categor-
izing and evaluatively labelling information which is
being processed. The state of this psychological filter
would be expected to vary as a function of several con-
ditions internal and external to the individual. Under
conditions which have only just begun to be delineated
(and about which more will be said later), it is hypothe-
sized that the processing filter would operate so that
the information being processed is not labeled "alien,
to be defended against" and negative emotional responses
would not be attached to the material. Under these con-
ditions it would be said that the person has an unbiased
evaluative disposition. Under other conditions, such
as when the procedures force attention to the exploitative
or otherwise undesirable nature of the source or sponsor
of the counterattitudinal material, the filter would be
conceived as coating all information with the label
"alien, to be defended against" and attaching negative
emotional reactions to the material. It would be expected
that under these conditions the person would be refractory
to the new point of view, and the person would be
described as having a biased evaluative disposition.

To summarize, the susPension of critical judgment
approach asserts that the more the role-playing task or

situation inhibits or prevents those interfering responses

57

commonly associated with the encounter of attitude dis-
crepant material, the more attitude change in the
direction of the public statement can be expected to
occur. On the other hand, the more the role-playing

task or situation forces attention to the biased nature
of a communication or the undesirable aspects of the com-
municator, the more the effectiveness of role-playing
should be reduced. In fact, under some circumstances
active participation may lead to less change than passive
exposure because the role-playing more forcefully_high—
lights negative aspects of the position to be portrayed
or of the sponsor.

Since there is probably more interest in the role-
playing conditions which facilitate attitude change, it
would seem to be desirable to attempt a preliminary
specification of the conditions under which disengagement
of defensive reactions will mediate between role-playing
and attitude change. Perhaps an appropriate place to
begin is by noting that the effectiveness of role-playing
is hypothesized to be due to the fact that when a person
attempts to present counterattitudinal arguments on an
issue as though they were his own, he is focusing on the
counterattitudinal material and striving to adequately
fulfill the task requirements. This preoccupation with
t the authentic portrayal of a counterattitudinal position

is presumed to reduce significantly the opportunity for

58

such defensive reactions as counterargumentation, source
and communication derogation, judgmental distortions, and
denial or out-and-out rejection. The involvement in the
counterattitudinal performance, per se, then is assumed
to be the major aspect of role-playing which results in

a disengagement of defensive reactions. However, the
entire role-performance situation is a very complex set
of variables. This leads to the possibility that factors
other than involvement in the role task could also have
an effect on defensive reactions. Thus, under some cir-
cumstances certain variables in the role-playing complex
may be inhibiting defensive reactions while other variables
may be intensifying defensive reactions.

More concretely, the implication of the preceding
discussion is that when the role-playing situation forces
attention to the negative or undesirable aspects of the
sponsor of the role performance (or of the counter-
attitudinal performance itself, or the particular con-
tent issue) defensive reactions will tend to be aroused
and attitude change will tend to be impeded. In order for
the hypothesized relationship between disengagement of
defensive reactions and attitude change to obtain, the
role sponsor or role performance must not be so unpleasant
or repugnant that, on balance, defensive reactions are
intensified rather than reduced. Role-playing situations

which could be expected to activate interfering reactions

59

include: those in which the counterattitudinal behavior
is considered by the role-player to be morally repre-
hensible, those in which excessive force is used to
induce the counterattitudinal behavior, those in which
the sponsor of the counterattitudinal behavior is
blatantly insincere or exploitative, and those in

which an untrustworthy sponsor creates suspicions

about his covert intentions. One critical issue then,
is the extent to which the role-playing situation permits
the role-player to ignore or disregard negative aspects
of his counterattitudinal act or the sponsor of the act,
rather than forcing the role-player's attention to these
negative characteristics.

Another matter of obvious importance is the rela—
tive salience and ego-involvingness of the attitude issue.
The suspension of critical judgment proposition is clearly
most relevant to more involving issues with respect to
which defensive reactions are likely to be aroused. For
issues about which there are likely to be few if any
defensive reactions, counterattitudinal advocacy would
be expected to be no more effective than a passive form
of exposure to the counterattitudinal material.

Another important characteristic concerns the
extent to which the role-playing procedures permit cog-
nitive contact with counterattitudinal material. One

aspect of this characteristic is the length of the

60

performance. It is assumed that the counterattitudinal
performance must be of such duration that role-supporting
arguments can be processed with an unbiased evaluative
disposition. That is, the performance must be of such
length that an inhibition of defensive reactions could
make a difference between the effectiveness of active
and passive forms of exposure to counterattitudinal
material. Another aspect to this characteristic is the
extent to which role performance interferes with the
reception and/or retention of the processed counter-
attitudinal arguments. There must first of all be an
awareness of the content of the performance before the
proposed unbiased evaluative disposition can be a
determinant of attitude change.

Another somewhat related consideration involves
the extent to which the role-player has sufficient infor-
mation to enable him to fulfill the role requirements.
Before the disengagement of defensive reactions can be
an influential factor, it is necessary for the role-
player to express some counterattitudinal views. Role-
playing would be expected to be of little consequence
when the role-player has little knowledge about the issue
and/or counterattitudinal arguments. Under these latter
conditions the potential effect of diSengagement of
defensive reactions would not have an opportunity to

be actualized.

61

On the basis of a superficial appraisal, it might
be assumed that the proposition being presented herein is
a negative or incomplete statement which only explains
why change does not occur, and that some other mechanism
is necessary to explain why change does occur. As a
corrective for such a misinterpretation, it is noted
that the suSpension of critical judgment approach
assumes as a matter of definition that there is some
content to the counterattitudinal performance. That is,
it is assumed that in presenting an assigned position a
person will draw upon a pool of information to fulfill
the task requirements. (This pool of information will
of course vary from person to person, from time to time,
sand from issue to issue.) Role-playing induced change
occurs from reprocessing parts of this pool of information
with the cognitive dispositions activated by role per-
formance as described above.

It should be noted that it is assumed that dis-
sonance, satisfaction, attention, effort, improvisation,
and possibly other mechanisms can have some affect on the
attitudinal outcome of a role-playing experience. How-
ever, it is further assumed that, under the conditions
specified above, these various mechanisms will be of
only secondary importance relative to the effect role-

playing has in disengaging defensive reactions.

62

Research to verify the suspension of critical
judgment approach could begin at many places. The very
general objective of the present study is to demonstrate
that defensive reactions play a critical role in moderating
the attitude change effects of a role-playing experience.
This will be attempted by varying an aspect of the role-
playing situation which seems likely to effect differences

in the extent to which defensive reactions are aroused.

Hypotheses and Rationale

 

Hundreds, perhaps thousands, of studies have shown
that an experimenter can change the attitudes of subjects
by presenting the subjects with some form of a persuasive
communication. A much smaller body of research suggests
that, in general, actively involving a subject in express-
ing a counterattitudinal position will lead to more atti-
tude change than would occur by passively exposing the
subject to the counterattitudinal position. The main
purpose of the present research is a demonstration of
the utility and applicability of the "suspension of
critical judgment" approach in accounting for the par-
ticular effectiveness of role-playing in producing atti-
tude change. To accomplish this purpose, the present
study focuses on both active and passive forms of par-
ticipation. Generally, it is assumed that while pre-
senting a subject with a persuasive communication or

counterattitudinal material will lead to attitude change

63

in the direction advocated, actively involving the subject
in the presentation of the material will lead to even
greater change.

The preceding considerations lead to the following
hypothesis:

H1: Both passive exposure to a counterattitudinal com-
munication and active involvement in the expression of
a counterattitudinal communication produce a signifi-
cant amount of attitude change in the direction of the
counterattitudinal position.

Now, if the presence or absence of implicit inter-
fering responses are critical determinants of the cogni-
tive consequences of counterattitudinal advocacy, then
it should be possible, by manipulating their presence, to
affect the amount of attitude change produced. Furthermore,
it could be said that, if role-playing is particularly
effective because it reduces the presence of interfering
responses, then by introducing (by experimental manipu-
lation) interfering responses into the role situation, it
should be possible to decrease or eliminate the superiority
of active participation over passive exposure.

Janis and Gilmore (1965), in a comment consistent
with the above assumptions, suggested that the gain in
attitude change produced by role-playing would not be
likely to occur if there were negative incentives in
the role-play situation such as information about the

manipulative intentions of the sponsor. A related finding

64

was provided by Brock (1967) who asked subjects to list
their thoughts before receiving a counterattitudinal
message. Among several variables, persuasive intent
was manipulated and it was found that high persuasive
intent led to a higher level of counterargument pro-
duction.

Thus, under the circumstances of the present study,
it was assumed that implicit interfering responses would
be activated by explicitly mentioning the persuasive
intent of the author(s) of the communication used to
provide a basis for role-performance. Two factors added
extra credence to the anticipation of the efficacy of
the manipulation: first of all, the research was con-
ducted just after the resumption of classes following a
student strike over the first 0.8. invasion of Cambodia
and secondly, the subjects (male) were asked to argue
against an all-volunteer army and at the time of the
study, students were very much in favor of an all-
volunteer army.

This prediction was made with the recognition
that it has been demonstrated that suspiciousness of
intent and warning about the issue and direction of per-
suasive material do not necessarily diminish the extent
of attitude change. Excellent reviews by Papageorgis
(1968) and McGuire (1969b) have pointed out the complex

effects of these variables. The predictions in this

65

study are not about how mention of intent, in general,
influences attitude change. The question is, first of
all, whether the specific manipulations of this study
induced negative, potentially interfering psychological
reactions and, secondly, how this manipulation of inter-
fering reactions affected the relative difference in per-
suasive effectiveness between an active and passive form
of exposure to counterattitudinal material.

To the extent to which interfering responses are
produced by the mention of persuasive intent, the superi-
ority of active role-playing over passive exposure should
be reduced. This hypothesis is more formally presented as:
H2: Active involvement in the expression of a counter-

attitudinal communication leads to a greater amount of
attitude change than passive exposure to the counter-
attitudinal communication when there is no mention of
the persuasive intent of the counterattitudinal com-
mun1cat1on.

In the current research, all subjects were pre-
sented with a five-page 1,000-word essay which was to
provide a common background for the role-performance.

The basic design of the present study required role-
players to read over a counterattitudinal communication
(which served to structure the role performance), then to
use the information to prepare an informal talk, and
finally to deliver the talk. A question arises as to

which point in the procedure to introduce a comment about

the manipulative intent of the source of the communication.

66

In order to obtain a clearer meaning of the
manipulation of intent to persuade and to gather infor-
mation about the crucial time parameters in the change
process, it was decided to mention the persuasive intent
at one of three different points in the procedure: before
the subject initially reads the communication, after the
subject reads the communication, or after the subject
delivers (hears) the informal talk and just before atti-
tude measurement. For one set of subjects there was, of
course, no explicit mention of persuasive intent.

It is assumed that the earlier that persuasive
intent is mentioned, the more interfering responses are
created in the interval preceding and during role per-
formance; and consequently the more equal in persuasive
effect are active and passive forms of participation.
Mention of persuasive intent would be expected to have
less of an inhibiting effect if given after preparation
for and delivery of a counterattitudinal performance.
The second hypothesis then predicts an interaction between
type of participation and mention of persuasive intent.
It is assumed that active participation will be at least
as effective as passive exposure in all mention of
intent conditions; although it is uncertain whether the
main effect of type of participation will be obscured by

the interaction mentioned.

67

A number of explanations of the role-playing effect
have been considered: (a) dissonance, lb) improvisation,
(c) biased scanning, (d) attention, (e) satisfaction,

(f) effort, and (9) suspension of critical judgment.
Although the present research is primarily an attempt

to validate the suspension of critical judgment approach,
evidence relative to several of the other alternatives
will be collected.

Recall data were gathered in order to examine the
attention hypothesis. Although the "effort" hypothesis
was not considered particularly compelling when applied
to attitude change concerning nontask related arguments,
role-playing and non-role-playing conditions were equated
as nearly as possible in terms of effort required to ful-
fill the task requirements. To evaluate the satisfaction
hypothesis, self-ratings of role performance can be com-
pared to see if they correspond to condition differences
in attitude change.

The dissonance theory explanation may or may not
have been relevant to the subjects' behaviors in the
present study. A number of studies have suggested that
"choice" seems to be an essential element in the arousal
of dissonance (e.g., Linder g£_al., 1967). That is, it
is assumed that counterattitudinal advocacy will arouse
cognitive dissonance only if a person is given a choice

of whether or not to engage in the attitude-discrepant

68

behavior. If choice is a necessary condition for the
arousal of dissonance, and therefore a necessary con-
dition for testing theory-related predictions, then the
present study did not establish adequate conditions for
testing dissonance predictions. Subjects committed them-
selves over the telephone to take part in an experiment
looking into the ways in which people organize their
thoughts. In addition, before engaging in any issue
related behaviors or receiving any information about
the experimental topic, subjects were paid and signed
receipts. During the experiment, active and passive
roles were arbitrarily assigned and the subjects were
not queried about their desire to continue with the study.

If, however, dissonance was aroused by the manipu-
lation in the current study, then predictions from the
theory are possible. The clearest prediction that can be
made is that active participation will lead to greater
attitude change than passive exposure. Subjects in active
participation conditions should experience more dissonance
than passive exposure subjects since they would have the
dissonance produced by the attitude-discrepant role per-
formance in addition to the dissonance produced by the
counterattitudinal communication.

Predictions about the effect of the mention of
persuasive intent are uncertain. It seems most likely

that mentioning the persuasive intent would increase the

69

dissonance and consequently, the amount of attitude
change. Thus, dissonance theory would predict that the
mention of persuasive intent would lead to an increase in
the amount of attitude change produced by role-playing--a
prediction contrary to the one made by the suspension of

critical judgment approach.

CHAPTER I I
METHOD

Overview

An after-only design was used to examine the effect
of role-playing and mention of persuasive intent on atti-
tude change. Experimental subjects were exposed to a
1,000-word essay arguing against an all-volunteer army.
Half of the subjects then used the material to improvise
a counterattitudinal speech, while the other half outlined
the essay material. Two subjects were scheduled for each
session. The two role-playing conditions were randomly
assigned so that one member of the pair was assigned to
improvise a "talk" while the other member was asked to
outline the material. After the role playing, the effect
of the two independent variables on several dependent
variables was measured.

The role-playing manipulation was crossed with a
manipulation of mention of persuasive intent. One-fourth
of the subjects were warned about the intent of the essay
before reading it; one-fourth of the subjects were warned

after reading the essay; another fourth were warned before

70

71

the dependent variable measures were distributed; and
finally, for one-fourth of the subjects the persuasive
intent of the essay was not mentioned. Supplementing
this 2 x 4 design was a control group which was not
exposed to any of the experimental procedures but merely

filled out the attitude rating scales.

Subjects

One hundred forty-nine undergraduates at Michigan
State University were recruited from a larger pool of
subjects who had responded to a school-newspaper adver-
tisement offering to pay subjects for participating in
motivational research. Subjects were contacted by tele-
phone and given a chance to earn $2 for taking part in an
experiment "looking into the ways in which people organize
their thoughts." After the subject had agreed to partici-
pate and had made an appointment for the experiment, he
was informed that, because of scheduling requirements
and the fact that two people were taking part in each
session, it was important to be prompt. While a small
percentage of the subjects had never taken part in psy-
chological research before, the mean number of experiments
per subject was 2.77. Sixteen subjects were randomly
assigned to each of eight experimental conditions and 21

were assigned to the control group condition.

72

Attitude Topic and Essay Material

Three weeks prior to the first session of the
experiment students in several undergraduate classes at
Michigan State University were surveyed about their atti-
tudes on a number of topics of national interest. The
topic of an all-volunteer army was selected for manipu-
lation because there was a high degree of consensus and
polarization on the issue. Approximately 80% of the sub-
jects surveyed checked a point on the side of the neutral
point which favored an all-volunteer army. On an 11-
point scale most respondents checked one of the first
three choice points.

The fact that there was a great deal of consensus
in this student population on a position strongly in
favor of an all-volunteer army made the topic highly
desirable for a number of reasons. First of all, since
subjects were randomly selected for the experiment with-
out knowledge of their attitude on the issue, there was a
high probability that a position against an all-volunteer
army would be counterattitudinal for most subjects.
Secondly, the extremity of the majority responses rein-
forced the assumption that this was a highly ego-involving
issue. Finally, the attitudinal consensus was desirable
because the greater the consensus in initial attitude,
the less individual differences in initial attitude con-

tribute to the total variability of attitude change

73

scores and therefore the greater the likelihood of detect-
ing the influence of the independent variables.

To provide material for the role-playing experience
an essay was developed which took a position that an all-
volunteer army is an undesirable way to maintain an army.
The 1,000-word essay entitled "Concerning an All-Volunteer
Army” was written in a clear and highly organized style.
The first page noted that the issue of an all-volunteer
army was one of widespread concern. The second page made
a case that the United States did need an army. The
remainder of the essay contained four arguments suggest-
ing that an all-volunteer army is not the answer to
American military manpower needs: the cost would be
too high; civilian concern about military involvement
would wane, the military institution would become more
inflexible, and the poor would have to carry the burdens
of the society for the rich. The source of the essay was
unspecified, §_merely indicated that the material was
taken from a recent campus position paper. (See Appendix A

for a copy of the essay.)

Procedure

 

Two subjects were scheduled for each session.
Immediately upon arriving at the experiment, subjects
were paid $2 and asked to sign a receipt. E then assured
the subjects of the confidential nature of their responses.

After reminding §§ that the purpose of the research was to

74

examine the ways in which people organize their thoughts,
§_passed out copies of the essay on an all-volunteer army
and proceeded with the following instructions:

First of all what we'd like to have you do is read
over the following material. It's of some current
interest and we thought it would be something you
could work with. You'll be given six minutes to
finish reading the essay.

Manipulation of role:playipg variable. After §§
completed the essay, §_read the following instructions
which functioned to define the two experimental tasks.

Now I'd like each of you to do something dif-
ferent with the material in the essay. I'd like one
of you to use the points listed in the essay and pre-
sent a "talk" in your natural speaking style. When
the time comes, I'll ask you a "lead-in" question and
you should answer in the way you normally would when
having a discussion between friends. As sincerely_and
convincingly as possible present the same position as
the one taken in the essay. Use all the arguments
listed and any examples, illustrations or other argu-
ments you want. You'll be given eight minutes to pre-
pare the talk. A sheet of paper will be provided so
that you can take whatever notes you think will be
helpful. Your talk will be tape-recorded for purposes
of analysis.

Now, while one of you is preparing a talk, the
other will be asked to make up an outline of the
essay in whatever form you decide is appropriate.
While the talk is being presented, the other person
should carefully listen to and think about the presen-
tation. After the talk, I'll ask you both some
questions about it.

 

At this point E briefly summarized the instructions
and asked for questions concerning the two different tasks.
Active role-players were informed that they could use what-
ever notes they should desire to take.

At the end of the eight-minute preparation (or out-

line) period, § asked §§ if they were ready. Upon

75

acknowledgment from both S5, E reminded the passive role-
player to listen very carefully and told the active role-
player to begin with the talk after §_had given him a
lead-in question.

After turning on a tape recorder which had been
in full view from the beginning of the session, §_cued the
active role-player by saying, "Well, I don't know. I've
been doing a lot of thinking about it and I just can't
decide although I think I'm leaning toward an all-volunteer
army. How do you feel about it?" Subjects were allowed to
talk as long as they wanted; however, most "talks" ranged
between three and five minutes in length. When the active
role-player had finished his talk, §_collected the notes
of the active role-player and the outline of the passive
role-player. Q then passed out two sets of questions
which he said would help him understand their performance.
(See Appendix B.)

Subjects were not debriefed at the experimental
session but were given several options about receiving
further information about the experiment. Two separate
evening discussions were scheduled for times later in the
quarter. §§ also had a chance to sign-up for a "two- or
three-page description and discussion of the study." For
all §§ to whom the above options were undesirable or

impossible, §_extended an invitation for office

76

appointments. Before leaving, subjects were thanked for
their c00peration and asked not to discuss the experiment
with friends.

Manipulation of persuasive intent. The persuasive
intent variable was intended to induce implicit interfer-
ing responses (see introduction). To instigate these
affectively negative reactions, E told the subjects, "As
you'll see [or 'As you've seen'] the essay is a recent
campus-position paper which was intended to influence
people so that they would be against an all-volunteer army."
For one-fourth of the experimental §§ the "forewarning" was
delivered before they read the essays, for one-fourth the
"forewarning" was delivered afpgp the essay, for another
fourth the "forewarning" came before the attitude measure-
ment, and for another fourth of the subjects there was no
mention of the persuasive intent of the essay.

Control condition. After being paid $2 and signing

 

a receipt and without being exposed to any experimental
material or manipulations, control subjects answered a
set of questions concerning their feelings about national
issues of current interest and filled out a set of items
composed of two personality scales. §_then informed the
‘§§ that they were in a control group and described the
activities of experimental §§° As with experimental §§v
the control §§ were thanked and asked not to discuss the

experiment with friends.

77

Dependent Measures

 

Attitudes toward national issues. The first page
of the first set of questions asked subjects to express
their opinions about six issues of national interest by
checking a point on an ll-point rating scale. The third
item of the set, which is included below as an example
of the format of these questions, was the measure of the
major dependent variable of the study.

The United States should have an all-volunteer army.

 

1 2 3 4 5 6 7 8 9 10 ll
Strongly Strongly
Agree Neutral Disagree

(See Appendix B for a list of the entire set of questions
used.)

Evaluation of the improvised talk. Both the pas-
sive and active role-players rated the talk on four
dimensions. §§_rated how convincing, interesting, sincere,
and logical the presentation had been. Role—playing sub-
jects rated themselves and passive participants rated the
performance of the active participants with whom they were
paired. Ratings were made on a scale such as the one
below. (See Appendix B.)

How convincing was the "talk" which was just

 

given?
1 2 3 4 5 6 7 8 9 10 11
Not at all Very

convincing convincing

78

Perception of essay and its source. Besides
rating the talk, subjects were also asked to rate the
original essay. §§_indicated how "fair“ the "essay" had
been and how "knowledgeable" and "trustworthy" the source
(or sources) had been. S5 also checked how much they had
thought about the essay's intent to persuade. Again 11-
point scales with end-point labels like those used for
evaluating "the talk" were provided.

Recall. The last question of the first set of
questions asked §§_to list the main thesis and each of

the supporting arguments of the essay.

CHAPTER III

RESULTS

 

Check of Premanipulation Equivalence
of Experimental subjects

 

The after-only nature of the design used in this
study makes it impossible to specify definitely whether
or not there were significant premanipulation differences
among the various experimental conditions in terms of
attitudes toward an all-volunteer army. Indirect evi-
dence does suggest, however, that the procedures used to
assign subjects to conditions were effective in creating
experimental groups relatively equivalent with reSpect
to the critical attitude issue.

After the experimental manipulation, subjects were
asked to indicate their opinions on five other issues of
nationwide concern in addition to the critical issue of
an all-volunteer army. Nonsignificant main effects and
interactions on these nonmanipulated topics would suggest
that the assignment procedures were successful.

The values of the F statistics and associated pro-
bability levels obtained from the analysis of variance of

opinion data for each issue are presented in Table l.

79

F Statistics and Associated Probability Values

80

Table 1

from Analyses of Variance of the Non-

manipulated Attitude Issues

 

Attitude

 

Main Effect Main Effect

 

 

 

 

 

Type of Mention of Interactionb
Issue Exposurea Intentb
F P F P F P
Question 1
(development
of ABM) .642 (.424) 1.046 (.375) .087 (.967)
Question 2
(demon-r
strations) .001 (.976) 2.152 (.097) .324 (.808)
Question 4
(college
deferments) .559 (.456) 1.802 (.150) 1.155 (.330)
Question 5
(family size) .868 (.353) 1.363 (.257) 1.191 (.316)
Question 6
(Supreme
court) .185 (.668) .818 (.486) .758 (.520)
adf = 1,120
bdf = 3,120

81

None of the main effects or interactions on any of the
issues obtained conventional levels of significance and

it was therefore assumed that the various experimental
conditions were equivalent with respect to premanipulation
attitudes toward an all-volunteer army.

Assessment of Experimental
Manipulations

 

Type of exposure. No measure of this variable was

 

taken. It was assumed that each participant would be
aware of and could identify the behaviors associated with
the condition to which he was assigned.

Mention of persuasive intent. As part of an exper-
iment on "the ways in which people organize their thoughts,"
subjects were presented with an essay against an all-
volunteer army. In some conditions the manipulative
intent of the essay material was explicitly stated to
the subjects by the experimenter. The purpose of the
manipulation was to arouse implicit interfering responses
(see introduction).

It will be recalled that the time at which this
observation was introduced was systematically varied:
for some subjects the comment was delivered before they
read the essay; for others the comment was delivered
after they read the essay and before they prepared for
an informal talk or outlined the material; in yet another

condition, the comment was delivered after the

82

role-performance had been given and just before post-
manipulation questionnaires were administered. In one
condition the experimenter did not explicitly mention the
persuasive intent of the essay. It was hypothesized that
this mention of persuasive intent would greatly reduce
the superior persuasive effect of active participation

as compared with passive participation.

As a check to see if the manipulation was success-
ful, subjects were asked, "How much did you think about
the essay's intent to influence your feelings?" and were
given an ll-point rating scale (anchored at l with the
label "not at all" and at 11 with the label "constantly")
to indicate their reaction. An analysis of variance of
the scores revealed that there were no significant dif-
ferences in concern about intent among the experimental
conditions. The means and analysis of variance of the
subjects' self-perceptions concerning their thoughts
about the essay's persuasive intent are presented in
Table 2 .

The data seem to suggest that the manipulation -
was unsuccessful. There was no tendency for subjects
to report more thoughts about the essay's intent to
influence when the statement about intent was delivered
earlier in the experimental proceedings. In fact, sub-
jects in the condition in which no mention of intent was
delivered reported virtually the same degree of interfer-

ing thoughts as did subjects in the other conditions.

83

Table 2

Means and Analysis of Variance of Subjects' Self-
perceptions of Their Thoughts about
Persuasive Intent

 

 

 

 

Conditiona M1 M2 M3 M4
Active Participation 5.56 5.94 6.25 5.56
Passive Exposure 6.81 4.56 7.06 6.31
Source of Variance SS df MS F P
Type of Exposure 4.13 l 4.13 .54 ns
Mention of Intent 33.02 3 11.01 1.43 ns
Interaction 33.27 3 11.10 1.44 ns
Error 924.56 120 7.70

Total 994.99 127

 

Note.--The higher the score the more thought
about intent to persuade; n = 16 in each cell.

aMl

reads essay.

mention of intent before subject initially

M2 = mention of intent after subject reads essay.

M3 = mention of intent after informal talk
delivered.

M4 = no mention of persuasive intent.

84

The subjects may have differentially experienced
thoughts about the persuasive intent of the essay as
intended but the differences may have been obscured when
subjects attempted to reconstruct their thoughts concern-
ing what was clearly a biased essay about an all-volunteer
army. It is thus possible that the manipulation was
effective but that the question used to check it was so
direct that it introduced a demand to report the presence
of thoughts about the manipulative intent. An analysis
of subject ratings of how fair the essay was offers some
support for this possibility.

That is, for subjects in active role-playing con-
ditions, the earlier in the procedure mention of intent
was delivered, the greater the tendency for subjects to
perceive the essay as biased. Table 3 presents the means
and analysis of variance for the ratings of fairness of
the essay.

The means for passive exposure conditions do not
parallel the trend of means for the active-participation
conditions. In fact, the mention of intent condition
which induced the highest ratings of bias among role-
players, induced the lowest ratings of bias among non-
role-playing subjects. This suggests that the manipulation
was effective for active participants but had an unintended
effect for passive exposure subjects. Subjects were also

asked to indicate how knowledgeable and how trustworthy

85

Table 3

Means and Analysis of Variance of Subject's Ratings
of Fairness of the Essay

 

 

 

 

Conditiona M1 M2 M3 M4
Active Participation 9.50 8.94 7.94 7.38
Passive Exposure 6.38 8.25 8.00 8.25
Source of Variance SS df MS F P
Type of Exposure 16.53 1 16.53 2.85 .094
Mention of Intent 11.78 3 3.93 .68 ns
Interaction 71.53 3 23.84 4.11 .008
Error 695.38 120 5.79

Total 795.22

 

Note.--The higher the mean, the more the essay was
perceived to be biased; n = 16 in each cell.

aMl = mention of intent before subject initially

reads essay.

M2 = mention of intent after subject reads essay.

M3 = mention of intent after informal talk
delivered.
M4 = no mention of persuasive intent.

86

they felt the source of the counterattitudinal essay to be.
None of the main effects or interactions approached con-
ventional levels of significance.

When the evidence concerning the manipulation of
‘intent is reviewed, it unfortunately becomes quite clear
that any interpretation of the data with respect to the
mention of intent variable must be purely speculative.
The meaning of the manipulation is at best ambiguous.
As a result of these findings, the validity of Hypothe-
sis 1 can still be properly examined but the examination
of Hypothesis 2, the major hypothesis of the study, becomes

a questionable exercise.

Test of Hypotheses

 

Hypothesis 1. The first hypothesis predicted that

 

both types of participation, active role-playing and pas-

sive exposure, would be effective in changing attitudes.

This hypothesis specifically predicted:

H1: Both passive exposure to a counterattitudinal com-
munication and active involvement in the expression
of a counterattitudinal communication produce a sig-
nificant amount of attitude change in the direction
of the counterattitudinal position.

Essentially, this tests whether or not the pro-
cedures were effective in changing attitudes. A supple-
mentary control group unexposed to any of the experimental

manipulations was added to the 2 x 4 after-only design to

provide a baseline against which change could be assessed.

87

The means for the experimental conditions and the control
condition on the all-volunteer army issue are presented
in Table 4.1

Results of the analysis of the mean differences
indicate that only the passive exposure--Ml subjects
(mention of intent given before subjects read the essay)
and passive exposure--M3 subjects (mention of intent given
after the informal talk had been presented) differed sig-
nificantly (p < .05) from control subjects. (The critical
value for Dunnett's t statistic, p < .05, for 9 means and
120 degrees of freedom was 2.41, one-tailed.)

This indicates that only subjects in the two
mentioned passive participation conditions demonstrated

a significant amount of attitude change. However, the

difference between the mean of the experimental condition

 

1A modified version of Dunnett's test for compar-
ing all means with a control was used to test Hypothesis 1.
Dunnett's test (Winer, 1962, p. 89) is appropriate for
comparing a control condition of K conditions with each
of the rest of the conditions. The level of significance
chosen applies to the whole set of K-l comparisons and not
to each of the individual comparisons. Dunnett's t-statis-
tic is:

t = T.-T
41 o
/2MS error/n

 

In this case a harmonic mean was computed for n with 16
subjects in each of the eight experimental groups and 21
subjects in the control group. This computed value for
n was 16.43.

88

Table 4

Means of Attitude Toward an All-Volunteer Army for
Experimental and Control Groups

 

 

Conditiona M1 M2 M3 M4
Active Participation 4.62 4.31 4.12 4.81
Passive Exposure 7.12 4.12 5.94 4.75
Control 3.42b

 

Note.--The lower the mean, the more favorable
toward an all-volunteer army.

aMl = mention of intent before subject initially
reads essay.

M2 = mention of intent after subject reads essay.

M3
delivered.

mention of intent after informal talk

M4 = no mention of intent.

bn = 16 in each cell but this one where n = 21.

and the control condition was in the predicted direction

for each of the experimental conditions (p < .01, binomial

test) suggesting that, in general, the procedures tended

to change the subject's attitudes in the intended direction.

Thus, the first hypothesis receives some confirmation

although the support is rather weak.

Hypothesis 2. The second hypothesis states:

H2: Active involvement in the expression of a counter-
attitudinal communication leads to a greater amount
of attitude change than passive exposure to the
counterattitudinal communication when there is no

mention of the persuasive intent of the counter-
attitudinal communication.

89

The means for the attitudes toward an all-volunteer
army were presented in Table 4. Table 5 presents the
analysis of variance of the attitude scores for the

experimental subjects.

Table 5

Analysis of Variance of Attitude Toward an All-Volunteer
Army Scores for the Experimental Subjects

 

 

Source of Variation SS df MS F P
Type of Exposure 33.01 1 33.01 3.26 .073
Mention of Intent 45.52 3 15.17 1.50 .218
Interaction 43.58 3 14.52 1.44 .235
Error 1212.81 120 10.10

Total 1334.92 127

 

The predicted interaction was not significant nor
was the main effect for the "mention of intent" variable.
Since the manipulation of mention of intent variable did
not seem to have the intended effect, the failure to
find the predicted interaction is to be expected. Rather
unexpectedly, the main effect for "type of exposure"
approached conventional levels of significance (p < .08).
This latter outcome was in the opposite direction from
that expected. Contrary to a great deal of research,
passive exposure had a tendency to produce more attitude

change than an active form of counterattitudinal advocacy.

90

Supplementary Analyses

The attention hypothesis. An item in the post-

 

manipulation questionnaire asked the subjects to list
the main thesis and each of the supporting arguments
presented in the essay and permitted an examination of
the attention hypothesis. The interjudge reliability of
the two judges' ratings of recall was .91. For purposes
of comparison a subject's recall score was the combination
of the two judges' ratings. The means and analysis of
variance of the recall scores are presented in Table 6.

The main effect for mention of intent and the
interaction between type of participation and mention of
intent were not significant. The main effect for type of
participation approached significance (.10 < p < .05).
That is, there was a trend for actively involved subjects
to recall more of the counterattitudinal essay. This is
what the attention hypothesis predicts. However, the
attention hypothesis also predicts greater attitude change
along with greater recall and the data did not support
this prediction. While subjects actively involved in
presenting a counterattitudinal position had greater
recall scores than passively exposed subjects, actively
involved subjects demonstrated less attitude change.

The improvisation hypothesis. To explore the
predictions of the improvisation hypothesis, two general

dimensions of the counterattitudinal performance were

91

Table 6

Means and Analysis of Variance for Recall Scores for
Experimental Subjectsa

 

 

 

Conditiona M1 M2 M3 M4
Active Participation 11.75 11.12 12.50 9.68
Passive Exposure 10.38 10.00 10.50 10.31
Source of Variation SS df MS F P

 

Type of Exposure 30.03 1 30.03 3.70 .05 < p < .10

Mention of Intent 40.12 3 13.38 1.65
Interaction 30.34 3 10.11 1.25
Error 973.38 120 8.11

Total 1073.88 127

 

Note.--The higher the mean, the greater the recall;
n = 16 in each cell. The potential range for the recall
scores was 0-16, the actual range was 2-14.

am

reads essay.

mention of intent before subject initially

M2 = mention of intent after subject reads essay.

M3 = mention of intent after informal talk
delivered.

M4 = no mention of persuasive intent.

92

examined. First of all, the length of time spent in
delivering the counterattitudinal talk was obtained by
timing the tape-recorded talk with a stop watch. In
addition, role-player's talks were rated on a 7-point
scale for overall persuasiveness. The interjudge
reliability of the judge's ratings of persuasiveness
was .81.

Since there were no significant differences among
active role-playing conditions in either overall per-
suasiveness (F < l) or length of performance (F < l),
the improvisation hypothesis was tested by examining
within cell and overall correlations between performance
and attitude. These correlations are presented in
Table 7. The within condition correlation of attitude
score with overall persuasiveness ranged from -.l64 to
.145. The within condition correlation of attitude score
with length of performance ranged from .114 to .320. None
of the correlations were significantly different from zero
(Critical value of r .05, df = 14, one-tailed = .426).
The overall correlation between attitude toward an all-
volunteer army and length of performance was .2055. A
value this large or larger has approximately a .05 proba-
bility of occurrence. The overall correlation between
attitude toward an all-volunteer army and overall per-
suasiveness was -.0091 and did not begin to approach

significance.

93

Table 7

Correlation Between Attitude Toward an All-Volunteer
Army and Length and Rated Overall Persuasiveness
of Performance for Active Participation Subjects

 

Conditiona M1 M2 M3 M4 Overall

 

Correlation of
attitude with:

Length .2411 .1728 .3203 .1140 .2055*
Overall Persua-
siveness -.1643 .0321 -.0859 .1449 -.0091

 

Note.--for within cell correlations r.05, df = 14,
one-tailed = .426; for the overall correlation r.05, df =
60, one-tailed = .211 (Ferguson, 1966, p. 413).

*

p < .06.

aMl = mention of intent before subject initially
reads essay.

M2 = mention of intent after subject reads essay.

M3 = mention of intent after informal talk is
delivered.

M4 no mention of persuasive intent.

The satisfaction hypothesis. The essence of the
satisfaction hypothesis is that a role-player will be
influenced in the direction of his counterattitudinal
performance to the extent to which he feels he has done
a good job. To gather data for this hypothesis active
role-playing subjects were asked to indicate how convinc-
ing, interesting, sincere, and logical they had been in
giving the informal talk. That is, the subjects rated

themselves on the mentioned dimensions.

94

There were no significant differences among
mention of intent conditions for role-playing subjects
on any of the self-ratings. There was no theoretical
basis in the present research to use non-role-playing
subjects' ratings of the active participants' performance
and consequently these ratings were not used in any
analyses. The within cell and overall correlations
between attitude and self-rating for each dimension
are presented in Table 8. Both self-ratings of sincerity
and convincingness were found to be consistently related
to attitude toward an all-volunteer army. The overall
correlation between attitude toward an all-volunteer
army and the rating of sincerity was .356 (p < .005, df =
62, one-tailed). The overall correlation between attitude
toward an all-volunteer army and the rating of convincing-

ness was .270 (p < .025, df = 62, one-tailed).

95

Table 8

Correlation Between Attitude Toward an All—Volunteer
Army and Self-Rating of Performance for
Active Participation Subjects

 

Conditiona M1 M2 M3 M4 Overall

 

Correlation of
Attitude with

Self-Rating
"Convincing" .4229 .2896 .0919 .2184 .2701**
"Interesting" .2789 .4613* -.0120 -.1043 .1150
"Sincere" .4154 .4822* .3458 .2228 .3559***
"Logical" .5131a -.1579 .0442 -.1349 .1119

 

Note.--n = 16 in each experimental condition:
df = 14, one-tailed: r .05 = .426, r .01 = .574
df = 62, one-tailed: r .05 = .295, r .001 = .325
(Ferguson, 1966, p. 413).

aM1 = mention of intent before subject initially
reads essay.

M2

mention of intent after subject reads essay.

M3 = mention of intent after informal talk
delivered.

M4

no mention of persuasive intent.

*
p < .05, one-tailed.

**
p < .025, one-tailed.

***
p < .005, one-tailed.

CHAPTER IV

DISCUSSION

A large number of studies (Janis & King, 1954;
Culbertson, 1957; Elms, 1967; Janis & Mann, 1965) have
shown that active participation in the expression of
counterattitudinal position leads to a significant amount
of attitude change in the direction of the public state-
ment. Many more studies have demonstrated that an experi-
menter can produce desired attitude change by passively
exposing subjects to a persuasive communication. It is
somewhat surprising then that the first hypothesis (which.
predicted that both active and passive forms of contact
with counterattitudinal information would lead to a sig-
nificant amount of attitude change), received only weak
support. Only two of the eight experimental groups were
significantly different from the control group on the
primary dependent variable measure (see Table 2).

One partial explanation of the limited amount of
attitude change may be related to a finding by Greenwald
(1970). In that study it was obvious that the role-

playing effect did not occur when subjects had an

96

97

opportunity to consider and reject the counterattitudinal
argument before they were assigned the counterattitudinal
task. The role-play effect was observed, however, when
the counterattitudinal material was received after the
role assignment.

In the present study subjects in all experimental
conditions read the counterattitudinal essay before role
assignment. This procedure was used to ensure equal
attention by both role-playing and non-role-playing sub-
jects to the content of the attitude-discrepant communi-
cation. The unintended effect may have been to reduce
the amount of attitude change.

Yet another explanation of the limited attitude
change involves an aspect of the counterattitudinal essay
which may have highlighted peer group opinion on the
attitude issue under investigation. Specifically, the
introduction of the essay attacked student support of an
all-volunteer army as short-sighted and self-serving. A
matter of historical accident, a campus-wide boycott of
classes because of the first U.S. invasion of Cambodia,
may have increased the importance and salience of peer
opinion. Pilot studies just prior to the boycott revealed
that student opinion was overwhelmingly in favor of an
all-volunteer army. It has often been observed that
reference groups can function as a force not only in

developing attitudes but also in maintaining attitudes.

98

Thus, in the present study at a time when there were
strong anti-war feelings, the comment in the essay about
peer group opinion may have reduced the effect of the
counterattitudinal essay.

Uncertainty about the effectiveness of the
manipulation designed to induce implicit interfering
responses presents even a more serious problem. Two
possible approaches can be considered with respect to
the manipulation. First of all, it may be assumed that
the manipulation was effective as intended but that the
effect was inappropriately measured. It was previously
suggested that the manipulation-check item may have been
too direct, thus obscuring actual differences.

Another possibility is related to the fact that
the critical aspect of the manipulation involved the
time dimension; yhgp during the processing of the counter-
attitudinal material the thoughts about manipulative intent
occurred may have been of major importance. It may not
have been possible for subjects to report accurately the
onset of their reactions about intent. As a final con-
sideration it is suggested that it may have been better
to ask subjects how strongly they reacted to this per-
suasive intent rather than how much of the time they
thought about the persuasive intent.

The other approach, and the one which seems most

likely, assumes that the manipulation was not effective as

99

intended. If the manipulation had been effective, the
attitudes of active role-playing subjects in conditions
where intent was explicitly mentioned would have been
significantly lower than the actively role-playing sub-
jects who did not receive mention of persuasive intent.
Table 4 indicates that the means of former group (4.62,
4.31, and 4.12) were essentially the same as the mean
of the latter group (4.81).

In addition, the attitude data suggest that the
manipulation tended to have different effects on active
role-players than on passive participants. In two of
the three passive exposure conditions, the explicit
mention of persuasive intent tended to facilitate atti—
tude change.

Apparently, for active participants, the manipu-
lation provided little information that was not immediately
available without explicit mention of intent.

Very much related to the question of the actual
effect of the mention of intent manipulation is the find-
ing that there was a near significant difference in mean
in attitude toward an all-volunteer army between active
role-players and passive participants. Passive partici-
pants were more influenced by the counterattitudinal
material. Clearly the difference was due to the effect
of the mention of intent variable on passively exposed

subjects (see Table 4). It can be seen that passively

100

exposed subjects who were told about the persuasive intent
before initially reading the essay material and passively
exposed subjects who were told about the persuasive intent
of the essay after the informal talk had been given evi-
denced attitude scores substantially larger than the
attitude scores of subjects in the other conditions.

It should be noted that the change facilitating
effect of role-playing has not been universally acknowl-
edged. McGuire (l969a) has been responsible for some of
the more forceful objections. Several studies were cited
which have found greater attitude change with passive than
active participation. For his skeptical readers, no
doubt, McGuire provided several reasons for the obvious-
ness of the observed decreased efficacy of active partici-

pation. The major reasons included the facts that:

(a) Aspects of active participation may interfere

with learning.

(b) The active participant has little interest in or

acquaintance with the topic.

(c) The active participant is unmotivated or unpre-

pared to utilize the participation opportunity.

(d) The active participant typically does not have a
supply of relevant arguments within his cognitive

repertory.

101

It is clear, however, that none of the above con-
siderations provide a very cogent explanation of the atti-
tude change data of the present study. They do not help
explain the greater attitude change demonstrated by passive
participants. First of all, all subjects were supplied
with a set of counterattitudinal arguments. While active
participants may have been unprepared to use the material,
the nature of the experimental task would suggest that
active participants would be more strongly motivated than
passive participants to use the information since active
participants expected to make a public statement based on
the information. Furthermore, pilot studies indicated
that the issue was in fact a very salient one for college
students--particularly male students. Finally, active
participants obtained near significantly greater recall
scores than passive participants, suggesting that the
active participation did not interfere with learning.

Although the suggestions by McGuire were of
little value in the present study, they should not be
lightly dismissed. The "suspension of critical judgment
approach" also suggests that there are situations in
which active participation in expressing a counter-
attitudinal position will be no more effective and
perhaps even less effective than passive exposure to
the same counterattitudinal material. In situations

in which the role-playing calls attention to the negative

102

aspects of a sponsor, communicator, or the role-playing
situation, role-playing is not expected to be superior
to passive exposure as far as attitude change is con-
cerned.

In another context, Greenwald (1969) has suggested
that the less biased evaluation disposition associated
with assignment to a counterattitudinal position may be
of little or no consequence on highly familiar issues
or on issues which provide little new information. As
a related point it was previously noted that the sus-
pension of critical judgment approach suggests that dif-
ferences between active and passive forms of exposure to
counterattitudinal information will be minimal on low ego-
involving issues--issues about which few defensive reactions
and little resistance can be expected.

An examination of the means on the attitude toward
an all-volunteer army issue (Table 4) suggests that there
is no simple explanation of the data. Particularly
troublesome for any sort of parsimonious speculation are
the data from the conditions in which the attention of
passively exposed subjects was directly called to the
persuasive intent of the essay material. Passive partici-
pation subjects who received mention of intent after
reading the essay material demonstrated attitudes most
similar (of all eight experimental groups) to unexposed

control subjects, while subjects in the other two

103

conditions where there was explicit mention of persuasive
intent were least similar to the unexposed controls. The
mention of intent did not have a uniform effect on pas-
sively exposed subjects. All that can be safely inferred
is that some aspect(s) of the experimental procedure was
(were) particularly effective in inducing subjects in some
conditions to change their attitudes. Perhaps a dif-
ferent explanatory proposition is necessary for each
mention of intent condition. Further speculation about
reasons for the observed effect seems rather fruitless.
None of the various theoretical positions on role-playing
would have predicted the obtained results.

In view of the complex and ambiguous effect of the
mention of intent variable, little will be said about the
failure of dissonance theory to predict the attitude
change effects. The findings were not inconsistent with
the suspension of critical judgment approach, if it is
assumed that the manipulation led to less interfering
responses for "passive participants." That this was
partially the case is seen in the ratings of essay fair-
ness (see Table 3). Nevertheless, as in the case of
dissonance theory, speculation about the explanation of
the obtained data is of dubious value. The only role-
playing hypothesis about which conclusive evidence was
obtained was the attention hypothesis. An examination

of Table 4 and Table 6 indicates that the hypothesis was

104

not confirmed. Active participants demonstrated higher
recall scores but less attitude change than passive par-
ticipants.

Replicating a finding in many other studies,
objective ratings of the overall persuasiveness of the
role performance demonstrated no support for the improvi?
sation hypothesis. However, within cell correlations
between length of role-play performance and attitude
change were weak but consistently positive and the over-
all correlation between the two variables approached con-
ventional levels of significance.

Self-ratings of performance (how sincere, and how
convincing) were found to be consistently correlated with
attitude change across the four role-playing conditions.
This can be taken as partial support for the satisfaction
hypothesis, that a subject will tend to be convinced by
his performance the more he feels he has done a good job.
However, it is possible that the ratings of sincerity and
convincingness are only indirect attitude measures not
independent from the critical attitude issue. Or perhaps
they are indirect measures of the absence of interfering
responses. Stronger tests of all the various hypotheses
can be made when significant between-condition attitude
effects are obtained-~a condition not met in the present
study.

What does become clear from an examination of the

data is the importance of more closely examining the

105

procedures used in the present study. While the conditions
established for active participants were quite similar to
those employed in many other studies, the circumstances

for passive participants were somewhat unusual. In no
other study were conditions preceding role assignment so
suspenseful. Before the actual coin toss to decide which
of the pair of subjects would deliver an informal talk

and which would outline the material, both subjects were
well aware of the strongly discrepant stand that it would
be necessary for one of them to publicly project. Informal
observations indicated that the subjects receiving the
outlining task expressed a great deal of relief (smiling,
sighs of relief, exclamations, etc.).

This can be contrasted to the more sober expressions
often accompanied by explicit references by the subjects
who had "won" the task of delivering an informal talk to
uncertainty about ability to deal with the assignment. To
what extent relief about task assignment mediated the
attitude change effects is not clear, but it seems plausible
that the relief may have accounted for some of the observed
change. The mention of persuasive intent jog; before the
second processing (preparation for talk on outlining) of
the counterattitudinal material may have dampened the
facilitating effect of the positive effect resulting from
task assignment. The study by Janis and King (1954)

employed procedures most similar to the ones used here.

106

The differences are that in the Janis and King study
there were three subjects per experimental session and
each subject was asked to deliver a speech. Since each
subject had to deliver a speech the "relief" factor
would not seem to be relevant.

The foregoing discussion points to certain
troublesome aspects of the particular paradigm chosen
for the present study. Specifically, the decision to
schedule both an active and a passive participant in
each experimental session seems to have introduced cer-
tain unintended dynamics. That is, the role assignment
procedures seem to have also manipulated the subject's
feelings of relief, which apparently interacted with the
mention of intent manipulation. The complex pattern of
attitude differences is taken as a reflection of these
unintended dynamics. As was mentioned in the introduction,
this design decision derived from a concern about the
serious problem of inequalities between active and pas-
sive participants in exposure to counterattitudinal
information.

On the basis of the present study, it is recommended
that future research exploring role-playing related hypothe-
ses and propositions schedule only one subject per experi-
mental session. To equate for exposure to counter-

attitudinal information, passive participants could be

107

exposed to written material, audio tapes, or video tapes
of an active participant's counterattitudinal performance.

Another aspect of the present study which should
receive at least passing attention concerns the nature
of the task set out for passive participants. Phrased
as a question of general research import, the issue is,
how passive is the passive participant? In this study,
so-called passive participants read a persuasive essay,
outlined the essay material, and listened to the counter-
attitudinal performance of a peer. Only in a very rela-
tive sense does it seem correct to call these activities
passive participation or passive exposure. Perhaps overt
and covert role-playing would have been more appropriate
labels. That the passive participants were so active may
partially account for the nearly identical persuasive
effect of active and passive forms of participation for
subjects not explicitly told about the persuasive intent
of the essay.

The present research does not support or refute
the propositions of the suspension of critical judgment
approach. The utility of this new position remains to
be fully examined. The possible directions to be taken
are many. A modified replication of the present design
was previously suggested; specifically that was to
schedule only one subject per session but take precautions

to equate for exposure of information. Another possible

108

method of validating the suspension of critical judgment
approach would be to use the technique presented by
Osterhouse and Brock (1970). This would involve giving
active and passive role-playing subjects a brief time
period following the counterattitudinal performance to
list their ideas about the experimental issue. This
material could then be content analyzed for counter-
arguments. The suspension of critical judgment pre-
diction would be that active participants would obtain
lower counterargument scores and demonstrate more atti-

tude change.

REFERENCES

REFERENCES

Aronson, E. The psychology of insufficient justification:
An analysis of some conflicting data. In S.
Feldman (Ed.), Co nitive consistency. New York:
Academic Press,_Ig66.

 

Brehm, J. W., & Cohen, A. R. Explorations in cognitive
dissonance. New York: Wiley, 1962.

 

 

Brock, T. C. Communication discrepancy and intent to
persuade as determinants of counterargument pro-
duction. Journal of Experimental Social Psychology,
1967, 3, 296-309.

Carlsmith, J. M., Collins, B. E., & Helmreich, R. L.
Studies in forced compliance: I. The effect
of pressure for compliance on attitude change
produced by face-to-face role playing and anony-
mous essay writing. Journal of Personalipy and
Social Psychology, 1966, 5, 1513.

 

Chapanis, N. P., & Chapanis, A. Cognitive dissonance:
Five years later. Psychological Bulletin, 1964,
61, 1-22.

 

Cohen, A. R. An experiment on small rewards for dis-
crepant compliance and attitude change. In J. W.
Brehm & A. R. Cohen, Explorations in cognitive
dissonance. New York: Wiley, 1962.

 

Cohen, A. R., Brehm, J. W., & Fleming, W. H. Attitude
change and justification for compliance. Journal
of Abnormal and Social Psychology, 1958, 5_,
276-278.

Collins, B. E. Social psychology. Reading, Mass.: Addi-
son-wesley, 9 .

Collins, B. E., & Helmreich, R. L. Studies in forced com-
pliance II: Contrasting mechanisms of attitude
change produced by public-persuasive and private-
true essays. Journal of Social Psychology, 1970,
81, 253-264.

109

110

Crano, W. D., & Messé, L. A. When does dissonance fail?
The time dimension in attitude measurement. Journal
of Personality, 1970, 88, 493-508.

 

Culbertson, F. M. Modification of an emotionally held
attitude through role playing. Journal of Abnormal
and Social Psychology, 1957, 88,230-233.

 

 

Elms, A. C. The influence of fantasy ability on attitude
change through role—playing. Journal of Personality
and Social Psyohology, 1966, 4, 36- 43.

 

 

Elms, A. C. Role playing, incentive, and dissonance.
Psychological Bulletin, 1967, 88, 132-148.

Elms, A. C., & Janis, I. L. Counter-norm attitudes induced
by consonant versus dissonant conditions of role-
playing. Journal of Experimental Research in
Personality, 1965, 1,50- 60.

 

 

Ferguson, G. A. Statistical analysis in psychology and edu-
cation. New York: McGraw-Hill, 1966.

 

Festinger, L. A theogy of cognitive dissonance. Evanston,
111.: Row, Peterson,il957}

 

Festinger, L., & Carlsmith, J. M. Cognitive consequences
of forced compliance. Journal of Abnormal and
Social Psychology, 1959,58, 203- 210

 

 

Greenwald, A. G. Cognitive learning, cognitive response
to persuasion, and attitude change. In A. G.
Greenwald, T. C. Brock, & T. M. Ostrom (Eds.),
Psychological foundations of attitudes. New
York: Academ1c Press, 1968.

Greenwald, A. G. The open-mindedness of the counter-
attitudinal role player. Journal of Experimental
Social Psycholo , 1969, 5, 375- 388.

 

 

Greenwald, A. G. When does role playing produce attitude
change? Toward an answer. Journal of Personality
and Social Psychology, 1970 ,_16, 214- 219.

 

 

Greenwald, A. G., & Albert, R. D. Acceptance and recall
of improvised arguments. Journal of Personality
and Social Psychology, 1968, 8, 31-34.

 

 

Harvey, O., & Beverly, G. Some personality correlates of
concept change through role playing. Journal of
Abnormal and Social Psychology, 1961, 88, I25-I30.

 

111

Helmreich, R., & Collins, B. E. Studies in forced com-
pliance: Commitment and magnitude of inducement
to comply as determinants of opinion change.
Journal of Personality and Social Psychology,
1968, 1_O, 75-81.

 

Holmes, J. G., & Strickland, L. H. Choice freedom and
confirmation of incentive expectancy as determi-
nants of attitude change. Journal of Personality
and Social Psychology, 1970, 11, 39-45.

 

 

Hovland, C., Janis, I., & Kelley, H. Communication and
persuasion. New Haven: Yale University Press,

Insko, C. A. Theories ofJatEitude change. New York:
Appleton§Century Crofts, 1967.

 

Janis, I. L., & Gilmore, J. B. The influence of incentive
conditions on the success of role playing in modi-
fying attitudes. Journal of Personality and
Social Psychology, 1965, 1, 17—27.

 

Janis, I. L., & King, B. T. The influence of role-playing
in opinion change. Journal of Abnormal and Social
Psychology, 1954, 88, 211-21 .

 

Janis, I. L., & Mann, L. Effectiveness of emotional role-
playing in modifying smoking habits and attitudes.
Journal of Experimental Research in Personality,
I965, I, 84F90.

Kelman, H. Attitude change as a function of response
restriction. Human Relations, 1953, 8, 185-214.

 

King, B. T., & Janis, I. Comparison of the effectiveness
of improvised vs. non-improvised role playing in
producing opinion changes. Human Relations, 1956,
8, 177-186.

Linder, D. E., Cooper, J., & Jones, E. E. Decision freedom
as a determinant of the role of incentive magnitude
in attitude change. Journal of Personality and
Social Psychology, 1967, 8, 2454254.

 

McGuire, W. J. Attitudes and opinions. In P. Farnsworth
(Ed.), Annual Review of Psychology, 1966, ll,

McGuire, W. J. Personality and susceptibility to social
influence. In E. F. Borgotta & W. W. Lambert
(Eds.), Handbook ofspersonalisy theory and

research. Chicago:_§and McNally, 1968.

112

McGuire, W. J. The nature of attitudes and attitude change.
In G. Lindzey & E. Aronson (Eds.), The handbook of
social_psycholo . Vol. 3. Reading, Mass.:
Addison-wesley, 969(a).

 

 

McGuire, W. J. Suspiciousness of experimenter's intent.
In R. Rosenthal & R. L. Rosnow (Eds.), Artifact
is_Behaviora1 Research. New York: Aca em1c
press , 1569 (b) .

Mann, L., & Janis, I. L. A follow-up study on the long-
term effects of emotional role playing. Journal
of Personality and Social Psychology, 195§v.§r
339-342. *

Mann, L. The effects of emotional role playing on desire
to modify smoking habits. Journal of Expcrimental
Social Psychology, 1967, 8, 354-348.

Nel, E., Helmreich, R. L., & Aronson, E. Opinion change
in the advocate as a function of the persuasibility
of his audience: A clarification of the meaning
of dissonance. Journal of Personality and Social
Psychology, 1969, 12, 117-124.

 

Osterhouse, R. A., & Brock, T. C. Distraction increases
yielding to propaganda by inhibiting counter-
arguing. Journal of Personality and Social Psy-

chology, 1976, 15, 531-358.

Papageorgis, D. Warning and persuasion. Psychological
Bulletin, 1968, 18, 271-282.

 

Rosenberg, M. J. When dissonance fails: On eliminating
evaluation apprehension from attitude measurement.
Journal of Personality and Social Psychology,
1 -4
I_I -

Rosenberg, M. J. Some limits of dissonance: Towards a
differentiated view of counterattitudinal per-

formance. In S. Feldman (Ed.), Co nitive con-
sistency. New York: Academic Press, 1966.

Scott, W. Attitude change through reward of verbal
behavior. Journal of Abnormal and Social Psy-

Scott, W. Cognitive consistency, response reinforcement,
:33 attitude change. Sociometry, 1959, 88, 219-

 

113

Sears, D. 0., & Abeles, R. P. Attitudes and opinions.
In P. Farnsworth (Ed.), Annual Review of Psy-
chology, 1969, 88, 253-288.

 

Steiner, I. D., & Darroch, R. K. Relationship between the
quality of counterattitudinal performance and
attitude change. Journal of Personalisy and
Social Psychology, 1969, 88, 312-320.

wallace, J. Role reward and dissonance reduction.
Journal of Personality and Social Psychology,
1966, 8, 305-312.

Watts, W. A. Relative persistence of Opinion change
induced by active compared to passive partici-
pation. Journal of Personality and Social Psy-
chology, 1967,8, 4-15.

 

Winer, B. J. Statistical principles in experimental
design. New York: McGraw-Hill, 1962.

 

Zimbardo, P. G. The effect of effort and improvisation
on self-persuasion produced by role-playing.
Journal of Experimental Social Psychology, 1965,
8, 163-120.

 

Zimbardo, P. G., & Ebbesen, E. B. Experimental modifi-
cation of the relationship between effort, atti-
tude, and behavior. Journal of Personality and

Social Psychology, 1975, 88, 207-213.

 

APPENDICES

APPENDIX A

COUNTERATTITUDINAL COMMUNICATION

APPENDIX A

CONCERNING AN ALL-VOLUNTEER ARMY

An issue of growing interest, especially to
draft-age people, centers around the question of an All-
Volunteer Army. It is the thesis of this paper that
under present conditions an all-volunteer army is neither
attainable nor particularly desirable.

Much of the controversy is by an increasingly
active student populace. Unfortunately, the cries for
change in the present draft-based system are often louder
than they are logically reasoned. Despite humanitarian
pretentions, it often seems that student advocates of
"end the draft" are really concerned with simply keeping
themselves out of the service and Vietnam. While this
may be an admirable goal, ending the draft may be a rather
short-sighted means of accomplishing it. Very little was
heard about the "draft" when Selective Service was not
conscripting college students or their friends.

Support for an all-volunteer army has not come
solely from students. President Nixon himself (backed by

the Gate's Commission Report) has said that an

114

115

all-volunteer army is a goal which he is working towards.
Perhaps this fact in itself is enough to convince us that
we should take a long, hard look at the consequences of
such a decision (e.g., all-volunteer force vis-a-vis

student dissent of government policy).

Why do we need an army at all?

The nation's present foreign policy is inextricably
intertwined with questions of military manpower. Although
the American public would not readily accept intervention
in another Asian conflict, in all likelihood the U.S.
foreign policy will continue to be one in which we sup-
port existing democratic governments against outside
aggression or those internal problems directly caused
by outside agitation. To carry out this policy--even to
a limited extent--we need the capacity to apply restrained
but appropriate force; that is, something short of nuclear
confrontation. Potential trouble spots in Central and
South America, in Berlin, in Greece and the ever-escalating
Arab-Israeli conflict, will require that the United States
maintain a strong, mobile striking force which can be
expanded with little effort. However desirable it might
seem, the abolition of the military structure in this
country is not practical in the foreseeable future.

Given that there will be a continuous need for at
least some defensive manpower and that there is no ideal

solution to the problem of military recruitment, what we

 

116

must do is make the best out of a bad business. It is
hoped that the following points will clearly illustrate
why an all-volunteer army is not the answer to our man-

power needs.

(1) Cost of an All-Volunteer Army

A rather obvious objection to an all-volunteer
army is that the United States cannot afford it. Sub-
stantial pay hikes are included in all proposals which
present alternatives to the current system of providing
for military manpower. President Nixon's commission
(which supported the all-volunteer army) concluded that
the pay scales necessary to attract enough volunteers
would range between four and seventeen billion dollars
a year--a substantial increase over present levels. Even
if the true cost is only four billion a year, this
increase is more than the country can afford.

The problem of cost becomes important when we con-
sider the great social needs of our country today. Per-
haps the extra billions for a volunteer army would be
justified if social equality were somehow increased.

But this cost is the cause of deep concern if it merely
puts further strain on a budget already over-committed
to our fighting forces. Since there are so many unmet
domestic needs--including housing, education, transpor-
tation, and pollution to mention only a few broad cate-

gories--even a few extra billion dollars in military

117

expenditures could be enough to tip the social equilibrium
toward widespread rioting and disruption.
(2) Civilian concern about
military involvement

Fully as important as the costs involved is the
fact that an all-volunteer force would lessen civilian
concern about the use of military forces. The attentions
of Congress, the press, the federal courts, and anxious
parents would drift far from the indignities of military
life as soon as that life was proclaimed to be voluntary.
There would similarly be less popular concern about the
uses to which volunteer soldiers were put. How many would
care and how deeply, about the Vietnam war if their sons
and brothers were not being conscripted to fight it?
What if proponents of that war could say, "What are you
so upset about--every American boy over there is a volun-
teer?"
(3) Keeping the military insti-

tution flexible

A third objection to an all-volunteer force is
that the already sluggish, tortoiselike military insti-
tution would be even more resistant to change. An end
to the draft would shield the army from the influx of
citizen-soldiers who are the yeast of internal change.
The army needs Yossarians, Ronald Ridenhours, independent-

minded R.O.T.C. junior officers, and J.A.G.

 

118

lawyers--soldiers who do their jobs but who are not com-
mitted to the cover-your-ass system, whose loyalties are
to civilian not careerist values. Given the absence of
countervailing powers within the military, it is the
civilian-in-uniform who is most likely to point out,
articulate and test the areas for change.

To look at this aspect from another perspective,
it could be said that an all-volunteer force would tend
to lead to the development of a separate military ethos.
The result would be an increased probability of military
adventures (like those of C.I.A.) and an even less
responsible foreign policy.

(4) A Rich man's war, a poor
man's fight!

Finally, it has been said that an all-volunteer
army would put an end to peacetime induction. But in so
doing, it would transfer the burden of military service
to those who are most susceptible to being inducted.
These would be primarily the poorer, less educated, and
less sophisticated segments of our youth--the people who
can't get a decent job, for whom the future holds little
promise.

Is it proper that our nation should be defended by
those who have been favored by it least? Is not the burden
of common defense something that all segments of society

should share equally, or at least run an equal risk of

 

119

sharing? And how much justice is obtained by ending con-
scription, but replacing it with a recruiting system
that feeds on poverty, ignorance, and gullibility of
our most disadvantaged youngsters?

There is no such thing as a large painless mili-
tary. The more socially just course lies not in trying
to hide the pain, or transfer it, but in sharing the pain

and seeking, through reform of the Army, to reduce it.

 

APPENDIX B

POSTEXPERIMENTAL QUESTIONNAIRE

 

APPENDIX B

POSTEXPERIMENTAL QUESTIONNAIRE

Please check the scale position which most closely repre-
sents your true opinion.

(1) Development of the ABM is a realistic response to
China's recent military advancements.

 

l 2 3 4 5 6 7 8 9 10 ll
Strongly Strongly
Agree Neutral Disagree

(2) We need a much tougher approach toward violent demon-

 

strations.
l 2 3 4 5 6 7 8 9 10 ll
Strongly Strongly
Agree Neutral Disagree

(3) The United States should have an all-volunteer army.

 

l 2 3 4 5 6 7 8 9 10 ll
Strongly Strongly
Agree Neutral Disagree

(4) Given that there is a draft, there should not be college
student deferrments.

 

1 2 3 4 5 6 7 8 9 10 ll
Strongly Strongly
Agree Neutral Disagree

(5) The federal government should do everything in its
power to limit family size.

 

l 2 3 4 5 6 7 8 9 10 ll
Strongly Strongly
Agree Neutral Disagree

120

 

121

(6) Nixon's behavior toward the Supreme Court has
threatened the dignity of that institution.

 

l 2 3 4 5 6 7 8 9 10 ll
Strongly Strongly
Agree Neutral Disagree

(1) How convincing was the "talk" which was just given?

 

1 2 3 4 5 6 7 8 9 10 11
Not at all Very
Convincing Convincing

(2) Was the presentation interesting?

 

1 2 3 4 5 6 7 8 9 10 11
Not at all Very
Interesting Interesting

(3) Was the presentation given sincerely?

 

l 2 3 4 5 6 7 8 9 10 11
Not at all Very
Sincerely Sincerely

(4) How logical was the presentation?

 

l 2 3 4 5 6 7 8 9 10 11
Not at all Very
Logical Logical

(5) How fair was the essay which you read at the beginning
of this session?

 

1 2 3 4 5 6 7 8 9 10 11
Not at all Very
Biased Biased

122

(6) How knowledgeable about the issue do you think the
person (persons) who wrote the essay was (were)?

 

l 2 3 4 5 6 7 8 9 10 11
Not at all Very
Knowledgeable Knowledgeable

(7) How trustworthy a source of information about the
issue would you estimate the person (persons) who
wrote the essay to be?

 

1 2 3 4 5 6 7 8 9 10 11
Not at all Very
Trustworthy Trustworthy

(8) How much did you think about the essay's intent to
influence your feelings?

 

l 2 3 4 5 6 7 8 9 10 11
Not at all Constantly

(9) How many psychological experiments have you taken
part in prior to this one?

(10) In the space below, list as accurately and yet as
briefly as possible the main thesis and each of the
supporting arguments presented in the essay you read
at the beginning of this session. (Use the back of
the page if necessary.)

va 1 2 T5?)