CDUNTERARGUMENTATION INHIBITION
AS A MEDIATOR OF THE
ROLE - PLAYING - -ATTITUDE- CHANGE.
RELATIONSHIP

Thesis for the Degree of Ph. D.
MICHIGAN STATE UNIVERSITY .
DOUGLAS M. LITTLE
1972

 

    
     

I. .‘ 5' 4\
Michigan. State
University

7 ‘0141 ‘1»,

This is to certify that the

thesis entitled
COUNTERARGUMENTATION INHIBITION AS A MEDIATOR OF THE

ROLE—PLAYING—éATTITUDE-CHANGE RELATIONSHIP

presented by

Douglas M. Little

has been accepted towards fulfillment
of the requirements for

Meme in _P_$1Qh_0.logy

 

, 3
(d/fl'flfl/ \K:/‘Lfl/I/\//>/-

Major professor

Date 8/10/7 2

0-763!

 

ABSTRACT

COUNTERARGUMENTATION INHIBITION AS A MEDIATOR OF THE
ROLE-PLAYING--ATTITUDE-CHANGE RELATIONSHIP

By

Douglas M. Little

Two experiments were conducted to test the proposition that role-
playing is an especially effective attitude change technique because
it inhibits a person's ability to counterargue. Generally, the.
manipulations failed to have the predicted effects. Conditions which.
involved improvising a counterattitudinal speech (Study 1) or a letter
(Study 2) had a tendency to lead to more rather than less counterargumen-
tatiOn and did not lead to more attitude change than conditions in
which subjects were more passively involved and merely read the counter-
attitudinal material. Furthermore, in Study 1, a .42_second delay in
auditory feedback while speaking did not lead to less counterargumen—
tation or to more attitude change than the no delay condition. Finally,
in Study 2, specific instructions to counterargue led to less supportive
argumentation but not more counterargumentation or less attitude change
than a condition in which subjects were not given such instructions.‘

Only the unexpected interaction in Study 1 between Delayed Audi-
tory Feedback (DAF) and Sex.of Subject resulted in effects consistent
with the predicted inverse relationship between.counterargumentation
and attitude change. Male subjects counterargued more and demonstrated

less attitude change when delivering a counterattitudinal speech under

Douglas M. Little
conditions of no delay in auditory feedback than under conditions of
a .42 second delay. Female subjects, on the other hand, counterargued
more and demonstrated less attitude change under the .42 second delay
interval. Females were affected by the DAF manipulation in a manner
contrary to that expected but there was an inverse relationship between
counterargumentation and attitude change.

The only significant effect on attitude change in Study 2 was for
the interaction of Role (Improvise versus Read) and Instructions to
Counterargue (Present versus Absent).‘ When subjects were given specific
instructions to counterargue, improvisation led to more attitude change
than a more passive type of exposure. There were no differences in
attitude change between the Improvise and Read conditions for those
not given instructions to counterargue. These data in conjunction_
with supplementary evidence suggested that the enhanced effectiveness
of role-playing may occur only when the sponsor or source appears to
have benign and nonexploitative intentions.

Although it was found in Study 2 that Improvise conditions led to
more attitude change than Read conditions when instructions.to counter-
argue were given, there were no differences between these conditions
in terms of the amount of counterargumentation that was generated. 0n
the other hand, when no instructions to counterargue were given,
Improvise conditions led to more counterargumentation but no more
attitude change than Read conditions. Thus, Study 2 provided no
evidence to indicate that role-playing is effective because it leads
to a reduction in the incidence of counterargumentation.

The fact that, in both studies, counterargumentation was affected
by more treatments and treatment combinations than was attitude change

further highlighted the limitations of the prOposed relationship

~A-

 

Douglas M. Little
between role-playing, counterargumentation, and attitude change.
However, the evidence with respect to the DAF-Sex of Subject interaction
in Study 1 along with certain within condition correlational evidence
indicated the need for further research.

It was noted that the hypotheses of this research failed to dis-
tinguish between the ability to counterargue and the motivation to
counterargue. This latter factor was presumed to account for much of
the obtained data. In a related matter, it was observed that pretreat-
ment attitudes have a substantial impact on the degree of counterargu-
mentation. As a result it was recommended that future research utilize
pre-post designs. Effort justification, satisfaction with performance,
and recall were not able to account for the attitude change data. Some
of the differences in counterargumentation appeared to be attributable
to demand characteristics and modifications.of the measurement tech-
nique to eliminate such sources of contamination were suggested.
various methodological contributions in this research were also

discussed.

 

COUNTERARGUMENTATION INHIBITION AS A MEDIATOR OF THE
ROLE-PLAYING--ATTITUDE-CHANGE RELATIONSHIP
By

a“

l
Douglas Mia Lit tle

A THESIS

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

DOCTOR OF PHILOSOPHY

Department of Psychology

1972

‘M-
FH- - -. A

 

To Karen

11

 

ACKNOWLEDGMENTS

I am grateful to Drs. Bill Crano, Larry Messé, Eugene Jacobson.
and Bob Ruh for their help as dissertation committee members. They
provided as much guidance and advice as I requested and as much
freedom as I needed. My chairman, Bill Crano, is especially responsible
for the strengths of this research and has been exceptionally generous
with his time and counsel. I must also note my appreciation for the
"free-time" provided by Bob Ruh during the summer of '71.

Others to whom I am indebted include Karen Sanderson, Ken_White,
Ralph Cooper, and Gerry Stornant. There is not room for their stories
but in various ways they provided essential support.

It is also appropriate here to recognize the sacrifices and con-
tributions of my mother and father and grandparents. The road has
been as tough for them as for me.

Finally, I would like to acknowledge the help of Lawrence O'Kelly
and Roger Halley who expeditiously provided the funds for the coding

and the analysis of the data.

iii

TABLE OF CONTENTS
Chapter
I INTRODUCTION . . . . . . . . . . . . . . . . . . . .
II EXPERIMENT ONE . . . . . . . . . . . . .‘. . . .'. .
Method. . . . .'. . . . . . . . . . . . . . .

Overview . . . . . . . . . . . . . . .
Subjects . . . . . . . . . . . . . . .
Procedure. . . . . . .‘. . .'. . . . .
Dependent Measures . . . . . . . . . .

RESUltS‘. o o o o 0.0 o o 0.0 o o o o o o e 0

Assessment of the Effort Manipulation.
Tests of Hypotheses. . . . . . .
Alternative Explanations of the Role-
Playing Effect--Secondary Analyses .
Correlates of COunterargumentation and
Total Defensiveness. . . . . . . . .

III EXPERIMENT TWO 0 o o o o o o o o o o o o o o o o o o
MethOdo o o o o .,. o o o o o o 0'. o 0 0‘. 0

Overview . . . . . . . . . . . . . . .

SUbjects o o o o o o e o o o 0‘. o o 0'

Procedure. . . . . . . . . . . . . . .
Dependent Measures . . . . . . . . . .

RESUICS o o o o o 0‘. o o o o o o o o o 0 0'0

Tests of Hypotheses. . . . . .‘. . . .
Secondary Analyses:. Covariates of»

Message Acceptance, Counterargumenta-

tion, and Total Defensiveness. . . .
IV GENERAL DISCUSSION . . . . . . . . . . . . . . . . .
Possible Determinants of Counterargumentation

Developing a Valid Measure of Counterargumen-
tationo o o e o o e o o o a o e 0.0 '0 o o 0

Evidence Concerning the Relationship Between
Counterargumentation and Attitude Change. .

iv

Page

10
10
14
21

21
26

40
43

50

' 52

52
52
53
59
62

62

72

80

81

88

95

Chapter Page E ,7 I
REFERENCES. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 IHJI
APPENDICES. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105

A. INSTRUCTIONS FOR "READ" CONDITIONS (STUDY 1). . . 105

B. INSTRUCTIONS FOR "IMPROVISE" CONDITIONS
(STUDY 1) I C C I O O O O O O O O O O O O O O O O 107

C. FREQUENCY OF OCCURRENCE OF EACH RESPONSE
CATEGORY (STUDY 1). . . . . . . . . . . . w . . . 109

D. PRETEST MEASURE OF ATTITUDE TOWARD COEDUCATIONAL
LIVING (STUDY 1).. O I O O O I O O O I O C O O O i 110

E. MATERIALS INCLUDED IN POSTEXPERIMENTAL
QUESTIONNAIRE (STUDY 1) . . . . . . . . . . . . . 112

F. FREQUENCY OF OCCURRENCE OF THE MAJOR RESPONSE
CATEGORIES (STUDY 2) O O I O O O O O O O O I O O O 115

G. MATERIALS INCLUDED IN POSTEXPERIMENTAL
QUESTIONNAIRE (STUDY 2) . . . . . . . . . . . . . 116

 

Table

10

11'

12

13

LIST OF TABLES

Carrelations of "Difficulty" and "Effort" Ratings
with Other Self-Perceptions. . . . . . . . . . . . .'.

Summary of Analysis of Variance of Scores for Support-
ing Arguments, Counterarguments, Total Defensiveness,
and Total Number of Arguments--With Separate Listings
for Each of the Four Scoring Methods . . . . . . . . .

Mean Counterargumentation as a FUnction of Delayed
Auditory Feedback Interval and Role. . . . . -.- . . .

Summary of Analysis of Variance of Counterargumenta-
tion and Total Defensiveness . . . . . . . . . . . . .

Mean Total Defensiveness and Number of Supportive
Arguments as a Function of Order of Measurement. . . .

Mean Total Defensiveness as a Function of Sex of
Subject and Order of Measurement . . . . . ... . . . .

Mean Total Defensiveness as a Function of Sex of
Subject and Delayed Auditory Feedback Interval . . . .

Mean Attitude Change as a Function of Delayed Audi-
tory Feedback Interval and Role. . .'. . . . . . . . .

Mean Attitude Change as a Function of the Order of
Measurement 0 I o o c e e o o o o o o o o e o o o o o O 0

Mean Attitude Change as a Function of Delayed Audi-
tory Feedback Interval and Sex of Subject. . . . . .‘.

Coefficients of Selected Correlates of Counterargumen-
tation and Total Defensiveness . . . . . . . . . . . .

Mean Counterargumentation, Total Defensiveness, Pre-
treatment Attitude, Rating of the Logic of Persuasive
Materials, and Rating of Source' 8 Knowledge in Rela-
tion to Delay-Role Condition . . . . . . . . . . . . .

Mean Counterargumentation, Pre—test Attitude, and

Ratings of~the Bias of the Arguments, as a Function
of Sex and Delay in Auditory Feedback Interval . . . .

vi

Page

24

29

31

*33

34

34

35

37

~ 38

38

45

47

49

Table

14

15

16

17

18

19

20

Page

Summary of Analysis of Variance of Scores for Sup-

porting Arguments, Counterarguments, Total Defensive-

ness, and Total Number of Arguments--With Separate,

Listings for Each of the Four Scoring Methods. . . . . . 64

Mean Counterargumentation as a Function of Role and
Instructions to Counterargue . . . . . . . . . . . . . . 66

Mean Total Defensiveness as a Function of Order of
Measurement and Sex of Subject .'. . . . . . . . . . . . 67

Mean Acceptance of Counterattitudinal Arguments as
a Function of Role and Instructions to Counterargue. . .' 69

Correlates for Communication Acceptance, Counter-
argumentation and Total Defensiveness (Product-

moment correlations) O O O O O O I O O O I O O O I O O O O I 74

Mean Rating of Source's Knowledge as a Function of
Role and Instructions to Counterargue. . . . . . . . . . 77

Mean Subject-Ratings of Bias of Persuasive Material'
as a Function of Role and Sex of Subject . . . . . . . .' 79

vii

 

CHAPTER I

INTRODUCTION

A sizeable body of evidence has been collected which substantiates
the hypothesis that counterattitudinal-advocacy can be an effective
technique for modifying attitudes (see reviews by Collins, 1970; Elms,
1967; Insko, 1967; Sears and Abeles, 1970). More intriguing are the
findings which indicate that active participation in the expression of
counterattitudinal material leads to a significantly greater amount of
attitude change than does a more passive form of exposure to the countere
attitudinal material (Culbertson, 1957; Harvey and Beverly, 1961;

Janis and King, 1954; Janis and Mann, 1965; King and Janis, 1956;
Scott, 1957). This superiority of an active over a passive form of
participation in changing attitudes has also been found in terms of
temporal persistence (Elms, 1966; Mann and Janis, 1968; Watts, 1967).
Termed the role-playing effect, this tendency for role-playing to lead
to greater change than a more passive form of exposure is the primary
focus of the present research.

While the occurrence of the role-playing effect has been.we11
documented, numerous and repeated efforts to determine the mediating
mechanisms responsible for its effectiveness have met with much less
than complete success. In addition to the classical dissonance inter-
pretation (Festinger and Carlsmith, 1959) and a dissonance-based effort-
justification explanation (Zimbardo, 1965), improvisation (Janis and
King, 1954), biased scanning (Janis and Gilmore, 1965), satisfaction

1

2
(Janis and King, 1954), attention (Janis and King, 1954) and self-
perception (Bem, 1967) hypotheses have been put forth. Each of the
prOposed mechaniSms has been shown to be limited or inadequate in
certain respects. No single proposition has been able to account for.
all of the data with respect to the role-playing variable (Little, 1971).
The present discussion presents the possibility of an integrative solu-
tion and examines yet another potential mediator of the role-playing
effect.

Since the earliest investigations of counterattitudinal-advocacy,
.researchers have speculated about the critical role of defensive reactions
(such as counterargumentation) in the role-playing--attitude-change
relationship.l Indeed, in the early study by King and Janis (1956)
it was observed that

improvised role-playing might have been successful in

helping to overcome resistance by reducing the intensity

of those internal responses which normally interfere

with the acceptance of persuasive messages. (p. 183)

By interfering responses, the authors had in mind such things as concerns
about the consequences of accepting the discrepant material, concerns
about the trustworthiness of the communicator of the counterattitudinal,
role-structuring material, and counterargumentation. In an investigation
of an especially involving and life-like type of role-playing, Janis and
Mann (1965) attributed the persuasive effectiveness of the technique to
its ability to break through a person's "defensive facade." Elms (1966)

linked empathic fantasy ability to the increased effectiveness of active

as compared with passive forms of role exposure. The fantasy ability

 

1Terms such as "counterattitudinal advocacy," "role-playing,"

and "active participation" have been used, more or less synonymously to
refer to the behavior of a person who has been induced to attempt publicly
an authentic portrayal of a point of view contrary to his private
.opinions.

3
trait was presumed to act so that, in some manner or another, one's‘ own.
psychological defenses are circumvented, thus permitting cognitive
contact with the new attitudinal position.

In addition to various speculation, secondary data from several
studies are consistent with the notion that the persuasive impact of a
counterattitudinal performance is related to the manner in .which the
performance affects processes of psychological resistance. In studies
by Kelman (1953) and Cohen, Brehm and Fleming (1958) high justification
conditions. led to less attitude change than low justification conditions.
More important for the present discussion, high justification subjects
in both studies demonstrated more negative reactions to the manipula-
tions than did low justification subjects. While various aspects of
these studies force moderation inthe confidence of conclusions that
can be drawn from them, they do provide at least suggestive evidence
that the effect of role-playing procedures on defensive reactions will
determine the degree of attitude change. Further supportive evidence
18 contained in a study by Collins and Helmreich (1970) , who observed
that "consequence" subjects demonstrated less attitude change and a
~81‘eater tendency to dissociate themselves from their performance than
did "process" subjects.

Other evidence with a more direct bearing on the question, was
p“Ovided by‘ Greenwald (1969. 1970) who, in a series of studies, demon-
strated that the assignment to play a counterattitudinal role leads to
a less biased judgmental disposition with respect to the initially
counterattitudinal arguments. Furthermore, it was found that this less
biaSed disposition was followed by greater attitude change in the
direction of the assigned role. (This shift in evaluative disposition

was not found, however, when the counterattitudinal material was

4
presented just prior to role assignment.) Greenwald explicitly sug-
gested that the impartial disposition resulting from role assignment
led to less counterarguing and, consequently, more attitude change.

In spite of the suggestions and suggestive evidence, little has

been done to assess directly the mediating role of‘ defensive reactions
in the roles-playinguattitude-change relationship. Only the studies
by Greenwald (1969, 1970) were expressly conducted to gauge the extent
to which role-playing procedures disengage or moderate processes of
resistance. Especially in an area that remains as controversial as
role-playing, every alternative should be thoroughly considered. To
correct for the relative neglect mentioned above, the present research
was-developed to explore the role of defensive reactions in mediating
the attitudinal consequences of counterattitudinal advocacy.

While persuasion researchers have presented little systematic
discussion or documentation of the nature of psychological resistance
(Vi th exceptions, e.g. McGuire, 1964), the rough outlines of the terrain
have been noted. Freedman, Carlsmith, and Sears (1970) discussed
8ever-a1 psychological processes commonly related to resisting the
influence of a persuasion attempt, including: out-and-out rejection,
sour ce derogation, communication derogation, counterargumentation, and.
Perceptual distortion of the material in the persuasive attempt (a
coutrast effect).

That these defensive processes (or their absence) could influence
the outcome of a role-playing experience canbe easily appreciated by
cotlSidering the orientation of a person who, in accommodating the
recl‘-l:l.rements of a psychological experiment, the demands of his occupa-
tion orrsocial group, or legal prescriptions and prescriptions, expresses

a p0Bfition that is inconsistent with hisprivate convictions. It seems

5
reasonable to assume that the overall set of a person in such a situa-
tion would be to use the available information (whether provided by
the experimenter, normative communications, or formal laws) to ensure
the satisfactory performance of the role requirements. As a consequence
of this set, out-and-out rejection would be less likely since the person
could not totally disregard the material. Furthermore, opportunities
for counterargumentation and communicator and communication derogation
would be greatly curtailed.

How counterattitudinal advocacy moderates any or all of the above
mentioned defensive processes is yet to be empirically demonstrated.‘
The most likely defensiveness variable to be affected by-a verbal role-
playing sequence is counterargumentation. A person who has been asked
to present publicly only one side of a controversial-topic would
probably find it difficult to engage in any substantive amount of
covert, subvocal cognitive work inconsistent with the public position
being presented. At least there would be less opportunity for a person
to counterargue while actively constructing a new attitude position
than while being asked to do nothing more than receive.the counteratti-
tudinal message. Some research by-Osterhouse and Brock (1970) has
linked attitude change effects to counterargumentation inhibition and
may have special relevance for role-playing research.

Basically, Osterhouse and Brock (1970) explored the relationship
between distraction and communication acceptance. Experimental subjects
in.distraction conditions verbally monitored a sequence of flashes on
a panel of lights while they simultaneously listened to a persuasive
communication. As predicted, subjects in the high distraction condi-
tion produced fewer counterarguments than Subjects in the other condi-

tions, and were more favorable toward the advocated tuition increase.

6
It was concluded that distraction mediates attitude change by inhibiting
counterargumentation.

Of significance for the present discussion were the contentions
that in order for the distraction--acceptance relationship to hold, the
"distraction" should interfere with subvocal counterarguing but not
interfere with the reception of the counterattitudinal material.

If role-playing can be assumed to be a special case of the more
general category of events referred to as "distraction" then the rele-
vance of the Osterhouse and Brock study for an understanding of the
role-playing—-attitude change relationship is clear. Consistent with
the conditions noted for a replication of the distraction-acceptance
relationship, it would appear that role-playing procedures involve
subjects in activities which make vocal or subvocal counterarguing
extremely difficult. Furthermore, as research on the attention
hypothesis has shown (Janis and King, 1954; King and Janis, 1956;
Zimbarde, 1965), role-playing procedures do not seem to interfere with
the-reCeption of the counterattitudinal material. 4

From the preceding discussion, it is hypothesized that role-
playing, like distraction, drastically reduces a personIs ability to
counterargue and, as a result, makes the person more susceptible to
the impact of the attitude discrepant material. Thus, the major
hypothesis of the present investigation is that role-playing leads to
significantly less counterargumentation and, consequently, signifi-
cantly more attitude change than does a more passive form of exposure

to the counterattitudinal material.

CHAPTER II

EXPERIMENT ONE

One way of demonstrating the usefulness of the major hypothesis
of the present study is to show that it can.be used to explain the
results of previous role-playing investigations. Space permitting, a
number of studies of counterattitudinal advocacy could be reinterpreted
in terms of the effect of the manipulations on defensive reactions.
Results of Festinger and Carlsmith (1959), Carlsmith, Collins, and
Helmreich (1966), Linder, Cooper, and Jones (1967), and Zimbardo
(1965), for example, can be explained in terms of defensive reactions.
The choice here will be to attempt to replicate the Zimbardo (1965)
study. The effort-justification interpretation by Zimbardo has been
perhaps the least challenged of all the current explanations of the
role-playing effect. Interestingly, a recent empirical confirmation
of the effort-justification hypothesis has warned about an alternative
explanation in terms of counterargumentation. Linder and Worchel
(1970) commented:

The manipulation of effort in this experiment may have

had the concomitant effect of preventing subjects from

engaging in the use of counterarguments to reject the~

communication as it impinged upon them. (p. 444)

To assess something of the utility and robustness of the above
preposition, two investigations are presented. The first study is a

conceptual replication of Zimbardo (1965) and investigates the effort-

justification interpretation of the role—playing effect. The second

8
study explores other parameters which can be expected to moderate
defensive reactions.‘

In Zimbardo's (1965) eXperiment, subjects were either actively or
passively exposed to counterattitudinal material (participation condi-
tion) under conditions of either long or short delayed auditory feedback
(effort manipulation). Consistent with Zimbardo's predictions, sub-
jects in high effort conditions demonstrated greater attitude change
than low effort subjects. In extending this study, the most basic
modification is the inclusion of a measure of counterargumentation.
along with other dependent variable measures.

It is assumed that Zimbardo's manipulation of effort actually
established conditions varying the ease with which one could.counter-
argue. His results then were the effect of differences in ability to
counterargue rather than effort, per se. Thus, it is predicted that,
when Zimbardo's effort.manipulations are replicated, high effort condi-
tions will lead to less counterarguing and, as a result, greater attitude
change than low effort conditions. I

In the present study, half of the subjects were asked to present,
aloud, a speech based on a list of eight arguments against coeducational.
living. The other half of the subjects were asked to read aloud, as a.
speech, an essay of approximately 900 words containing the same eight
arguments. Subjects delivered their "talk" under conditions.of a .42-
second delay in auditory feedback or no delay in-auditory feedback. It
is specifically hypothesized that active participation in the exposure
to counterattitudinal material will lead to less counterarguing and.
more attitude change than a more passive form of exposure. It is also
predicted that delayed auditory feedback will lead to less counterarguing

and greater attitude change than no delay inrauditory feedback.

9
The general hypotheses presented above are detailed more formally
as follows:

HI: Improvising and presenting a speech from a list of arguments
which support an attitudinal position discrepant from one's
own leads to less counterargumentation than occurs from merely.
overtly reading as a speech an essay based on the same dis-
crepant arguments. '

HII: Improvising and presenting a speech from a list of arguments
which support an attitudinal position discrepant from one's
own leads to more attitude change than occurs from merely
overtly reading as a speech an essay based on the same dis-
crepant arguments.

HIII: Presenting a speech supporting an attitudinal position with
which one does not agree results in less counterargumentation
when there is a .42 second delay in.auditory-feedback while
speaking than when there is no delay in auditory feedback.

HIV: Presenting a speech supporting an attitudinal position with
which one does not agree results in more attitude change when

there is a .42 second delay in auditory feedback while speaking
than when there is no delay in auditory feedback.

Method

Overview-

A 2x2x2x2 factorial design was used to determine the influence of
type of exposure (improvising or overtly reading a counterattitudinal
communication), effort (.42 seconds delayed auditory feedback or no
delay), order of measurement of attitudes and counterargumentation
(attitudes before counterargumentation, or counterargumentation after
attitudes), and sex on the cognitive and affective consequences of role-
playing. Subjects were randomly assigned to the various treatment.
conditions in this pre—post design. Dependent variable measures were
collected on attitudes, counterargumentation. performance related

perceptions, and recall of the counterattitudinal materials.

10
Subjects
Eighty introductory psychology students (40 male, 40 female) were
recruited for experimental credits to participate in a project entitled
"Communications Research." Since the attitudinal issue to be used in
the study concerned coeducational living in the campus residence halls,
participation was limited to on-campus residents only. Only one subject

was present at each experimental session.

Procedure

Common to all subjects was a set of introductory comments. As-a
matter of first order, the subjects were given their experimental
credits. Next, in an attempt to highlight the salience of choice in
the situation, the experimenter delivered a few remarks about the
Psychology Department's policy with respect to research participation.
In brief, the subjects were told that they were perfectly free to leave
the study at any point that they found the procedures unpleasant or
otherwise intolerable. The experimenter added that, in any case, the
subject could retain the experimental credits, merely for having good

faith and showing up.

Read--improvise manipulation. Both active and passive participants
were told that the project was part of a program of research exploring
interpersonal communication. "Read" subjects were told, "What we'll
be looking at in the present study is the verbal presentation of topical
material." "Improvise" subjects were told "What we'll be looking at in
the present study is most commonly called improvisational speech."

After further elaboration common to both groups (See Appendices A
and B for complete details of the instructions) the experimenter digressed

from the experimental proceedings to ask for assistance with his efforts

11
to "find issues which are interesting to use in studies like these."
The-experimenter then administered a booklet containing 10 issues of
local campus concern to be rated on a number of semantic differential'

scales. This served as a pretest; one of the issues, "coeducational

I

living on the same floor where you live,' was used as a.premeasure of

the major dependent attitudinal variable.
"Read" subjects were told that they would be provided with some
materials on the issue of coeducational living and were instructed to:

read these arguments aloud in your natural argumentative
style as though you were presenting a speech. Try to be
as sincere and convincing as possible as you present the
material you will be given entitled: "A Position Paper:
Coeducational Living at Michigan State University." You
should try to be as natural as possible while reading the
prepared material. You will be given three minutes to
rehearse the speech silently."

Improvise subjects were similarly told about the materials and were

instructed
to use the arguments to present a speech in your natural
argumentative style. Try to be as sincere and convincing
as.possible in presenting the same point of view as the
one taken in the material you will be given entitled: "A
Position Paper: Coeducational Living at Michigan State
University." Use all the arguments listed and feel free
to add any other arguments you want as long as they are
consistent with the ones that have been provided. Try to
be as comprehensive as possible when presenting the speech.

You will be given three minutes to rehearse the speech
silently.

The experimenter suggested to all subjects that they try to imagine that
they actually were a person with beliefs like those expressed in the
material on coeducational living, and further, that they were speaking
to a small group of peOple eager to hear what they had to say about the
issue. Subjects were reminded to be sincere and convincing and told to
begin speaking when signaled at the end of the three minute preparation

period.

12

Performance-structuring counterattitudinal materials. "Improvise"

 

subjects were given a list of eight arguments to support the position
that male and female occupation of the same residence hall would be
undesirable. The arguments on the single sheet of paper headed "From
A Position Paper: Coeducational Living at Michigan State University"
were: (1) more inconvenience (noise, privacy), (2) less adequate study-
atmosphere, (3) an increase in room and board fees because of the cost
of altering bathroom facilities, (4) agitation among many parents and
other taxpayers, (5) more artificial male-female relationships (have
to be "on-stage"), (6) more sexual incidents (pregnancies and physical
assaults), (7) more difficult to get many types of hall activities
going, and (8) there are enough adjustments to make as it is.- "Read"
subjects were presented with an essay which was ostensibly a student
position paper.‘ This essay of approximately 900 words contained in
elaborated form the same eight arguments provided to "improvise"

subjects.

Effort manipulation. Zimbardo (1965) manipulated effort by varying
the interval between speaking and hearing one's own comments. The .25
second delayed auditory feedback (DAF) condition in Zimbardo's study
created a considerably more effortful task than the .01 second DAF
condition. Equipment available for.the present research permitted
feedback delay intervals of .OO, .21, .27, .42, and .54 seconds. Pilot
Work suggested that either .21 or .27 delay interval would lead to
BUbstantial "dropout" problems (subject mortality). However, the .42
Second interval did produce disruption of speech patterns and higher
Perceived difficulty ratings thanthe .OO delay interval. - Thus, the

'OC’ and .42 delay intervals were judged to be suitable parallels to

13
Zimbardo's manipulations and were used in this study to define effort
operationally. The .42 second delay condition was the more effortful.
Shortly before distributing the material on coeducational living,
the experimenter cautioned each subject not to be overly concerned
about the "modifications of your voice by the reCording equipment."
subjects in the .42 delay condition were further told: "You will

notice a slight time lag in your speech which might be disturbing at.

first."

Manipulation of order and measurement.. The final manipulation
involved the order in which the attitude and counterargument measures
were taken. Half of the subjectsjwere randomly assigned to indicate
their attitudes before a measure of counterargumentation was obtained.
The other half of the subjects were given an Opportunity to express
counterarguments before they were asked about their attitudes. While
there is no apparent reason for assuming that the order of measurement
should make a difference in this situation, there is also no strong
confidence that the order of measurement will be inconsequential.

There is no precedent in this case; there are no investigations of
role-playing which have also looked at counterargumentation.

An additional reason for examining order of measurement has to do
with Miller and Baron's (1971) warning about the ambiguity surrounding
causal interpretations of the relationship between attitude change and
CGUnterarguing:' Miller and Baron (1971) correctly noted that even.
With perfect measures it would not be possible to establish the tenporal
Priority of counterarguing. Thus it is possible that counterarguing is
the consequence and not the cause of attitude change. Perhaps differ-

ences in order‘of measurement will shed some light upon'the complexities

14
of this problem. In any case, at this primitive stage. an exploration

of the potential of this manipulation seems justified.’

Dependent Measures

 

At the completion of the subject's speech, the experimenter dis-
tributed a four-page postexperimental questionnaire. The pretest
measure of attitudes had been distributed before subjects were given

the role-structuring counterattitudinal material.

Measures of attitudes. Semantic differential scales were used

 

to measure attitudes toward the critical issue, "coeducational living
on the same floor where you live." The bipolar adjectives used on the
pretest were: good-bad, beautiful-ugly, sweet-sour, nice-awful,
fair-unfair. These bipolar adjectives were mixed with six other sets
of bipolar adjectives. On the post-test, a parallel set of bipolar
pairs was employed. The five sets of scales loading on the evaluative
dimension were: honest-dishonest, clean-dirty, pleasant-unpleasant,

fragrant—foul, kind-cruel.

Measurement of counterargumentation and other thought processes.

As is to be expected with newly deve10ped measures, procedures for
eliciting and categorizing the subjectsf previously covert thought
processes have not been standardized. A number of different approaches

to this measurement problem appear to be appropriate within the bounds

of agreement which have already develOped. In recognition of this

lack of consensus, the present study has incorporated several alternative
te<2hniques for assessing counterargumentation and other thought processes.
Many ,of the decisions with respect to this problem have been influenced
bYthe research by Osterhouse and Brock (1970) and the excellent

theheretical paper by Miller and Baron (1971).

15

It will be recalled that, in general, the basic mediating variables
in this study are assumed to be covert processes which occurred during
the reception and encoding of the counterattitudinal material. Counter-
argumentation is the type of thought process of most direct interest
in this research. In establishing a valid measure, it is essential-
that the instrument reflect rather than create these thought processes.‘
As Miller and Baron (1971) have warned, "it is quite possible that the
measurement attempt itself and not the earlier anticipation of or
actual exposure to a persuasive communication instigates counter-
arguing" (p. 4).”

One suggestion for meeting the problems of reactivity is to use
instructions which do not specifically ask the subject to produce
counterarguments.' To this end, the instructions for the counterargu-
mentation measure were modeled after those used by Osterhouse and
Brock (1971) and stated:

While reading over and presenting the material concerning

coeducational living at MSU you probably had many dif-

ferent reactions to the various ideas. List below any

of these thoughts - either pro or con - which may have

occurred to you. State your own thoughts and ideas as

briefly as possible--IGNORE SPELLING, GRAMMAR, AND PUNCTU-

ATION. You will have three minutes to write your ideas.

Please stop writing immediately and go to the next page
when asked to do so.

Paralleling the procedures of the Osterhouse and Brock study, 15
7-inch horizontal lines followed the instructions and filled out the
remainder of the page. , Each line was preceded by the word "idea."

To further ensure that those thoughts which are listed by the
Bul>Jects were the same ones which spontaneously occurred to them during
the processing of the counterattitudinal message, a limit was imposed
on the amount of time given to record prior thoughts. The assumption

is Inade that the more time provided to list thoughts, the less the

'3:

iv!

.L

16
recorded thoughts wOuld correspond to those actually evoked during
the initial processing of the counterattitudinal material. "Readily
available" as opposed to "spontaneously used" ideas (to use a dis-
tinction of Miller and Baron) would be more frequent. With unlimited
time, the differences between subjects on a thought listing_measure
would be more a function of the subjects' initial motivation. atti—
tudes, and knowledge about the issue. On the other hand. the less
time available, the more likely it is that the previously occurring
thoughts would be listed since presumably they should be more salient.
under such circumstances, those‘subjects who had been counterarguing
during-the.reception of the persuasive material would be the ones
most able to reproduce these thoughts.

A question is created, then, as to the most appropriate time
limit to employ. Clearly, there are practical limitations on the
time restriction. For instance, in all probability a limit Of 5 seconds
would not be sufficient to detect actual treatment effects. The Optimal

time interval (where Optimal is defined in terms of the measure's
sensitivity to theoretically appropriate treatment differences) can

only be empirically determined. Osterhousewand Brock (1970)*used a

three minute limit and noted that a one minute period was not sufficiently
sensitive. Miller and Baron (1971), in commenting on some of their
previous research, indicated that a 4S-second interval was sensitive to.
exPerimental manipulations.’

In the present study, measures with respect to two time periods
wereobtained.’ While subjects were given.three minutes within which to
list their ideas about the coeducational living issue, the experimenter
also obtained almeasure of counterarguing after .45 seconds. To do this,

the: experiment.unobtrusively glanced over the subject's shoulder after

17
45 seconds to determine which idea line number the subject was working
on at the time. Consequently, it is posSible to compare the pattern
of treatment effects for the 45-second measure with those for the
3-minute measure.“

The criteria for scoring counterarguments were adapted from Brock
(1967), Osterhouse and Brock (1970), and Mihler and Baron (1971). An
idea was scored as a counterargument if it was a declarative statement
which mentioned unfavorable or undesirable consequences of not having
coeducational residence hall living. Refutations of arguments and
prOposed alternative solutions to the coed living plan were also scored
as counterarguments. Consistent with the recommendation of Miller and
Baron (1971), certain rhetorical questions were classified as counter:
arguments (i.e. rhetorical questions which seemed to challenge the
validity or accuracy of an idea presented in the persuasive materials).

A statement was not scored as a counterargument if it was a state-
ment of simple support (coeducational living is undesirable), an affect-
ive-reaction (those arguments make me mad), or an attack on something
other than the arguments against coeducational residence hall living

(if people would only let Jesus into their lives, we wouldn't have
these problems). -

Brock (1967) had recommended that only one argument be counted per
"idea" space.‘ However, subjects are generally untrained in the use of
the instrument and might list more than one idea per.space. Thus,
although the structure of the measure was set up for only one idea,
morethanjone idea could be counted per space.

An_ambiguous response to the counterargumentation measure is an
idea which is repeated one or moretimes. It is possible that a subject

cOllildLhave been perseverating on an idea or looking at a single idea

18
from several perspectives. Repeated thoughts of this nature would seem
to constitute more of a particular type of response than a single idea
and thus should be weighted and counted apprOpriately. On the other
hand, a repetition may indicate that the subject merely tried to
clarify a thought previously written down. Or it" may mean that the
subject was stuck for an idea and repeated himself while trying to think.
of something different.

In the study by Brock (1967), ideas which appeared to be a repe-
tition or mere rephrasing of an "idea" previously listed were not
counted again. In the present study, a dual system was employed. In
doing the content analysis of the item tapping subjects' thoughts,
frequency counts for each variable were made both including and exclud-
ing repeated thoughts. This procedure permits a comparison of the
utility of the two different assumptions by examining the sensitivity _
of the two (measures to the varioustreatments.

Thus far, the discussion has focused oncounterargumentation.
There are, however, other types of likelyreactions to a persuasive
message. For one thing, source and communication derogation are con-.
sidered to be, at least conceptually, distinct from responses of-counter-
argumentation (Aronson, Turner, and Carlsmith, 1963; Brewer-and Crano,
1968; Festinger and Maccoby, 1964; Freedman and Sears, 1965; Freedman,
Carlsmith, and Sears, 1970) . A response of general disagreement may
be yet another independent type of resistance. In agreement with
Miller and Baron's suggestion concerning defensive reactions. and in
1Line with existent conceptual distinctions in the persuasion area,
8eparate.cateiigo‘Irries were created for source derogation, communication
derogation, general disagreement, and counterargumentation. Other

categories existed for arguments supporting the persuasive material,

19
for ambivalent comments, and for comments irrelevant to or unclearly
related to the attitude position in question. These seven categories
exhaustively partitioned the response domain.
The-rules for assigning thoughts to the counterargument category
were previously notedg'brief descriptions of the other categories

follow below:

(1) General disagreement--any statement rejecting the main thesis of
the arguments without mentioning specific reasons (e.g. "I didn't
agree with any of the ideas.").

(2) Source derogation--any negative comment about the author(s) Of the
material about coed living (e.g. "The peOple who wrote this are weird.")

(3) Communication derogation--any negative comments about the struc-
tural aspects of the counterattitudinal materials (e.g. "These arguments
were inconsistent with each other.").

(4)- Supporting arguments--any statement consistent with the intent of'
the essay.' This category for positive statements was not further sub-
divided as was the case for negative statements.

(5) Ambivalence--any single sentence indicating mixed feelings about
coeducational living (e.g. "There are some good points and there.are
some bad points.").-

(6) Unclear--any comment not related to the issue of coeducational
living or to the performance with respect to this issue. Generally,
anything not codeable into any other category.

Inter-rater agreement was employed to assess the reliability of the
thought-listing instrument. Two judges, working independently and with-
out knowledge of the subject's treatment condition, initially agreed on
the placement of 92.5% of the responses to the thought listing question.
A consensus was reached between the judges concerning the appropriate
Classification of the remainder of the responses. (For each of the fourv
Scoring systems, Appendix C lists the frequency of occurrence of each
of the response categories.)

There are several advantages to the-scoring system used in the

Present study. Of particular merit is the fact that the system makes

20
it possible to determine whether the several types of defensive reac-
tions are differentially responsive to the particular types of manipula-
tions employed. That is, for example, counterargumentation may be the
typical mode of response for one set of circumstances while source and
communication derogation may be used for others.'

In the present study, roughly 8% Of all responses were described
by the categories of general disagreement. source derogation, and
communication derogation. Since so few Of the defensive responses-
fell into categories other than counterargumentation, analyses were
completed for only two measures of psychological resistance. One
measure is of counterargumentation alone; the other, referred to as
total defensiveness, included source derogation, communication deroga-
tion, and general disagreement, in addition to counterargumentation.l
While the total defensiveness measure is heavily loaded with counter-
argumentation, it is of interest to determine if the broader measure
is more sensitive to the treatments. It is also possible to determine
if the supporting argument category is the mirror image of the defensive-
ness categories. Conceivably, responses consistent with and supportive
of the persuasive message and responses antagonistic to the persuasive
material are or can be independently influenced by various treatment
combinations.

The reader should be forewarned that the wealth represented by the
number of options for scoring the thought processes (evoked by persuasive
material) is not without its liabilities. When the dependent variables
are.analyzed, the number of comparisons to be made (and potentially the.

alnount of confusion) are manipulated correspondingly.

21

Other dependent variable measures and manipulation checks. In
order to assess the effort manipulations and to test other interpretav
tions Of the role-playing effect, other items in the questionnaire asked
subjects to rate: (1) the sincerity and convincingness of their per—
formance; (2) the extent to which the arguments they presented were
logical, and biased; (3) how difficult the performance was and how
much effort and involvement it entailed; and (4) how satisfied they
were with their performance. Subjects were also asked to evaluate
the expertise and trustworthiness of the author of the role-structuring
materials. The.last question was a.measure of recall of the supporting
arguments.‘ (See Appendix D for an example of the pretest measure of
attitude toward coeducational living. Appendix E contains the materials

included in the postexperimental questionnaire.)

Results
The purpose of this replication was to demonstrate that attitude
change differences which Zimbardo attributed to effort-justification
induced dissonance reduction could more accurately and efficiently be
eXplained via treatment effects on defensive processes such as counter-

argumentation.

Assessment of the Effort Manipulation

Two questions were used to index self-perceptions of the effort
involved in the counterattitudinal performance. In terms of the first
question, (Q1): "How difficult was it to present the speech?", both the
delayed auditory feedback (DAF) and the improvisation manipulations had
the intended effect. Subjects who performed under .42 second delay in=
auditory feedback reported substantially more difficulty (p=.0005) in

Presenting the speech than those who performed under no delay in.

22
auditory feedback. "Improvise" subjects reported marginally more
(p-.O6) difficulty than did "Read" subjects. These findings are in
accord with the experimenter's Observations with respect to stuttering,
body posture, foot.shuffling. handdwringing, sighs of relief upon
completion of the task, and.other similar subject behaviors. Subjects
in the-.42 second delay condition were quite visibly under more pressure
or tension than subjects in the no delay condition. To a slightly,
lesser degree, the differences between read and improvise groups
were.apparent.

,The data with respect to the second question (Q2) cast a certain
ambunt of ambiguity over the just mentioned findings. In terms of the
question, "How much effort did you put into presenting the speech?",
there were no significant treatment effects. All subjects expressed a
high level of effort; the mean for all subjects on the lO-point scale.
‘was 7.13. The means for .42 second delay, no delay, and improvise and
read conditions were, respectively: 7.08, 7.20, 7.25, and 7.03. The
correlation between the two questions intended to index the same
phenomenological state was a nonsignificant -.118.

Perhaps the responses on this second question indicate a ceiling
effect. Alternately, the data may indicate that the subjects found the
question ambiguous. It may have been that the subjects initially found
the second question highly similar to the first one asking about diffi-
culty; but, since there were two questions, decided the second must be.
inquiring about a different aspect of their performance. The subjects
may have perceived the firsthuestion as related to physical effort
and the second to psychological. Zimbardo (1965) reported significant
differences for physical effort (p < .001) but more attenuated effects

for psychological effort (p < .10).

 

'

«.2

23

In view of the high degree of similarity between the present study
and Zimbardo's, it seems unlikely that the kind and amount of effort'
elicited in the two studies would be radically different. As in they
Zimbardo study, all subjects were given a few minutes within which to
rehearse their speech. In both studies "read" subjects read a prepared
essay and "improvise" subjects worked from a prepared list of eight
arguments. Also, both studies involved a comparison of performance.
under high- and low-disruptive delayed auditory feedback.

If the results Of the direct question (02) concerning effort are
taken at face value, what modifications Of Zimbardo's procedures could
be responsible for the fact that Zimbardo found differences in per-
ceived effort while there are none in the-present study? Actually,-in;
the former study effort ratings differed only between those subjects
receiving less than .01 second delay between speaking and hearing and.
those subjects receiving a .25 second delay.’ Zimbardo found no dif-
ferences between "read" and "improvise" subjects. As mentioned in the
method section, the delayed auditory feedback (DAF) manipulation of
this study was established by having half of the subjects perform under
a .42 second delay, while the other half of the subjects performed
under.direct feedback. While the difference between DAF conditions in
the present study may have been Of a lower magnitude than that in the
Zimbardo study, the effect of the difference between DAF conditions in
the present study still appears to be substantial.

subjects may have been puzzled by the second question on effort
and wondered what it was asking that was not involved in their answer
to the first question. While the situation may have been an ambiguous
one for them, they responded to it in something, other than a random

fashion, The support for this observation can be obtained by examining

24

the correlations of the two effort questions with other self-ratings.
If the subjects were totally confused by the second effort question:
(02) and responded arbitrarily, then this second question should not
be very highly correlated with other items. As Table 1 indicates, the
responses tO both effort questions were substantially related to
responses to other items. It appears that both items were reliably
measuringnsomething.

Table 1

Correlations of "Difficulty" and "Effort" Ratings
with other Self-perceptions

 

 

Difficulty in Effort You Put
Correlation Presenting Presenting

between Speech Speech,~
Satisfaction with task
fulfillment -.493* .337*
Performance rated
relative to others -.552* .338*
Ability to get involved
with task, -.388* .504*
Sincerity of performance -.080 .469*
Convincingness of performance -.540* .381*
Initial confidence in

 

Note.--The coefficients in this table are based on the total sample of
80 subjects.‘

a
p < .01.

Since some resolution of this problem is necessary, the suggestion
is hazarded that the first question. asking about difficulty, is in fact

a valid measure of Zimbardo's concept of effort. When responding to

25
this question, it is assumed that subjects were indicating how hard
they had to work in presenting the speech.’ When faced with the second
questiOn ("How much effort did you put into presenting the speech?"),
subjects probably assumed that something different was involved. Since
the question specifically focused on their personal contribution (by
emphasizing the word you), subjects probably were indicating the extent
to which they were able to perform unencumbered by aspects of the
experimental materials and manipulations. That is, subjects were indi-
cating hOw free they were to innovate with respect to content and/or
style of delivery. The less "difficulty" (physical trouble) one had
in delivering his speech, the more he should feel as though he had the
freedom to innovate. The more "difficulty" one had in merely deliver-
ing the speech, the more he should feel that his ability to innovate
was limited. The.more difficult the conditions, the less he should
feel that he was actually able to add to the already structured per-
formance. The tendency for subjects confronting high difficulty to
devalue their creative efforts and for subjects confronting low diffi-
culty to enhance-their elaborative efforts would result in a homogenizing
effect on the second effort question. Consistent with this specula-
tion, the second effort item did have the lowest variability of all 11
of the manipulation-check items.

The patterns of the correlations between the two effort questions
and other self-perceptions can be seen as supportive of this speculation
concerning the meaning Of the items (see Table 1). With respect to the
item asking about a.subject's "ability to get involved," it appears that
the more a person felt he was able to get involved in delivering the
speech, the less difficulty (r.-,333) and the more effort (translate,

ability to innovate), (r-.504) he reported. The finding that initial

26
confidence in one's Opinions was negatively correlated (-.373) with
ratings on the "difficulty" item but positively correlated (.312) with
ratings on the "effort" item is also consistent with the above specu-
lation about the subject's interpretations of the meaning of the two
items. Consistent, that is, if it can be assumed that subjects with
less confidence in their Opinion would be more threatened and, thus,-
feel more constricted in their performance.

Whether the manipulations in this study created differences in.
degree of effort comparable to those of the Zimbardo study is somewhat
ambiguous. In view of the fact that the procedures of the two studies
are highly similar and in view of the data relating to the subject's
rating of difficulty in delivering the speech, it is assumed that
effort was appropriately and substantially manipulated.“ The assumption.
that the subjects interpreted the meaning of the second effort check item
in terms of ability to perform in an unconstrained, unfettered fashion is
reasonable post hoc speculation. Certainly, the disparate findings with
respect to the treatment effects on the response to the two effort ques-
tions forces a degree of reserve in any interpretations concerning an
effort-justification hypothesis. What this means, unfortunately, is
that it will not be possible to refute or to support in any conclusive

manner-the role of effort in role-playing induced attitude change.
Tests of Hypotheses

Counterargumentation production. Before considering the hypothe-

 

sized effects of Role (Improvise versus Read) and Delayed Auditory.
Feedback (.428econd Delay versus No Delay) on counterargumentation,
'it is desirable to recall that the measurement of the thoughts evoked

by the persuasive material was-carried out with respect to four different

27
sets of procedures. That is, counterargumentation, as well as total
defensiveness and supportive argumentation, were assessed according to
four different though related sets of rules. These four variations
were created by:~ (a) tallying the number and type of thoughts listed
after 45 seconds and after 3 minutes, and (b) tallying the thoughts
with repeated ideas either included or excluded.)

At this stage in the develOpment of a technique for measuring
counterargumentation and similar covert processes there is a_painfully
noticeable lack of consensus as to the most apprOpriate procedures.

This is especially unfortunate in view of the fact that the different
scoring systems employed in this study are sensitive to different.
treatment effects. A comparison.of the procedures clearly indicates
that the differences that do exist among the scoring systems are not

due solely to such psychometric considerations as test length. In

spite Of a high degree Of overlap in terms of the sensitivity of the,
different techniques to the treatments, it is apparent that different.
conclusions might be drawn depending upon which of the different scoring
systems was chosen for interpretation.

In light of this unavoidable ambiguity, a somewhat flexible policy
with respect to interpretation Of*effects has been followed. Generally,
more confidence has been placed in the validity of an effect, the more‘
different systems of scoring detect it and the lower the probability~
level associated with its chance occurrence. This has led to the
neglect in some places of findings which have only been established
with respect to one scoring system or in terms of marginal probability
levels.‘ The effects receiving emphasis are those which appear to be
the strongest, most consistent, and/or theoretically interesting. While

this is certainly not the most desirable state,'it does seem justified

28
in view of the current status Of the construct validity of the measures
in question and in view of the probable return on the time invested.

Table 2 contains the results of the analysis of variance for four.
different dependent variables: supporting arguments,counterarguments,
total-defensiveness, and total number of arguments.‘ Table 2 also shows
the results for each of the four separate scoring systems. The data
are displayed in this manner to facilitate the simultaneous comparison.
of the treatment effects on the several dependent variables and of the
different scoring systems.

It is desirable to take time for an explanatory word or_two con-
cerning Table 2 before turning to the hypotheses—-a moment to render
the table more intelligible and less frustrating. Consider, for
instance, the cell representing the effect Of order of measurement on
supporting arguments. When repeated arguments were included in the .
tally for the 3 minute limit (3 min. RAI), the resulting F ratio for
Order of Measurement had a probability of occurrence of less than or
equal to seven times in a thousand (.007). The probability of the F‘
ratio when repeated arguments were included with a 3-minute limit was
.018. The results are highly similar for the other two techniques. To
take just one other example, the effect of Role on counterargumentation.
was reliable (p,< .05) only when the 45 second limit on thought listing
was imposed and repeated arguments were included.

It was predicted that subjects would be less able to counterargue
when:g (a) improvising a counterattitudinal speech than when reading a
prepared counterattitudinal speech, and (b) when performing under a .42
second delay in auditory feedback (.42 Delay) than when performing with
no delay in auditory feedback (No Delay). In addition to Role and

Delayed Auditory Feedback, Sex of subject and Order_of Measurement of

N vi — .mIV.-.

29

.Amom .uon mwv wovoaoxo nuooaowuo communes mafia udaaa mafia voooon o>«MIhuuom

'U

.Am4M .oHa nV wovnaoxo nuooaowuo venomous suaz uwaHH mafia ouoofia mouse
.AHam .oon an cocoaoow muooaowum coupons“ sows uHaHH moan ooooom o>amuhuuom

“3.0 U

.AHam .aas nV essences muamsamum umumuauu nous usage menu ausaaa_mmuas

. .uuumsa one on. w a
mafia m.m Had you mHo>oa.huHHanonoun may .voumwa moo .ouomouonu .ouo3 com GOfiuouovwmooo noon: monommoa
ucoooonov man so nuoommo on on: oOOfiOOoHOuofi a x o N m can .9 x m x < .0 x m x < .n x o .o N m o£HII.ouoz

 

 

«AAAo.V a a o x.m x <
ofimno.v
uAmmo.VmAaeo.V nAemo.V erAo.VmAm~o.V a x o x a
erNo. Vo Ameo. V eAHHo. VU Amao. V
nANso. V“ ANHo. V n.AmS. Vo mAmoo. V o x m
eANAo. VM Aemo. V ereo. Vo «AAOH. V
nANeo. V“ Aer. V nASAV. Vo mAomo. V a x a
«Aqmo. V uAHoa. Vm mAaS. V o x a
oAaao. Vm Ammo.V wAmeo. Vm «Acme. V m x a
WANAo. VM Ammo. V nAANO. V oAmHo. V
nAmao. VM Ammo. V nAmoo. Vm oAAAo. V a Aumuro
UANNOV nAAeo.V o AoHomV
m AamAmnV
onso.V a AxomV
nuooaowu< snowmen annoy muooadwuououosoo muooaowu< oowuofiuo>
annoy woauuonqom mo oouoom

 

unmanned mo ooze

 

moonuoz wsauoom noon man no zoom mom omowuoaa ououooom nuwzwumuoosomu< mo nonaoz annoy new
mmooo>finoomon Houoa .muooaowuououoooo .muaoaomu< wowuuomgnm How nououm mo ooooauo> mo owanoo< mo mumaanm

N 0.3mm.

3O

'Attitudes and Counterargumentation were included as independent'
variables in the 2x2x2x2 analysis of variance conducted on the various
(dependent measures. Assuggested by Table 2 the hypotheses were not
supported. There was no effect of Delayed Auditory Feedback (DAF) on
counterargumentation or total defensiveness. This result replicated
across the different scoring systems. The effect of Role on counter-
arguing was. detected only with the measurement system which counted
repeated arguments and used the 45-second limit (F - 2.81, df - 1,64,
p i .047). Contrary to predictions, "improvise" subjects listed more
counterarguments in the first 45 seconds than did "read" subjects.*

No effects of Role were detected when the other scoring systems were
employed. Regardless of the scoring system employed, Role had no
effect on the total number of defensive arguments listed. In sum, the
predicted differences in counterargumentation did not materialize. If
anything, there was tendency for the role manipulation to have an
effect Opposite to that expected.

An analysis of the significant interaction of Role and Delay on
counterargumentation (Role x Delay interaction for 3 min. REA system,
P - 5.91, df‘- 1,64, p - .018) further illuminates the noncorrespondence
of the data and the-predicted relationships. As shown in Table 3,2 the
condition which should have resulted in the most counterargumentation
(the No Delay--Read condition) actually resulted in the least counter-

argumentation. However, the condition which theoretically should have

 

2In Table 3 and throughout the remainder of this section, tables
containing.counterargumentation and total°defensivene§s measures are
based on the scoring systemuwhich used the-threeéminute limit and excluded
repeated arguments (3 min. RAE). As will be clarified later, all four,
of the scoring systems produced highly similar results. The various
methods of scoring were virtually equal in terms of sensitivity to treat-
ments.‘ The reason for prefering the 3 min. REA system is that this
measure demonstrates a sensitivity to effects which most directly cor-
,responds to the treatment effects on attitude change.

31
Table 3

Mean COunterargumentation as a Function of
Delayed Auditory Feedback Interval and Role

 

 

Delayed Auditory: Role
Feedback Interval Read - Improvise
.42~Second.
Delay 2.80 2.25
No Delay 1.80 3.15

 

Note.--The sample-size for each cell is 20.

permitted the least amount of counterargumentation (the .42 Delay--
Improvise condition) did result in a relatively low level of
counterargumentation.

An analysis of variance of simple effects reveals that improvisa-
tion led to more.counterargumentation than reading under the no delay.
condition (F - 5.97,. df - 1,64, p < .05)--an effect inconsistentwith ~
predictions. The difference in counterargumentation between improvise
and read conditions was in the direction predicted for subjects per-
forming under the .42 second delay condition. This difference was,
however, not significant (F - .959, df - 1,64). Controlling role, it
can be seen that the delay manipulation had a tendency to work as pre-
dicted for improvise subjects (simple effects F = 2.656, df - 1,64).
The .42 second delay produced less counterargumentation than no delay.
On the other hand, "rea " subjects had a tendency to counterargue more
under the .42 second delay than under the no delay (F - 3.279, df -.
‘l,64) condition. Again it is seen that the hypotheses cannot account

for the counterargumentation data.

32

In addition to the Delay-Role interaction other significant
effects in terms of counterargumentation and/or total defensiveness
occurred for the main effect of Order Of Measurement and.the Sex-Delay,
SexrRole, and Sex-Order interaction (See Tables 2 and 4). Table 4
has been added to give a clearer picture of what treatments and treat—
ment combinations had an impact on counterargumentation and total
defensiveness.

Subjects who indicated their attitudes after they were given a
chance to express counterarguments demonstrated more total defensive-
ness than did subjects who indicated their attitudes first (F - 4.79,.
df'- 1,64, p - .032). This difference in total defensiveness appears
to be closely related to the effect Of order of measurement on the
amount of supportive argumentation. As Table 5 indicates, those
subjects who expressed their_attitudes before counterarguments (A1)
listed significantly more supportive arguments than did subjects who
expressed their attitudes after being given an Opportunity to express
counterarguments (A2) (F'- 5.867, df - 1,64, p - .018). Table 5
illustrates that the number of supportive arguments listed was inversely~
related to the-number of defensive thoughts listed.‘ Order of Measure—
ment did not have a substantial effect on.counterargumentation.

Even though the Seerrder of Measurement interaction is of only
borderline significance for three of the four scoring systems (for all
but the 45 sec. RAI system), this interaction is quite important for any
interpretations of the main effect of order of.measurement. The cell
means.for the sex-delay interaction are presented in Table 6. Order
of Measurement had no effect on men (simple effects F'- .125, df-- 1,64);
women had fewer defensive thoughts when receiving the Al order than

when receiving the A2 order (simple effects F-- 7.523, df = 1,64,

33

Table 4

Summary of Analysis of Variance of Counterargumentation

and Total Defensiveness

 

Counterargumentation.

 

Total Defensiveness

 

 

Source MS F MS F
Sex‘ (A)' 1.25 <1 2.81 <1
Delay (B) .05 <1 .01 <1
Role (C) 3.20 1.05 2.11 <1
Order (D) 5.00 1.64 15.31' 4.79**
A x B 12.80 4.20** 19.01 5.95***
A x C 8.45 2.77* 6.61 2.077
A x D 8.45 2.77* 9.11 2.85*
B x C 18.05 5.92*** 13.61 4:26**
B-x D .05 <1 .31 <1

C-x D .20 <1 1.51 <1

A x B.x C 7.20 2.36 5.51 1.73
A x B x D .20 <1 .31 <1

A x C x D 4.05 1.33 2.81 <1

B x C'x D 1.80. <1 .61 <1

A x B x C x D 3.05- <1 .31 <1

 

*
p < .11, df - 1,64

*1:
p < .05, df - 1,64

***
p < .02, df-- 1,64

34
Table 5

Mean Total Defensiveness and Number of Supportive Arguments
as a Function of Order of Measurement

 

Attitudes Measured

 

 

Before After,
Type of Thought Counterargumentation Counterargumentation
Total Defensiveness. 2.47 3.35
Supportive Arguments 2.45 1.42

 

Note.--The sample size for each cell is 40.

p < .Ol)x Apparently, the main effect of order Of measurement is due
to female subjects who expressed their attitudes before listing their

counterarguments 0

Table 6

Mean Total Defensiveness.as a Function of‘
Sex of Subject and Order of Measurement

 

Attitudes Measured

 

 

Before After
Sex of Subject Counterargumentation Counterargumentation
Female. 1.95 3.50
Male 3.00 3.20

 

Note.--The Sex x Order of Measurement interaction F - 2.85; df-- 1,64;
p < .10.' The sample size in.each cell is 20.

35

The interval of delay in auditory feedback seemed to interact with
the sex of the subject in terms of its effect on counterargumentation.
For male subjects, the .42 second delay led to less total defensiveness
than the no delay treatment (Table 7). This is consistent with the
hypothesized effect of the DAF manipulation (simple effects F - 3.13,
df'- 1,64, p <.10).‘ For females, there was a tendency for the .42
delay condition to lead to more total defensiveness than the no delay
condition--a result contrary to the predictions concerning the delay
manipulation (simple effects F-- 2.86, df - 1,64, p < .10). The.pattern
of means for counterargumentation is highly similar to that for total

defensiveness presented in Table 7.

Table 7

‘Mean Total Defensiveness as a Function of Sex of
Subject and Delayed Auditory Feedback Interval

Delayed Auditory Feedback Interval

 

 

Sex of Subject NO Delay .42 Second Delay
Female 2.25 3.20
Male 3.60 2.60

fi'

Note.--The sample size in each cell is 20.’

The Sex-Role interactiOn seems to be the least stable of all the
effects. Using the three-minute limit with repeated arguments included
(3 man. RAI), the mean number of counterarguments for female "read" and
"improvise" subjects were, respectively, 2.70 and 2.25. The means for

males in the read and_improvise conditions were, respectively, 2.40 and

36
3.90. The interaction seems to be due to the fact that male-improvise
subjects counterargued to an especially high degree.‘

To summarize: (a) contrary to predictions, delay in auditory
feedback did not have an overall effect on-counterargumentation, (b)
Role (Improvise versus Read) did not affect counterargumentation as
predicted, (c) Delayed Auditory Feedback influenced counterargumentation
as predicted for improvise subjects, (d) the attitudes before counter-w
argumentation order of measurement led to more supportive argumentation
and to less total defensiveness than did the reverse order of measure-
ment, (e) Sex of Subject interacted with Delay, ROle, and Order of
Measurement, although some of these interactions were of only marginal

significance.

Attitude chang_.' The major hypotheses Of this study predicted
that the delayed auditory feedback (DAF) variable and the role-playing
variable would influence the amount of attitude change. Specifically,
it was predicted that improvising a speech should lead to more attitude
change than merely reading a prepared speech. A .42 second delay
(.42-Delay) in auditory feedback between speaking and hearing was
expected to produce more attitude change than no delay in auditory~
feedback (NO Delay). No predictions were made concerning the influence
of the Order of Measurement and Sex variables, which were included for
the purposes of control.

The differences between Improvise and Read conditions and between
.42-Delay and No-Delay conditions in terms of attitude change were in
the predicted direction bat did not attain conventional levels of sig-1
nificance. The Delabeole interaction was also nonsignificant. Upon

examination of the mean change scores for the various.RoleeDelay

37
conditions (Table 8), it is clear that the hypotheses with respect to

attitude change were not supported.

Table 8

Mean Attitude Change as_a Function Of Delayed
Auditory Feedback Interval and Role

 

 

 

Delayed Auditory Role - .
Feedback Interval Read Improvise-
.42-Second-
Delay -1.80 -1.80
No Delay .45- - .40

 

Note.--For the main effect Of Delay, F - 2.35; df - 1,64; p - .13. For
the Role main effect, F - .13, df - 1,64. For the Delay x Role inter-‘
action, Fr! .13, df - 1,64. The sample size in each cell of the table
above is 20.

The 2x2x2x2 analysis Of variance of the attitude change scores
reveals that the only significant effects were: a main effect for the
Order of Measurement of attitudes and counterargumentation (F - 7.57,
df - 1,64, p - .008) and an interaction of Sex and DAF-(F - 5.244,
df - 1,64, p - .025). Those subjects who expressed their attitudes.
before given an Opportunity to express counterarguments (Al) demon-
strated more attitude change than those subjects who expressed their
attitudes after being given an opportunity to express counterarguments
(A2) (Table 9).

(In terms of the sex—delay interaction (Table 10), the effect of
delayad~suditory feedback on attitude change was as predicted for male
subjects (simple effects F - 7.310, df - 1,64, p < .01). For women-
there was a tendency for the No Delay condition to lead to more attitude

change than the .42 Delay condition, although the analysis of variance

2"!

.r'

38
Table 9

Mean Attitude Change as a Function of
the Order of Measurement

 

Attitudes Measured

Before Counterargumentation After Counterargumentation

 

-2o52 .72

 

Note.--The more negative the number, the more change in the direction
advocated in the_persuasive materials. The sample size in each cell
is 40..

Table 10

Mean Attitude Change as a Function of Delayed
Auditory Feedback Interval and Sex of Subject

 

Delsyed Auditory Feedback Interval

 

 

Sex of Subject No_Delay .42 Second Delay
Female -l.60 - .70-
Male 1.65 -2.92

 

Note.--The more.negative the number, the more change in the direction
advocated in the persuasive materials. The sample size in each cell
is 20.

for simple effects indicated that this difference was not significant

(F - .286, df - 1,64).

The Correspondence Between.Counterargumentation and Attitude
Chang . The hypotheses of this study predicted that certain manipula—

tions would determine the amount of counterargumentation that is possible

39
while processing counterattitudinal material. The resulting dif-
ferences in counterargumentation were in turn predicted to result in
different amounts of attitude change. The more a person is able to
counterargue, the less he should be influenced by counterattitudinal
material.

Although counterargumentation and attitude change were not
influenced by the manipulations as expected, there were certain pat-
terns in the data which support the hypothesis concerning the mediating
role of counterargumentation in role-playing induced attitude change.
First of all, the relationship between order of measurement, counter-
argumentation, and attitude change was consistent with the predictions
of this study. Those subjects who expressed attitudes before counter-
argumentation (Al) demonstrated:: (a) less total defensiveness, (b)
more supportive argumentation, and (c) more attitude change than those
subjects who received the reverse order of measurement.

Secondly, the effect of the Sex-Delay interactions on counter-
argumentation and total defensiveness is paralleled by a Seerelay
interaction in terms of attitude change. Men demonstrated more attitude
change and less counterargumentation and total defensiveness when per-
forming under the .42 second DAF interval than when there was no delay
in auditory feedback. Women counterargued more, demonstrated more
total defensiveness, and changed their.attitudes less under the .42-
second delay interval than with no delay in auditory feedback. Finally,
the hypothesized relationship is.supported by certain within treatment
analyses.‘ The overall within-cell correlation between change and
counterargumentation was .299, the correlation of attitude change with

total defensiveness was .306, and the correlation of attitude change

40
with supportive.argumentation was -.373. All the correlations are

significant at p < .05, df - 77.
Alternative Explanations of the Roleeplaying:Effect--Secondapy Analyses

Effort. The effort justification explanatiOn of the rolevplaying
effect asserts that the more effort a person expands in counteratti-
tudinal behavior, the more the person should come to experience cognitive
dissonance, and subsequently, the greater the acceptance of the counter-,
attitudinal behavior or the position represented by it (Zimbardo,

1965). In this study, "improvise" subjects expressed more difficulty
with their counterattitudinal speech than did "read" subjects; those
performing under a .42 second delay in auditory feedback expressed more
difficulty with their task than did those with no delay in feedback.
Contrary to the results of the Zimbardo (1965) study, neither Of these
manipulations had an overall effect on attitude change.

Attitudes were significantly affected by the Order of Measurement
and by the Sex-Delay interaction. An analysis of variance Of self-
ratings in terms of the difficulty of the performance reveals that
neither the Order of Measurement (F < 1) nor the Sex-Delay interaction
(F < 1) had-a significant impact. An analysis in terms of the more
direct question concerning effort yielded similarly nonsignificant
results. The overall within-cell correlation between rated difficulty
of performance and attitude change was a nonsignificant -.O33; for rated
effort and attitudechange the obtained correlation Of -.l34 was also
nonsignificant. In sum, effort did not appear to be related to the

amount of attitude change.

41

Recall. Janis and King (1954) were among the first to speculate
that the superiority of an active form of exposure to counterattitudinal
material over other techniques may be related to the effects of active
participation on attention. That is, active participants may pay
closer attention to the persuasive appeal, remember more, and thus be
more influenced.u On the other hand, McGuire (1969) has indicated that
sometimes active participation may be less effective than other tech-
niques because recipients are prevented from receiving the message.
Results of a number of studies (e.g. King and Janis, 1956; Zimbardo,
1965) seem to suggest that recall of the counterattitudinal message.
can explain little if any of the attitude change variance in role-
playing studies.

All subjects were provided with material listing eight reasons
against the policy of coed living on the same floor of the residence
hall where they lived. The last question in the postexperimental
materials asked subjects to recall as many of-the arguments as they
could. A 2x2x2x2 analysis Of variance of the recall scores indicated
that (a) females recalled more arguments than males (F = 4.83, df -
1,64, p - .032), (b) "improvise" subjects recalled more arguments than
"read" subjects (F - 4.52, df - 1,64, p - .037), (c) a significant
interaction of Delay and Order of Measurement (F - 6.54, df - 1,64,

p - .013), and (d) a significant interaction of Role and Order of
Measurement (F - 3.92, df - 1,64, p - .052). None of these effects
were matched by corresponding effects in terms of attitude change.

The overall within condition correlation between recall and attitude
change was a nonsignificant .085.' Recall was not substantially related

to the amount of attitude change.

42

Satisfaction. The satisfaction hypothesis which grew out of the

 

early Janis and King (1954) study suggests that role-playing leads to
attitude change to the extent to which the role player feels he has
done a good job. In general, the higher the satisfaction with per-
formance, the more attitude change.expected.

\‘In this study, two questions were used to index satisfaction.with
performance. The first (QSl) asked: "How satisfied were you with the
way you fulfilled the task requirements?" The second (Q82) asked:

"How would you rate your performance relative to other subjects who
have taken part in this study?" Subjects responded by checking a
space on a ten-point continuum.

Analyses of variance Of the responses to the two items indicates
that (a) males were more satisfied with their performance than females
(for Q32, F - 8.09, df - 1,64, p.- .006), (b) "no delay" subjects
were more satisfied than ".42 second delay" subjects (for QSl F - 5.05,
df - 1,64, p - .028; for QSZ F - 4.40, df - 1,64, p - .040), and (c)
"read" subjects were more satisfied than "improvise" subjects (for QSl,
F - 5.97, df - 1,64, p - .017; for QSZ. F - 3.31, df = 1,64, p - .074).
None of these main effects were paralleled by main effects for attitude
change. The overall within cell correlation between attitude change
and satisfaction was -.023 when using Q81 and .076 when using Q82.
Neither of the correlations were significant and it can be seen that
satisfaction with performance was not related in any substantial manner

to the amount of attitude‘change produced by the manipulations.

Attitude Change vis-a—via Source and Communication

 

Evaluation. Ratings of the-knowledgeability and trustworthiness of the

anonymously presented list of arguments were submitted to an analysis

43
of variance, but none of the main-effects or interactions were sig-
nificant. The overall within cell correlation between attitude change
and rated trustworthiness of the source was -.126; the correletion
between change and rated knowledgeability was —.l39. The more knowledge-
able and trustworthy-the source was perceived, the more attitude change
in the direction advocated. However, these trends were not reliably
established. Improvise subjects saw the arguments as less biased than
did the read subjects (F - 16.47, df - 1,64, p - .0005). Furthermore,
there was an interaction between Sex and Order of Measurement, such
that women whose attitudes were measured after counterargumentation
order saw the arguments as more biased than women who received the
"attitudes first" order. For men, Order of Measurement had little
effect on ratings of bias (F - 6.43, df - 1,64, p - .014). The overall
within cell correlation between ratings Of bias and attitude change
was a nonsignificant -.145. MOre change was associated with the lower.
ratings of bias although the trend was not significant. An analysis
of variance of ratings of how logical the arguments were indicated no
significant effects.’ The overall correlation between attitude change
and ratings of logic was a nonsignificant -.O60. In sum, ratings of-
sourceds knowledge or trustworthiness or message logic or bias were.

not related to the Observed attitude change.

Correlates of Counterapgumentation and Total Defensiveness

In view of the fact that counterargumentation and total defensive-
ness were affectedfby so many treatment conditions and in such anticipated
ways, it is of special interest to examine the correlates of these two
variables. Table 11 presents the overall within condition correlations

between the two defensiveness variables and several other variables.)

44
While no speculatiOn about the time order of events is necessary with-
respect to such variables as sex, initial attitude, or initial confi-
dence, this is not the case with other of the variables in the table.
Thus, with respect to several variables it is not possible to know
with the data at hand whether they determined or were determined-by
counterargumentation or total defensiveness. Nevertheless, numerous
alluring hypotheses about the determinants of the level of counter-
argumentation can be generated from the data.

The high correlations between pre-treatment attitude and counter-
argumentation (.387) and total defensiveness (.430) are in line with
Brock's (1966) finding that communication discrepancy-is a determinant
of the amount of counterargumentation. In this study, the more dis-
crepant.the person's initial feelings from those advocated in the
persuasive materials, the more the person counterargued. 0f the
variables measured, pretest attitude had the strongest correlation
with counterargumentation and with total defensiveness.-

Without followup on any of the leads at this time,,if can be noted
that: (a) the more a person rated his performance as sincere, the
less counterargumentation and total defensiveness the person generated,
(b) the more the source of the material was perceived to be knowledgeable
and trustworthy and the more the material was perceived to be logical,
the less counterargumentation and total defensiveness the person
generated, (c) the more satisfied the subject was with his performance,
the more counterargumentation and total defensiveness be generated.
There was a slight tendency for women (relative to men) and those with
a lower level of initial confidence in their feelings (relative to those
with a higher level), to counterargue less. There was also a slight

tendency for those who remembered more of the arguments to counterargue

45

Table 11

Coefficients8 of Selected Correlates of Counterargumentation
and Total Defensiveness

 

 

 

 

 

 

 

 

 

,. . . Total
Correlate Counterargumentation Defensiveness
Pre-attitude .387**** .430****
Pre-confidence in Feelings .122 .147
Sex of Subject -.161 -.213*
Recall .174 .114
Ratings of Arguments

how logical -.161 -.252**

how biased -.029 -.081
Ratings of Source

trustworthiness -.244** -.336****

knowledgeability -.263** -.339****
Satisfaction with Performance

task fulfillment .288*** .335****

performance relative to others .194* .236**
Self-Ratings of Performance

convincing .069 .067

sincere -.198 -.293***

 

Note.--A11 variables have been transformed so that high values indicate
more of a value, with the exception of sex where females were arbitrarily

assigned the value 2 and males were assigned the value 1.

In terms of

the pre-attitude measure, the larger the score, the more discrepant the

persuasive materials from the person '8 initial attitude.

aoverall within cell correlation.‘

*
p < .10, df - 78

**

p < .05, df = 78

***

p < .02, df - 78

****

p < .01, df I 78

46
more. Interestingly, no direct relationship was apparent between
ratings of the bias of the arguments and either,~counterargumentation;
(r--.029) or total defensiveness (r--.081).

In one further attempt to clarify the pattern of effects for
counterargumentation and total defensiveness, this section contains a
comparison of the-mean amount of counterargumentation and tetal
defensiveness in various treatment conditions with corresponding
means of other variables potentialLy related to the-tendency to
counterargue. Comparisons are presented for several of the treatment
interactions which were significant for either or both of the defensive-
ness variables. The post hoc nature of these comparisons and their
inevitable limitations must of course be recognized.

One aspect of the unexpected Role—Delay interaction whiCh was
particularly disconcerting for the hypotheses of the-present study
was the.finding that "read-no delay" subjects, who should have counter-
argued more than any other group, actually counterargued less. Table
12 presents the Role-Delay interaction means for counterargumentation,
pre—treatment attitudes, ratings of the logic of the persuasive
material, and ratings of the knowledge of the source of these materials.
Notice that the read-no delay subjects had initial attitudes more con-
sistent with the persuasive materials than any other group. This group
also tended to find the arguments more logical and to give the source
of the materials as high a knowledgeability rating as any other group.
While_pretest attitudes alone cannot explain the pattern of counter-
argumentation, it is possible that pretest attitudes interact with
treatment differences to determine the degree of counterargumentation.
The message may have been of such small discrepancy from the-read-no
delay subjects' own feelings that there was little reason to argue

against the message.

47

Table 12

Mean COunterargumentation, Total Defensiveness, Pre—treatment Attitude,
Rating of the Logic of Persuasive Materials, and Rating of Source's
Knowledge in Relation to Delay-Role Condition

 

Treatment Condition

 

 

 

 

 

 

 

 

 

Probability..
.42 Second .42 Second ‘ Level
No Delay No Delay Delay ‘ Delay Role x Delay
Read ' Improvise ‘ Read ' Improvise Interaction
Counterargu-
mentationa 1.80 3.15 2.80 2.25 p - .018
Total
Defensivenessa 2.35 3.50 3.15 2.65 p - .043
Pre—treatmentb
Attitude 23.25 24.35 27.15 24.20 p - .115
Ratinggofc
Persuasive
Materials,
re: Logical 7.55 6.30 5.85 6.15 p - .180
Ratings of
Sourcec of
Materials,
re: Knowledgeable 7.20 6.50 6.40 7.20 p - .134

 

 

8Based on 3 minute time limit, repeated arguments excluded. The
higher the score, the more counterargumentation or total defensiveness.

bThe higher the score the more discrepancy between own position

and advocated position.

cThe higher the score, the more of the trait in question.

48

Thus far a case has been made that some of the differences in
amount of counterargumentation are due to attitude differences prior
to the treatments. In this vein, the fact that women were initially
more in agreement with the position advocated by the persuasive
materials (F - 4.96, df - 1,64, p - .029) may help account for the
interaction of sex with the other independent variables in terms of.
counterargumentation. Table 13 examines the sex-delay interaction and
the counterargumentation, pre-attitude level, and ratings of the bias
of the arguments used to structure the role performance. While it was
predicted that no delay in auditory feedback should result in more
counterargumentation than a .42 second delay interval, the opposite case
resulted for women. Part of the explanation of the particularly low
degree of defensiveness produced by "no-delay" females may have been
due to the fact that they were closer to the persuasive materials in
initial attitude position than any,of the other groups. It can-also
be noted that ".42 second delay" females had an initial.leve1 of
attitude more discrepant from that advocated than did the "no delay"
females.

No further comment on the results of Study 1 will be made at this
time. In view of the high degree of similarity between the first and
second studies, a general discussion of the results is deferred until

after the results of Study 2 have been presented.

49
Table 13
Mean Counterargumentation, Pre—test Attitude, and Ratings

of the Bias of the Arguments, as a Function of Sexr
and Delay in Auditory Feedback Interval

 

 

Treatment Condition Counterargumentation 'Pre-Attitude Bias
Male .42 Delay 2.25 26.45 4.6
Male No Delay 3.00 25.85 4.2
Female .42 Delay 2.80 24.90 3.1
Female No Delay 1.95 21.75 4.4

(Probability level of
Seerelay interaction) (p - .045) (p - .318) (p - .089)

 

CHAPTER III

EXPERIMENT TWO

A fundamental assumption detailed in the introductory comments is
that role-playing or counterattitudinal advocacy is a superior persuasive
technique because, relative to other techniques, it drastically reduces ‘
the Opportunities for the intended recipients to counterargue. General-
izing the logic of this assumption, it could be conjectured that other
ways of exposing peeple-to counterattitudinal material which affect
the extent of counterarguing should also produce differences in attitude
change such that those who counterargue least change their attitudes
most. One very straightforward method of varying the degree to which
a person will develOp thoughts contrary to those he has been told to
prepare for public presentation would be to vary whether or not the‘
person is specifically instructed to think about such contrary thoughts.
That is,.instructions to counterargue (the presence or absence of such
instructions) should have a strong and clear impact on this putative
defensiveness variable.

From this perspective, it can be seen that-the second study of
this project was designed to provide less equivocal answers about the
role of counterargumentation in the role-playing--attitude-change
relationship. -The second study includes two rather strong orthogonal
manipulations of the likelihood of counterargumentation--namely, reading
versus improvising and presence versus absence of instruction to counter-
argue. The inclusion of the instructional manipulation can provide a

50

51

more solid basis for conclusions since, unlike Read-Improvise or

Delay-No Delay Auditory Feedback manipulations, this variation in:

counterargumentation is not as likely to be contaminated simultaneously.

by differences in effort, satisfaction, perceived pressure to change,
or recall.

As was the case in the first study, it is predicted that impro-
vising a counterattitudinal position will lead to less counterargumen-
tation than reading the same counterattitudinal material. It is further
predicted that, as a result of the differences in counterargumentation,
improvisation will produce more attitude change than a more passive
form of exposure to the counterattitudinal position. The predictions
with respect to the instructional manipulation are parallel. Specific
instructions to think about arguments contrary to the attitudinal posi-
tion to be portrayed should produce more counterargumentation than a
condition without such instructions.‘ Consequently, specific instruc-
tions to counterargue should result in less attitude change than a
condition without such instructions.

The general hypotheses presented above are detailed more formally
as follows:

HI:. Improvising a letter from a prepared list of arguments which
support an attitudinal position discrepant from one's own leads
to less counterargumentation than occurs from.being exposed to
the counterattitudinal position by reading it.

HII: Improvising a letter from a prepared list of arguments which,
support an attitudinal position discrepant from.one's own leads
to more attitude change than occurs from being exposed to the
counterattitudinal position by reading it.

HIII: A person who has been induced to process counterattitudinal
arguments thinks of more arguments against this initially
counterattitudinal position when specifically asked to keep

in mind arguments contrary to the counterattitudinal material
than when no such request is made.

52
HIV: A person who has been induced to process counterattitudinal
' arguments accepts the counterattitudinal position less when

specifically asked to keep in mind arguments contrary to the
counterattitudinal material than when no such request is made.

Method

Overview

The design of the second study is highly similar to the first one.
The major modification was the substitution of an "instructional"
manipulation of counterargumentation for the Delayed Auditory Feedback
manipulation. Half of the subjects were directly asked to think.about
arguments against the position they would be exposed to and work.with.
The attention of the other subjects was not directed toward counter-
arguments.‘ In an orthogonal manipulation of "role-playing", half of
the-subjects improvised a letter to the editor from a list of 10
arguments. The other half of the-subjects read over the list of 10
arguments and a letter written by another subject in the study.‘ As
in the previous investigation, the influence of Order of.Measurement
(attitudes before counterargumentation or vice versa) and subject's‘
sex'were examined. Twenty subjects were randomly assigned to the
various treatment conditions of the 2x2x2x2 after-only design. In
addition to the major attitude and counterargumentation variables,

measures of various subject perceptions and recall were obtained.

subjects
Three hundred twenty introductory psychology students (160 female

and 160 male) volunteered in exchange for class credits to participate
in a study entitled "Cognitive Elaboration."' To ensure as much homo-
geneity on.the critical attitude domain (a tuition increase) as

possible, the sample was restricted to insstate students. The sign-up

53
sheets used to enlist participation permitted a maximum of 15 subjects

per.experimental‘session.

Procedure

As the subjects were arriving for the session, the experimenter
signed their experimental credit forms. At the scheduled time to
begin, the first set of materials was distributed face down.- In his
prefatory comments, the experimenter informed the subjects that there
were several different kinds of activities involved in the study and
that, as a result, different people would be working on different
things, and at different times.

The subjects were also told that several of the tasks would be
timed and that the experimenter was aware that these time limitations
might make some parts of the study seem a little difficult. The
experimenter,noted:; "I‘m aware of these problems and what I'd like
you to do is make the best out of the time-available." These comments

were included as a result of data-from a pilot study which suggested

that rolevplaying subjects involved in similar conditions to those of
the present design had a tendency to dissociate themselves from their
actions. This seemed to occur in the pilot study because subjects
found the counterattitudinal letter writing task not only somewhat
unpleasant but also extremely difficult to complete in the 17 minutes
allotted. It was felt that if the-experimenter explicitly acknowledged
the difficulty to be encountered by the subjects, the subjects would

be less likely to use this dissociation response to resist the impact
of the counterattitudinal material. (Subjects in this study were given

30 minutes to complete the letter writing task.)

54

In order to reduce the possibility of contamination of the unit
of analysis (the individual subject), the experimenter: (1) asked the
subjects not to talk or make any comments while the experiment was in
progress, (2) arranged the chairs so that they faced away from the
center of the room, and (3) asked the subjects to sit in every other
chair. The experimenter qualified his remarks by noting that it was
very important-that they clearly understand the instructions, but that
if they had any questions, they should raise their hand rather than-
speak out.

Since several subjects were to be present at each session, it
would have been ideal to schedule all treatment conditions simultaneously.
A drawback to such a procedure in this study was the fact that the.
Read-Improvise and the Order of Measurement manipulation required
overt instructions by the experimenter to advise subjects of pertinent
time limits of their respective tasks. (The sex variable was a subject
characteristic. ‘The counterargumentation instructions were written
and thus-manipulable unobtrusively without verbal comment or experie
menter's awareness.) It was felt that a great deal of confusion on
the part of the subjects would be created if it were necessary for the
subjects to work while, at the same time, listening to four different
sets of instructions.

In order to maximize the-number of treatment conditions concomir
tantly scheduled while minimizing subject confusion, it was decided
to hold Order of Measurement constant for any given session. Thus,-
Role (Read-Improvise), Instructions to Counterargue (Present-Absent),.
and Sex (Male-Female) treatment conditions were permitted to vary during
a given session and the Order of Measurement of attitudes and counter-

argumentation were held constant.

55

To further reduce the possibility of subject confusion related to
listening to different sets of instructions, "read" and "improvise"
subjects were given different colored slips of paper (either green or
yellow).- The experimenter explained to the subjects that they only
needed to listen to his instructions when their group color was'
announced. The experimenter prefaced all remarks by reference to
either "those in the green group," "those in the yellow group," or

"both those in the green group and those in the yellow group."

Role (Read-Improvise) manipulation. The major hypothesis of this
research was framed in terms of differences between active and passive
forms of exposure, because it was felt that this comparison most ade-
quately facilitated an isolation of the distinctive aspects of role-
playing with respect to its relationship to attitude change. It was
thus of great concern in the present study to equate, as nearly as
possible, the breadth of counterattitudinal material to which active
and passive participants were exposed. The need to equate exposure,
to information is obvious if unequivocal statements about the critical
differences between roleeplaying and non-role-playing conditions are
to be attempted (see Jones and Gerard, 1967, for example, for a comment
on the importance of the-exposure issue).

The procedures for the role manipulation were develOped specifi-
cally to minimize the differences in information exposure (in terms
of the range of information processed and in terms of the-number of
times the information was processed) between active and passive
participants.

The initial instructions, common to both rble—playing conditions
(Reed and Improvise), began as follows: On the following page you

will be presented with a list of arguments concerning tuition at

56
Michigan State University. Some pe0p1e~in this study are-asked to use
this material to write a_"letter to the editor."' Others are asked to

use the material to help them evaluate a letter written by another

person.

The instructions for non-role—playing (Read condition)-subjects

continued:

What I would like you to do is to use this list of argu-
ments as an aid in judging how well another person did at
writing the letter. The specific criteria fOr your judg—
ments will be presented later. Those who were asked to
write a letter were instructed to be as sincere and con?
vincing as possible, to take the same point of view as the
One represented by the-arguments, and to use all the
arguments listed. They were also told not to worry about
spelling, grammar, or punctuation and were told they could
add any other arguments, examples, or illustrations which
were consistent with the listed arguments.

vYou will be given two minutes to read the list of arguments

used as a basis for the letter writing. When instrutted,

please turn the page and begin reading the list. When the

two minute time limit is up, you will be given further

instructions about the letterkevaluation task.

0n the second page the list of 10 arguments was presented. On the
third page, the non-role-playing sUbjects were further,instructed:

On the following page you will find a "letter to the

editor" written by another person who has taken part in

this study. You will be given 10 minutes to read it

over; it is suggested that you read it over at least

twice.

At the end of the 10 minute period, you will be provided

with a set of questions. Please do not consult the

letter you will read to answer these-questions.

Non-role-playing subjects then read the letter improvised by an
active role-player in a corresponding treatment condition. That is,
for example, if a non-role-playing female subject had received instruc-

tions to counterargue, then she was assigned the letter of a female

role-player who.also had received specific instructions to counterargue.'

57
Thus, read subjects were matched to improvise subjects in terms of the
other independent variables.

After the 10 minute period for reading another person's letter,
non-rolevplaying subjects were.provided with a booklet containing the
various dependent variable measures. So that "read" subjects would be
involved with experimental tasks the same length of time as the "improvise"
subjects,-read subjects were given 18 minutes to answer another set of.
questions.‘ This set of materials, which was constructed merely as‘a
filler material, asked "read" subjects to write short essays about how
creatively and sincerely the letter had been written and how they would
have organized the letter differently.

After the instructions common to both read and improvise conditions,
"improvise" subjects read the following:

What I would like you to do is to use this list of»argu-
ments to write a "letter to the editor" of the Lansing
State Journal. As sincerely and.convincingly as possible,
present the same point of view as the one taken in the
material you will read. Use all the arguments listed and
feel free to add any other arguments, examples, or illustra-
tions you want to as long as they are consistent with the-
ones that-are presented to you.

Before you begin writing the letter, you will be given

two minutes to read the list of arguments. When this time
limit-is up, you will be given further instructions about
the letter writing task.

Following the page with the list of role-structuring arguments,
instructions for "improvise" subjects continued:

Following are four blank sheets of paper.‘ The first one
or two may be used to outline your letter, if you so
desire; the others can be used to write the letter. Even.
though it may be difficult, try to imagine that you.
actually are a person with beliefs like those expressed
in the-preceding list of arguments. In the 30 minutes
which you will be allotted try-to write the most.per-
suasive-letter you can in favor ofVa tuition increase at
Michigan State University.

58
Since your work will be used in other parts of this study,
please write as clearly, neatly, and legibly as possible.
However, you need not be overly concerned about spelling,
grammar, or punctuation.
When you have completed the letter, please wait until
requested to continue. Remember, use all the arguments
provided and any others which are consistent with these.
You may consult the list of arguments while writing your
letter. Please sign.your name on the back of the last
page of your letter.
Both the "read" and the "improvise" subjects were started on
. their respective tasks at the same time. The experimenter stated the

time limitations for the different conditions aloud.

Instructions to counterargue and other manipulations.- The manipu-
lation of presence or absence of instructions to counterargue was quite
simply.effected by inserting, for half of the subjects, a brief para-
graph in the instructions which followed the list of role-structuring
arguments. The instructions for "read" subjects stated:

While reading over and evaluating this letter in favor

of a tuition increase, you should also keep in mind’argu-

ments against this position. You will be asked about

them later.-

The instructions for "improvise" subjects stated:-

While working on this letter in favor of a tuition increase,

you should also keep in mind arguments against this position--

even though you won't use them in the letter. You will be

asked about them later.

Other independent variables. Order of measurement.variable-was

manipulated as in the first study. An equal number of male and female

subjects were scheduled in each of the eight treatment conditions.

Performance-structuring counterattitudinal materials. All subjects
were presented with a list of 10 arguments supporting the need for a:

tuition increase at Michigan State University. To suggest that the

59
arguments had been developed by someone other than the experimenter,
they were headed by the title "from 'Tuition_Increase: A Necessary
Evil.'" In fact, nine of the ten arguments were culled from the argu-
ments most frequently given by introductory psychology students during
a pilot study designed to find out what arguments students could spon-_
taneously generate. The list of arguments suggested that a tuition
increase was-needed in order to:. (1) compensate for the increased cost-
in~running the university, (2) hire more and better instructors, (3)
provide smaller class size, (4) provide more and better equipment for
student use, (5) improve inadequate student parking facilities, (6)
offer more scholarships to those especially disadvantaged, (7) establish
more special services for students, (8) provide more time for student-
professor coordinating experiences, (9) provide better learning :.
atmosphere by discouraging those just along for the ride, and (10)
create a condition where students are paying a fair share. Examples.

were provided for arguments l, 4, and 7.
Dependent Measures

Attitude.measurement. After the subjects had either read or
improvised a letter, they were given a 5-page bOOklet-containing the
various dependent measures.‘ Contingent upon which order of measurement
condition the subject was in, either the first two or the second and
third pages asked a set of questions about the critical attitude issue, v
a tuition increase. The instructions for these 12 attitudinal items
were as follows:a

To help us-understand your reactions to the task require-

ments, we would like to have you answer several questions

related to the tapic you were working with. Circle the~

number under each statement which most directly corresponds
to your feelings. Work quickly-but carefully.

60

Under each item was an 11-point scale labeled: "Strongly Disagree"
at the extreme left, "Neutral" in the middle, and "Strongly Agree“ at
the extreme right. As examples of the type of items included, the
first and tenth items were, respectively: "A tuition increase at
Michigan State University would cause more problems than it would solve,"
and "It would be desirable for M.S.U. to raise tuition." Five of the
items were worded so that acceptance of the tuition increase was indi-
cated by disagreement with the-stem; five of the items were worded so
that acceptance of tuition increase was indicated by agreement with
the~stem.

Two items, numbers two and six, did not ask about a tuition
increase ("Out-of-state tuitiOn should be the same as in-state tuition."
and "Most students don't worry very much about the amount of tuition
lthey pay.”). These items were included to obscure somewhat the interest
in attitude change. It was hOped that the subject would feel that the
items were just about tuition in general and necessary to understand
his reactions to the task requirements. (See Appendix G for the complete

set of items.)

Counterarggmentation measurement. The question used to elicit

 

counterargumentation was virtually identical to the one employed in the
first study. Modifications were made apprOpriate to the new issue and
the new task requirements. It was also necessary to expand the category
system.‘ In the first study, both read and improvise subjects were (1)
presented with counterattitudinal materials, and (2) asked to make a
public presentation based on this material.

This was basically what "improvise" subjects were asked to do in

this study. "Read" subjects in the present study, however, had an

61
additional set of stimuli to respond to--the letter to the editor of
another subject. Thus it seemed likely that some "read" subjects would
focus on the structure of the performance rather than its content.
The result would be‘a statement of some ambiguity. How should a state—
ment such as "The essay did not discuss all the points originally pre-
sented." be classified? Does it indicate acceptance of the idea of
tuition increase, rejection of the idea, or ambivalence?

As a result of the-foregoing considerations, a category for per-
formance related comments was added. Statements were.classified into
this category if they referred to some aspect of the performance~of‘
the improvising subject without indicating pro or con feelings about
the tuition issue.

As in Study 1, inter-rater agreement was employed to assess the
reliability of the thought listing instrument. Two judges, working
independently and without knowledge of the subject's treatment condi-
tion, initially agreed on the categorization of 86.8% of the-responses
to the thought listing question. Those responses that the judges did
not initially agree upon were discussed and then assigned to the most
appropriate category. (For each of the four scoring systems, Appendix

F lists the frequency of occurrence of the major response categories.)

Other dependent variable measures. The last-two pages in the
dependent variables booklet contained questions to provide some informa-
tion about plausible alternative explanations of the data. All of these
items used lO-point scales. A number of the questions were related to
other explanations of the role-playing effect. There were five questions
asking for an evaluation of the constructed "letter to the editor."

The dimensions for evaluation included: sincerity, convincingness,

coherence, logic, and bias.‘ Two questions asked about the amount of

62
effort put into writing the letter. For the first seven items,
"improvise" subjects evaluated themselves; "read" subjects indicated
their perceptions or estimations concerning the "improvise" subject's
performance.

Two questions focused on an evaluation of the source of the role-
structuring materials (how knowledgeable and how trustworthy). There
were also two items designed to tap the subject's satisfaction with
his task performance. It was assumed that "read" subjects would rate
their own rather than the "improvise" subject's fulfillment of the task
requirements. Also, there was one question asking subjects to indicate.
the extent to which they were able to become involved with the experi-
mental task and a question asking about how much they felt pressured
to change their opinions. The final question in the booklet was an
open-ended recall item asking the subject to list the main thesis and
supporting arguments of the list of 10 arguments provided to all sub-
jects at the beginning of the study. (Appendix G-contains the materials

included in the postexperimental:questionnaire.)
Results
Tests of Hypotheses

Counterstgumentation and other thought_processes. In presenting
the results with respect to counterargumentation and other subject
reactions to the counterattitudinal material advocating a tuition
increase, the problem again arises as to which system is most appro-
priate. Again, more emphasis is given to an effect the more different,
‘measurement systems detect it and the more reliable the effect appears.
Paralleling the first study, a summary.table listing the probability

levels for the analyses of variance will be presented with separate

63

listings for each method of scoring. The basic distinctions between
scoring systems are whether the system used: (a) the three minute
limit with repeated arguments included in the total (3-mdn. RAI),.
(b) the 45-second limit with repeated arguments included in the-total
(45-sec. RAI), (c) the three minute limit with repeated arguments
excluded (3-min. RAE), and (d) the 45ksecond limit with repeated argu-td
ments excluded (45—sec. RAE). For the presentation of specific results:
in this section, the scoring system upon which the measurement was
based will be noted in the table where the data'appears.

Returning now to the hypotheses, counterargumentation was pre-
dicted to be greater (a) for those given specific instructions to
think about arguments on the other side of the issue that they were
being exposed to, relative to those for whom no mention of the other
side of the issue was made and (b) for those exposed to the counter-
attitudinal position by reading about it relative to those who improvised
a letter. The Order of Measurement of attitudes and counterargumentation.
and Sex of subject were included for purposes of control and were not
expected to affect counterargumentation. The probability values for~
the 2x2x2x2 analysis of variance of counterargumentation scores are
presented in Table 14. Also included in the table are the results.of
the analyses for total defensiveness, supportive argumentation, and'
total number of arguments. Results for the four different scoring
systems are also presented.

To summarize the-complicated appearing set of results presented
in Table 14, it can be said that each of the four independent variables
appears to have an overall effect on one or more of the variables
derived from.the content analysis of subjects' thoughts. The importance

of the interpretations of these effects is overshadowed, however, by

64

m.m Han now mHo>oH huqaanonoun 0:9

.Amem .oom nqv movoaono ouooaowum nauseous sows owned osfiu vacuum o>amuhuuom
.Amam .aaa_mV guesses» uueusswum couscous nuns oases uaau.uusaea mousse
.AHem .ooo mqv movoaunw ousoasmuo venomous sows ufiaaa mafia vacuum o>am:huuom

U

a

.AHdm .eHa mv wounded“ mundanwum communes suds uflaaa.oaau ounces moneys
.eaumua mum on. w a,susa

.vouowa uos .ouomouonu .ouoa mom aoaumuovfimdoo some: nonsense uaovoonom

emu do ouuomwo on we: moowuoeuouow n x o N m x 4 use .9 n u x.m .n x o N d on.n o .o n m .m x d osytl.ouoz

 

'fl‘

 

«Ammo.v ‘maeso.v n x m x <
oxmmo.v o x m a a
A sakoo.vuimqo.v efiouo.vufieoo.v
nieoo.v nfioao.vmxmmo.v sieuo.vuaemo.v n a m
Ammo.v
aimmo.vmfiumo.v a x a
. oiaoanv . .
gimme aminoa v cameo vmxooo V o a a
aAomo.v eAmmo.vnAeeo.v a Axomv
saneo.voaemo.v
oAmwo.VmAeso.v niooo.vmimmo.v o AmaoeuoauumaHv
efikmo.vuflaoa.v
nxmao.vaHeo.v nammo.vaAmmo.v eAmNo.V
cameo.vnfim~o.v m Auoeuov
nAeso.v eAmNO.V
eihoo.vmamao.v nAeao.vammo.v «Aeao.v , a Auaoav
oueoaowu< snowmen Hence ounceswumuouasoo mucosawudv .eoauofiue>
Hmuoa wawuuoannm mo oouoom

 

unoaouu< mo omww

 

. avenue: wowuoom unom.onu mo poem new museum A oumuonom nuazllnuooaswu< mo Honsoz House use
.mnooo>«osumon Houoa .nuoosnwuououoooo .ouauaawu< wnfluuonnsm you mouoom mo nonmauo> we ownhams< mo mumaaom

«H manna

65
the two significant interactions present in the data--the Role-
Instructions to Counterargue interaction and the Sex-Order of Measure-~
ment interaction.

Both the Role and Instructions to Counterargue manipulations
failed to have the predicted effects. While Role did have an overall
effect on counterargumentation (F - 4.80, df - 1,304, p - .028, 45
sec. RAI system; F - 6.08, df - 1,304, p - .014, 45 sec RAE system),
"improvise" rather than "read" subjects listed more counterarguments.
The Role manipulation had no substantial effect on total defensiveness.

In terms of the Instructions to Counterargue Manipulation, the
results are somewhat surprising. Basically, those peOple who were
instructed to think of arguments on the other side of the issue that
they were being exposed to produced significantly fewer supportive
arguments than did subjects not given such instructions (F - 4.43, df -
1,304, p - .034, 3 min. RAE; F - 4.51, df - 1,304, p - .033, 3 min.
RAI). Thus, instead of the-instructions producing more counterargu-
ments or total defensiveness as predicted, they led to fewer supportive
arguments. According to the hypotheses, the "improvise" subjects not
given instructions to counterargue should have demonstrated the least
amount of counterarguing. Aszable 15 reveals, the Improvise-
Instructions Absent condition resulted in the greatest amount of'
counterargumentation.

The significant interaction between.Role and Instructions to
Counterargue is the result of the low.amount of counterargumentation
listed by "read--instructions-absent" subjects. This treatment condi-
tion also appears to be.responsible for the main effect of Role. In
fact, "improvise" subjects only listed more counterarguments than-

"read" subjects when there were no instructions to counterargue.‘ The

66
Table 15

Mean Counterargumentation as a Function of Role
and Instructions to Counterargue

 

Instructions to Counterargue

 

 

Role Absent~ Present
Improvise 2.36 2.15
Read 1.78 2.21~

 

Note.--Scores were based on the three minute system with repeated :
arguments excluded (3 min. RAE). N - 80 in each cell.

analysis of variance for simple effects shows that the differences
between improvise and read conditions was significant when subjects
were not instructed to counterargue (F - 6.64, df - 1,304, p < .05) but.
was not significant when there were instructions to counterargue (F -
.075, df - 1,304).

There was no a priori reason to expect that Order of Measurement
and Sex of Subject would influence the degree of counterargumentation,
but it can be seen in Table 14 that both of them did-have an effect:
on both the amount of counterargumentation and total defensiveness.
Table 16 illustrates this interaction in terms of total defensiveness.
An analysis of variance for simple effects demonstrates that the inter?
action is a result of the effect of order of measurement on male subjects.
Male subjects listed significantly fewer defensive thoughts when they
indicated their attitudes after being given an opportunity to list.
counterarguments than when the order of measurement was reversed (F -
11.42, df - 1,304, p < .01). Order of Mhasurement had no effect on

female subjects (F - .208, df - 1,304). This interaction helps account

67
for the observed main effect for women to counterargue more than men
(F - 7.36, df>- 1,304, p - .007, 45-sec. RAE system) and for the-
"attitudes-before-counterargumentation" order of measurement to lead

to more total defensiveness (F -.4.27, df - 1,304, p - .037).

Table 16

Mean Total Defensiveness as a Function of
Order of Measurement and Sex of Subject

 

Order of Measurement
Attitudes Measured
Sex of Subject Before Counterargumentation After Counterargumentation

 

Male 1.22 .76

 

Note.--S¢ores were based on the 45 second system*with repeated arguments
excluded (45 sec. RAE).. N - 80 in each cell.

Attitude change. Very early in the experimental session, all

 

subjects read a list of arguments supporting a tuition increasee-a
position which most were expected to find counterattitudinal. Those
subjects who created a letter based on these arguments (improvise)
were expected to accept them to a greater degree than those subjects
who read a letter created by another subject (read). In addition,
those subjects who were specifically instructed-to think about arguments
on the other side of the issue they would be working with were expected
to demonstrate-less agreement with the point of view of the list of'
arguments than were those subjects who were not so instructed.

The instrument.constructed to measure attitudes toward a tuition
increase was composed of twelve items related to tuition at Michigan

State University. Ten of the items directly focused on feelings or

68
beliefs about the desirability of raising tuition. Half of these 10
items were worded so that agreement indicated acceptance of the tuition
increase; half were worded so that disagreement indicated acceptance.

For items 1, 5, 7, 11, and 12 (those items for which disagreement
indicated acceptance of the tuition increase) the item-total correla-
tions were, respectively: .628, .713, .716, .324, .749.- For items
3, 4, 8, 9, and 10 (those items for which agreement indicated accept-
ance of the tuition increase) the item-total correlations were,
respectively: .732, .607, .651, .565, and .537. The high and generally
uniform size of the coefficients suggests that the ten-item scale
measured a unidimensional variable.

A message acceptance score was determined for each subject by
summing the scores for the 10 items in the attitude toward tuition
scale. With Order of Measurement and Sex included with Role and
Instructions to Counterargue as independent variables, a 2x2x2x2 analysis
of variance of the acceptance scores indicated that the only significant
effect was.for the Role-Instructions to Counterargue interaction (F -
5.34, df~- 1,304, p - .020). Neither the overall effect of Role
(F - 1.60, df - 1,304, p - .20) nor of Instructions to Counterargue
(F - .42, df-- 1,304) was substantial.

Table 17 indicates that when subjects were given instructions to
counterargue while processing the counterattitudinal material, improvi-
sation led to more change than reading (simple effects F -.6.406,
df - 1,304, p < .05). When no mention was made of arguments on the
other side of the issue from.the one the subjects were working with,
there was no significant difference between "read" and "improvise"
subjects (simple effects F - .544, df.- l,304).’ The direction of the
difference in attitude change between read and improvise was Opposite

from that predicted.

69
Table 17

Mean Acceptance of Counterattitudinal Arguments as a.
Function of Role and Instructions to Counterargue

 

§pecific Instructions to Counterargge

 

 

Role Absent Present
Improvise 51.57 56.82
Read 53.42 50.47

 

Note.-—The higher the score, the greater the acceptance of the advocated
tuition increase. N - 80 in each cell.

The instructions to counterargue manipulation did not lead to dif-
ferences in attitude change for read subjects (simple effects F I 1.383,_
df.- 1,304), although there was a tendency for the instructions-absent
condition to lead to greater acceptance for read subjects. For improvise
subjects the effect of the instruction variable was apposite from.that
predicted (simple effects F - 4.379, df'- 1,304, p < .05). Specific
instructions to counterargue produced more acceptance than no instruc-

tion for improvise subjects.

The correspondence between counteratgumentation andgattitude changg.
The hypotheses of this study predicted that the manipulations would
first affect the degree of counterargumentation and that, in turn, the
degree of counterargumentation would moderate the amount of attitude
change. Instructions to counterargue were expected to produce more
counterargumentation and less attitude change than a condition without
such instructiOns.' Reading counterattitudinal material was expected

to lead to more counterargumentation (thoughts contrary to those being

70
read) and less attitude change than improvising a letter based on the
same counterattitudinal material.

While these independent variables did have main effects on the
types of thoughts generated in response to the counterattitudinal
material, they were not in the direction, or of the type, predicted
and they were not paralleled by main effects in terms of acceptance
of the counterattitudinal position. The "instructions-absent" condition
led to more supportive arguments than did instructions to counterargue,
and to more total number of thoughts listed. This manipulation had no
overall effect on counterargumentation, total defensiveness, or
attitude change, however. Contrary to predictions, "improvise"
subjects had a tendency to list more counterarguments than1”read"
subjects, although this effect is obscured somewhat by the fact that
there was also a tendency for "improvise" subjects to list more sup-
porting arguments and more total arguments. The role manipulation.
had no overall effect on attitude change.

It_must be noted that the results just presented do not constitute
a disconfirmation of the hypothesis that role-playing mediates attitude
change by inhibiting counterargumentation. Since the instruction to
counterargue manipulation of this study failed to have an effect on
counterargumentation, the major hypothesis of this study would not
predict an overall effect of instructions on attitude change. A simi-
larly ambiguous situation exists for the case of the-role manipulation.
There was a tendency for "improvise" subjects to produce more counter-
arguments than Vread" subjects, but "improvise" subjects also tended
to list more supportive arguments and a larger total number of argu-
ments. If the greater counterargumentation of "improvise" subjects

is merely a reflection of their overall higher response rate (a

71

difference balanced out by~a corresponding difference in amount of
supportive argumentation), then no differences in attitude change
would be expected according to the major hypothesis ofxthis study.

Partially supporting the predicted inverse relationship between
counterargumentation on message.acceptance was the fact.that the sig-
nificant interaction of role and instructions on counterargumentation
was paralleled by an interaction of role and instructions on acceptance
of the counterattitudinal message.

Unfortunately, the interactions do not match up as the hypotheses
would predict.“ While improvise subjects showed greater acceptance
of the tuition increase than read subjects when given instructions to
counterargue, there were no significant differences between these two
groups in terms of counterargumentation (compare“Tables 12 and 14).
On the other hand, while there were no significant differences in
acceptance of the counterattitudinal material between improvise and,
read subjects not given instructions to counterargue, these improvise
subjects counterargued toga greater extent than did read subjects
under similar conditions.- If the means for improvise subjects are
assumed to be inflated by-a constant of .50 (due perhaps to their
longer period of attention onzthe counterattitudinal material) and are
adjusted accordingly, the pattern of means for counterargumentation
correspond exactly to those for communication acceptance., Barring
this highly speculative possibility, the data do not provide much
support for the prOposed relationship between counterargumentation
and attitude change.

The most conclusive evidence for the mediating role of counter-
argumentation in the role-playing effect is provided by within treatment

analyses. The within-cell correlation between counterargumentation and

72

acceptance of the counterattitudinal position was -.304 (the counter-
argumentation measure was based on the 3-minute limit with repeated
arguments included). Acceptance of the advocated tuition increase:
was correlated -.346 with total defensiveness and .322 with the
number Of supportive arguments listed. With a sample size Of 320,
each of the correlation coefficients was significant.at p < .01.
Thus it appears that the more counterargumentation and total defensive-
ness the less acceptance of the persuasive material. The more support-
ive argumentation generated, the more acceptance.
Secondary analyses:‘ Covariates Of messgge accgptance, countergtgpmen-
tation, and total defensiveness

The following section takes a brief look at several possible
moderators of communication acceptance and of defensive processes.
In,a juxtaposition Of the standard order of review, it can-be summarily
noted that none of the possibilities considered after an explanation
Of pattern Of effects on the major dependent variables. The communica—
tion acceptance data do not seem to be related to recall, satisfaction
with performance, or to perceived pressure to change one's Opinions.
While there were no treatment effects on ratings Of the persuasive.
material7in terms Of its logic or bias corresponding to the pattern of
effects for acceptance, there was correlational evidence indicating
that those who perceived the message to be less biased, accepted the
persuasive materials more. There were also no treatment effects on
ratings of the source's knowledge or trustworthiness to correspondfi.\-
to the treatment effects on attitudes. However, those who accepted ‘
the message the most also saw the source as more trustworthy. Within—
treatment analyses revealed that-the more knowledgeable and trustworthy

the source was perceived, the less counterargumentation and total

 

73
defensiveness was generated. The more logical and the less biased the
arguments were perceived to be, the fewer counterarguments and total
defensiveness listed. (The cOrrelates Of communication acceptance
and counterargumentation and total defensiveness are listed in Table 18.)

Acceptance of the persuasive material was not determined in any.
direct manner by recall Of the persuasive arguments.' The only sig—
nificant effect arising from an,analysis.of variance Of number of
arguments recalled was a main effect for sex of subject.‘ Women
remembered significantly more arguments than did men (6.43 and 5.55,
respectively). There was.no difference in terms Of message acceptance
between male and female subjects and the overall within-cell correla-
tion between recall and communication acceptance was .00.‘ The correla-
tion coefficients for.se1ected correlates of communication acceptance
are presented in Table 18.

Through an oversight, subjects were only asked to estimate the
difficulty and effort involved in writing the letter. Thus, there is
no data on how difficult or effortful "read" subjects considered their.
own tasks. As a result, it is not possible to check on differences
between read and improvise conditions in terms of the effort expended.
Without the comparative data.a test of the effort justification explana-
tion of the role-playing effect is not possible.

Satisfaction with one's performance Of the experimental tasks
also fails to account for the attitude change Observed. Responses to
the question,,"How satisfied were you with the way you fulfilled the
task requirements?" were significantly affectediby the interaction of
Sex-and Instructions to Counterargue. Male's satisfaction wasfnot
influenced by the instructions to counterargue manipulation (mean

instructions—absent, 6.43; mean instructions-percent, 6.16). Females

74

Table 18

Correlates for Communication Acceptance, Counterargumentation
and Total Defensiveness (Product-moment correlations)

 

 

 

 

Communication Counter- Total
Acceptance Argumentation Defensiveness
Ratings of Communication .
logical .061 -.147** -.174***
biased -.126** .128** .l4l**
Ratings of Source
knowledgeable .068 -;ll6** -.l43**
trustworthy .159*** -.l74** -.212***

Satisfaction with Performance

 

 

 

 

task fulfillment -.043 -.017' -.036

relative to others -.104* -.004 -.013
Pressure to changg opinion .102* -.028 -.017
3335;; .000 .092 .092
Counteratggmentationa -.304*** - .948***
Total Defensivenessa -.346*** .948*** -
Sgpportive Arggmentationa .322*** -.496*** -.542***

 

 

8Based on the 3 minute time limit with repeated arguments included.

*
p < .10, df - 302

**
p < .05, df -.302

***
p < .01, df - 302

75

in the "instructions-absent" condition were significantly less satis-
fied with their.performance than were females given instructions to
counterargue (5.71 and 6.51, respectively).* In terms of the question,
"How would you rate your task performance relative to other subjects
who have taken part in this study?" there was a marginal tendency
(F - 2.78, df - 1,304. p - .09) for those with instructions to
counterargue to be more satisfied. The patterns of means for accept-
ance of the tuition increase did not duplicate the pattern for satis-
faction. The correlation between acceptance of the advocated tuition.
increase and satisfaction was -.043 when the first question above was
used as a measure of satisfaction and.-:104 when the second question
was used. While neither of the correlations are significant the
latter indicates a tendency for those who were more satisfied with
their performance to show less acceptance of the persuasive material.

There were no significant effects for ratings of trustworthiness,
but there were marginal trends for an Order of Measurement—Instruction.
to COunterargue interaction and an Instructions to Counterargue-Sex.
interaction. When there were no specific instructions to counterargue,
the attitudes after counterargumentation order led to higher ratings
of source trustworthiness.‘ With specific instructions to.counterargue,
Order of Measurement had no effect on the ratings of the source's
knowledge (interaction F - 3.37, df-- 1,304, p - .064). There was a
tendency for men to find the source less trustworthy when performing
under instructions to counterargue than when there were no instructions
to counterargue. Instruction to Counterargue did not affect the rating
of trustworthiness for female subjects. While these borderline inter?
actions did not correspond to treatment effects on.attitudes, the

overall withinrcell correlation of-.159 (see Table 18) does suggest

76
that message acceptance was greater the more the source was seen as
trustworthy.

Unexplicably, subjects who indicated their attitudes before being
given an Opportunity to counterargue, perceived the source of the list
of arguments as less knowledgeable (F‘- 6.56, df - 1,304, p - .0110).
There was also an interaction of Role and Instruction to Counterargue
on ratings of the source in terms of his knowledge (F - 8.79, df - 1,304,
p - .003) (see Table 18). Subjects who improvised a letter found the
source more knowledeable than read subjects did under conditions where
there were no instructions to counterargue (simple effects F - 6.004,
df - 1,304, p < .05). However, when there were instructions to counter-
argue, read subjects tended to find the source to be more knowledgeable
than write subjects did (F - 3.029, df - 1,304). It was this Role-
Instructions to Counterargue interaction which was also significant
for attitude change.. It becomes clear, however, that ratings of knowledge-
ability cannot explain the change data. A comparison of Tables 17 and
19 reveals that "instructions-absent--improvise" subjects and
"instructions to counterargue-read" subjects gave the source the highest
rating on knowledgeability but demonstrated the lowest amount of com-
munication acceptance. The correlation between message acceptance and
ratings of the source in terms of knowledgeability was a nonsignificant
.068.

In terms of discussions of reactance and demand characteristics,
it is relevant to draw attention to the_finding that "improvise"
subjects perceived themselves to be under much more pressure to change
their opinions than did "read" subjects (F - 23.86, df - 1,304, p's
.0005). Improvise subjects also had a tendency to counterargue more

than read subjects although this effect seemed to be due to the low

 

\

77
Table 19

Mean Rating of Source's Knowledge as a Function of
Role and Instructions to.Counterargue‘

 

Instructions to Counterargge-

 

 

Role Absent' Present
Improvise 6.45- 5.81'
Read 5.58- 6.42

 

amount of counterarguing produced by read subjects not given instructions
to counterargue. There was also an effect of Order of Measurement and
Sex of Subject on perceptions of pressure to change one's Opinion. The
interaction F of 7.13 (df - 1,304) was significant at p - .008. Male
subjects tended to feel less pressure to change their opinions when-
their attitudes were measured after they were given an opportunity to
counterargue (for simple effects F - 3.983, df - 1,304,.p < .05).
Female subjects tended to feel less pressure when their attitudes were
measured before they were given an Opportunity to counterargue. The
Sestrder-of Measurement interaction in terms of counterargumentation
did not exactly parallel that for perception of pressure to change.
Male subjects reported less pressure to change and produced fewer
counterarguments when attitudes were measured after counterargumenta-
tion. Females did not produce counterarguments differentially as a
function of the order of measurement even though there were differences
between order of measurement conditions in terms of perceived pressure
to change.. Acceptance of-the tuition‘increase was not significantly

influenced by Sex or Order of Measurement. The overall within-cell

 

 

78
correlation between communication acceptance and perceived pressure.
to change.one's opinions was a nonsignificant .102. ’In sum, there is
little reason to believe that perceptions of pressure to change one's
Opinions had a substantial impact on either communication acceptance
or counterargumentation or total defensiveness.

All subjects were asked to indicate how logical and biased they
felt the arguments presented in the letter to be. For the most part,
the arguments presented in the letter were taken from the list of
arguments which was provided for the subjects in the early part of
the experiment.' However, "improvise" subjects were encouraged and did
actually think of new arguments which were consistent with the ones
in the list. As a consequence, "read" and "improvise" subjects were
making their ratings from slightly different perspectives.‘

An analysis of variance of the ratings of how logical the argu-
ments had been revealed a Role-Sex interaction. Table 20 illustrates
that the "female-read" and "female-improvise" subjects did not differ
in their ratings. However, for male subjects, those in the read
condition rated the arguments in the letter as significantly less.
logical than those in the improvise condition did (for the RoleeSex
interaction, F - 5.87, df - 1,304, p - .015). The low ratings by.
male—read subjects accounted for the overall tendency for "improvise"
subjects to see the arguments as more logical (F -.ll,51, df - 1,304,
p - .001) and the tendency for males to rate the arguments as less
logical than females did (F = 2.84, df - 1,304, p - .089). This Role-
Sex interaction did not appear for communication acceptance or for
either of the measures of defensiveness. For subject's ratings of

how biased the arguments had been there were no significant effects.

79
Table 20

Mean Subject-Ratings of Bias of Persuasive Material
as a.Function‘of Role and Sex of Subject

 

. Sex
Role Male ' Female

 

 

Improvise 6.86 6.66,

Read 5.28 6.40

 

CHAPTER IV

GENERAL DISCUSSION.

In an introductory textbook discussion of role-playing Zimbardo
and Ebbesen (1969) recently observed:.

The conclusion derived from this body of research is con-

sidered to be the most reliable in the area of attitude

change:- Active participation is more effective in

changing attitudes than is passive exposure to persuasive

communications. (p. 57)

Echoing this position, Weiss (1971) more recently noted:

It is generally well established that opinion change is

greater when the subjects participate actively in the

persuasion (role playing) than it is when they passively

receive the persuasive communication. (p. 29)

In a subsequent commentary on the role-playing area Insko and SchOpler
(1972) acknowledged the widespread confirmation of the phenomenon but
candidly admitted they were not sure why it occurs.

From the vantage noted above and in the introduction, the failure
to demonstrate a causal link between counterargumentation inhibition
and the role-playing effect would not be particularly surprising.
However, the inability to replicate this "reliable" effect under well
controlled conditions should be reason for amazement if not alarm.

The findings of the present studies have done little to confirm
the robustness of this effect. In attempting to determine the value
of the present research for clarifying the role-playing controversy,

the discussion begins with an exploration.of the failure of the manipu-

lations to have the expected effects on counterargumentation. Quite

80

81
naturally attention is, in turn, focused on the validity of measure of
counterargumentation and other defensiveness variables. This is fol-
lowed by an interpretation of the observed attitude change patterns
and a re-analysis-of the hypotheses of the study. Throughout the dis-

cussion, suggestions for future research are provided.‘

Possible Determinants of Counteratggmentation.

In sorting through the various findings of the two studies, one
of the biggest hurdles in the way of any coherent and straightforward
interpretation is the complexity of the counterargumentation data.
Counterargumentation and the more general variable, total defensiveness,
were significantly influenced by an overwhelming number of treatments
and treatment combinations. Unfortunately, many effects appeared where
none were predicted, and those that were predicted were not obtained.

In part, the confusion that has resulted appears to have derived
from a set of expectations based on too narrow a.conception of-the
determinants of these defensive processes. The place to begin is with
the hypotheses. The major contention of this research was that the~
superior persuasive impact of role-playing results from the fact that
the active participant is less able to think of counterarguments.’ One
who 18 involved in the public presentation of a counterattitudinal
position was presumed to have less Opportunity to consider arguments
on the other side of the issue than someone who is exposed to the
counterattitudinal position with nothing else to do but.evaluate the
material. Thus the critical difference between active and passive
participation was conceiVed of in terms of the ability to counterargue.

One factor-overlooked in the develOpment of the argument presented
above is the subjects' motivation to counterargue. Indeed, there is

good reason to believe that motivational aspects of the situation are

 

 

82
central in determining the amount of defensiveness. From a theoretical
perspective, Brehm's theory of reactance (1966) predicts that the more
a person's freedom to believe as he wishes is threatened, the more
reactance experienced and consequently, the more resistance expressed.
Thus, if an improvisation condition created mOre perceived threat
than a passive exposure condition, Brehm's proposition would predict
more resistance for subjects in the improvisation condition. Brock
(1967) has demonstrated that counterargumentatiOn is a direct function
of the discrepancy between the message and the recipient's own position.
(It can probably be assumed that the more discrepant the message, the
more the subject was motivated to resist by counterarguing.) Brock
also found that intent to persuade had a tendency to increase counter-
argumentation. Osterhouse and Brock (1970) have demonstrated that the
degree of threat in the communication situation can determine the
amount of counterargumentation. It may also be the case that the more
frustrated a person is by the experimental procedures or events,
content issues aside, the more likely he is to counterargue.

Some of the patterns in the data do lend themselves to an inter-
pretation in terms of motivation to counterargue. For instance, in
Study 1 counterargumentation was affected by the interaction of Role
and Delay (see Table 3). "No Delay-Read" subjects who were predicted
to produce the most counterarguments actually produced the least. As
was previously noted, this no-delay-read group had pretreatment attitudes
more consistent with the attitudinal position being advocated than any
other group. Presumably, there was little reason for these subjects to
counterargue since they were initially more inclined to accept the

arguments.

 

 

83

A similar situation was found for the Sex-Delay interaction.
Female subjects in the No Delay condition_who should have counter-
argued more than their .42 Delay counterparts did not. On further
inspection it was found that these No Delay-Female subjects had pre-
treatment attitudes more in line with the persuasive arguments than
the other groups. Here we have two situations in which the ability
or Opportunity to counterargue was high but in which little counter-
argumentation resulted. It is assumed that, through the fortunes of
random sampling. those subjects in the treatment conditions described
had the least motivation to counterargue (since their pretreatment'.
attitudes were more consistent with the position in the persuasive
materials) and therefore produced less counterargumentation.'

A.word of warning is in.order.‘ The findings just discussed in
terms of the Role-Delay and Sex-Delay interactions and those concerning
the high within-cell correlations between pretest attitude and counter-
argumentation suggest the need for pre-post designs--perhaps even
with blocking on initial level of attitude. Initial attitude level
seems to have a sizeable impact on the amount of counterargumentation
produced in response to counterattitudinal material.

There are theoretically embarrassing aspects of the data for which
an explanation in terms of pretreatment attitude is not possible. In
the Sex-Delay interaction just mentioned, females in.the .42 second
delay condition had a tendency to produce more counterarguments than
males in the corresponding condition. Part of this higher level may
reflect the fact'that their motivation to counterargue was sufficient
to overcome the situational restraints. Perhaps the females were more
angered, frustrated, or threatened by the disruption of their speech

than the males were.

 

 

84

There are two other instances where the hypotheses were more
strongly contradicted.’ In examining the Role-Delay interaction (Study
1) it was found that under no-delay conditions, "improvise" subjects
listed significantly more counterarguments than did "read" subjects.
While part of this effect was previously accounted for in terms of the
pretest attitudes of the "No Delay-Read" subjects, reference to Table
3 suggests that there is more to be explained..

Improvisation was also found to lead to more counterargumentation
than reading in Study 2. Under conditions in which there were no
instructions to counterargue "improvise" subjects listed more counter-
arguments than "read" subjects. With instructions to counterargue,.
there were no differences between improvise and read subjects. When
there was no delay in auditory feedback in Study 1, and when there
were no instructions to counterargue in Study 2 (analogs of standard
role-playing manipulations to be found in literature), role-playing
led to more counterargumentation than a more passive technique of
exposure. '

One possible interpretation of these findings is that there are
conditions in which the enhanced motivation of the active participant
can overcome or cancel out the typical effect of the reduced oppor-
tunity to counterargue.‘ It is conceivable that improvise subjects
could have relatively higher motivation to counterargue than read
subjects, but a lOwer Opportunity to counterargue. If the motivational
factor is more powerful than the Opportunity factor, the net effect
could be more counterargumentation for improvise than for read subjects.
Clearly, research on the conjoint impact of these two variables is in
order. The implications of this are that the hypotheses were inadequate

in that they did not specify the limiting conditions for the effect of

 

85
role-playing on counterargumentation. The "all other things equal"
assumption implicit in the hypotheses of this study may be unreasonable
for a large number of settings.

Another possible interpretation_is that the observed tendency
for improvise subjects to list more counterarguments than read subjects
reflects a higher general response rate for improvise subjects. This
higher response rate may derive from the fact that improvise subjects
are performing in a manner that would function as a warm-up period for
the counterargumentation measure since they are improvising in terms
of the counterattitudinal material. It-is also possible that a higher
response rate could develOp from a longer period of attention to the
counterattitudinal material.

For response rate to be used to dismiss a finding of more counter-
argumentation for improvise subjects, it is necessary that "improvise"
subjects also show a tendency to produce more supportive argumenta-
tion as well. In Study 2, "improvise" subjects listed more total
number of arguments and more counterarguments than did read subjects.
While the effect was only detected by one measurement technique, there
was also a tendency for "improvise" subjects to list more supportive
argumentation. Thus, in Study 2 the greater amount of counterargumen-
tation produced by "improvise" subjects (with no instructions to counter—
argue) may be artifactual. However, even_if the relative degree of
counterargumentation is equal for read and improvise subjects, that
still represents a contradiction of the hypotheses of the study. This
suggests the-need for a method of standardizing the amount of output
of each type relative to the total output.

It should be noted that the conditions established in the two

studies provide a conservative test of the effects of role-playing on

86
counterargumentation. In Study 1 the differences in opportunity to
counterargue between read and improvise subjects were minimal. Read
subjects were required to present the essay orally as though they were
giving a speech. rWhile they may have been less involved and did report
less_difficulty with their performance than improvise subjeCts, it can.
still be seen that the read subjects were quite active. In Study 2'
improvise subjects were given 30 minutes to compose a letter to the_
editor; read subjects were given 10 minutes to examine a letter written
by another subject. In Study 1 the passive participants were nearly
as active as the active participants. In Study 2 active participants
spent three times as much time working with the counterattitudinal
material.~ Under these circumstances it should be difficult to find
that improvisation led to less counterargumentation.

These problems have arisen from an inherent difficulty in research
on the role-playing effect. That is, the role-playing effect is
formulated in terms of a comparison between.an active and a passive
form of exposure to counterattitudinal material. Therein lies the
problem since active participation differs from passive participation
simultaneously on several dimensions--including effort, exposure to
content. and time involved.

A partial solution to this problem is the inclusion of more con-
trol groups.1 In the present research the two studies were designed to
complement each other and thus, in combination, rule out alternatives
including effort and content differences. As a result the already
complex designs were not further elaborated with extra control groups.
Further attacks on the role-playing effect should be programatic efforts
including a number of methodological pilot studies designed to assess
the consequences of the confounds usually created in an active-passive

participation manipulation.

87

' ‘ Tan—m ' .L'M

Another explanation is highlighted by the finding that instruc-
tions to counterargue had an unanticipated effect. Those subjects
instructed to counterargue produced significantly fewer supporting
arguments than did subjects not given instructions to counterargue.

An examination of the significant Roleelnstructions to Counterargue
interaction indicated that the instructions had a tendency to work as
expected for read subjects. If anything, improvise subjects counter-
argued less with instructions to counterargue. It may be that impro—
vise subjects were put at ease about the exploitative intent of the
exercise by specific instructions to counterargue. The differences
can only be suggestive since the noted differences were not signifi-
cant. One interpretation of the differences in terms of supportive
argumentation is that the manipulation had no substantial effect on
counterargumentation. However, when-it came time to list one's thoughts
those subjects who had been given instructions to counterargue felt
that it would be inappropriate for them to list supportive thoughts.

If that is the case, then the measured differences do not reflect
actual differences in the number of supportive thoughts elicited by
the counterattitudinal material. This suggests a.degree of invalidity
in the measures which were intended to reflect the thought processes
of subjects while they were initially being exposed to counterattitudinal
material.. This very important question of the validity of the counter-
argumentation, tOtal defensiveness, and supportive argumentation measures
will be taken up following a summary of the conjecture presented above.

It is time to bring some closure to this discussion of the counter-
argumentation data. A number of factors have been summoned to account

for the rather large number of unexpected findings. Briefly, it was-

 

88

posited that the results were attributable to some combination of then

following:
(a) Even though situational constraints on counterargumentation
were low, subject characteristics (pretreatment attitudes) led
to a low level of counterargumentation.' (For example, the no
delay-read subjects and the no delay-female subjects of Study 1.)
(b) The situation activated or aroused a motivation to counter-
argue which was more powerful than situational constraints on
the opportunity to counterargue. (For example, the difference
between "improvise" and "read" subjects not given instructions
to counterargue.)
(c) Differences between the conditions in opportunity to counter-
argue were not substantially created or they were created in.a
manner that made confirmation difficult. (For example, the
improvise-read manipulations in both studies.)
(d) The manipulation may have inadvertently reduced the extent
to which the exploitative intent of the experimenter was salient.
(For example, the instructions to counterargue may have had such
an effect for improvise subjects.)
(e), The observed effect may be an artifact of measurement.' (For
example, the observed tendency for those with instructions to
counterargue to list significantly fewer supportive arguments than

did subjects without instructions to counterargue.)

Developing:a Valid Measure of Counterargumentation
There is nothing very parsimonious about the preceding speculation.
The total picture suggests that the hypotheses.were not adequate to pre-

dict the amount of counterargumentation.‘ Nevertheless there were some

 

89
conditions in the two experiments where treatment differences in
counterargumentation occurred, and a few conditions where the results
were as predicted. These instances will-be further examined and.
analyzed from the perspective of concomitant attitude change patterns.
But first it is appropriate to discuss the psychometric properties of
the measures used to assess the thought processes of the communication
recipients.

Much of the foregoing discussion was based on the assumption that
the scores derived from the content analyses were valid measures of
the extent to which a subject counterargued (thought of supportive
arguments, or whatever) during the initial processing of the counter-
attitudinal material. There is some evidence to suggest, however,
that this assumption is only partially correct. A related matter
involves the relative validities of the-different scoring systems.
This.section addresses the evidence that can be brought to bear on
these.issues.

The most clearcut indication that the measures of counterargumen-
tation were contaminated by unwanted factors is provided by the various
order of measurement effects.1 The observed interactions in terms of
counterargumentation, total defensiveness and supportive argumentation
indicate that some of the recorded thoughts may have been evoked by
the particular format of the measuring instrument rather than a reflec-
tion of the thoughts that had occurred during the initial processing
of the counterattitudinal material.

An explanation for the sex-order of measurement interaction is not
readily forthcoming. WOmen were affected by the order of measurement
of attitudes and counterargumentation in Study 1 but not in Study 2.

Men were influenced by the order of measurement in Study.2 but not in

 

 

90
Study 1. The attitudes befbre counterargumentation order suppressed
the number of defensive responses for female subjects while the atti-
tudes after counterargumentation order suppressed the amount of
defensiveness for male subjects.

An effect of measurement which does appear to be.more understand-
able was Observed in Study 2. A manipulation of instructions to.
counterargue (either present or absent) was found to have no effect
on the amount of counterargumentation produced. The manipulation did
have a significant effect on the number of supportive arguments listed.
One interpretation of the data.was that the manipulation failed to
have an effect on counterargumentation but that subjects who were told
to counterargue felt that listing supporting arguments would not have
been apprOpriate.' Thus, the differences may primarily represent the.
demand characteristics of the-situation.-

Partial support for the validity of the measure is provided by
those instances in which the manipulations had the expected effects.
Unfortunately, this evidence is clouded by the large number of instances
where the predictions were not confirmed. Nevertheless, previous
research (especially that of Osterhouse and Brock, 1970) suggests the
technique for assessing the nature of prior thought processes does
have.some.validity. In addition, there seem to be a number of Options
for eliminating the bias discussed above.

Miller and Baron (1971) have suggested a number of innovations
(including such things as the use of recognition rather than recall
questions) which might yield more valid measures. Another solution,
suggested by order of measurement effects, is to modify the instructions
given to subjects. Perhaps the question format employed makes it diffi-

cult for a subject to cooperate in providing the desired information.

 

91 . ‘
To illustrate the point, consider a hypothetical subject who has just Ea;
finished reading or creating a counterattitudinal essay. At this point
the subject is provided a booklet of questions. In short order, he

is told: "You probably had many different reactions to the various

 

ideas"; is faced with an overwhelming total of 15 numbered lines; and.
is told that he will be given three minutes to list any of his previous
thoughts. Under such circumstances a subject is likely to feel a great
deal of pressure to generate a large number of thoughts and may list
any thought which would help fill the page. Any situatiOnal pressures
encouraging a particular type of response would be readily noticed.

As a corrective it is suggested that subjects be informed that
it is not necessary to complete any specific number of lines. Subjects
might also be told that some participants only have a thought or two.
Furthermore, a comment like the following could be added: "Think.
back carefully and try to remember what your reactions were. Please
list only those thoughts which occurred to you while you were reading

(working with) the material on . .You will be given an

 

Opportunity later to express.any of your subsequent reactions." To
make sure that the instructions are attended to it might be helpful for
the experimenter to read them aloud while the subject follows along.

To facilitate this recall, the time limit would probably need to be.
extended. If it is possible for subjects to accurately recall their
prior thoughts without substantial distortions, additions, or other
modifications, the approach outlined above may be useful. This solu-
tion to the problems of validity is noticeably different from that
which advocates the use of the smallest practical time limit. Research

to-test the utility of the two approaches is clearly called for.

92

While the primary emphasis in this research was not on methodo-
logical matters there are several noteworthy contributions in this
regard which are relevant at this point during the discussion of
validity. First, consider the evidence with regard to the assumption
(by Brock, 1967) that repeated arguments should not be counted twice
during the content analyses of subjects' thoughts. It can be seen
from a comparison of Tables 2 and 14 that there is no reason to assume
that including repeated arguments lowers the validity of the response
measure. In fact, in Study 1 the three minute scoring system which
included repeated arguments was the only one to detect differences
in defensiveness between male and female subjects. Research by Brock
(1967) and Osterhouse and Brock (1970) on the effect of communication
discrepancy on counterargumentation would suggest that since the.
females had pretest attitudes more consistent with the persuasive
materials, they would counterargue less than males. The mentioned
finding strongly suggests that the scoring system is more valid when
repeated arguments are included.

Overall it can be seen that the system including repeated arguments
is virtually as sensitive to treatment effects as the system which
excludes repeated arguments.‘ Attitude change effects occurred in
Study 1 for the sex-delay interaction and in Study 2 for the role-
instructions to counterargue interaction. The 3 manute RAE system was
slightly more sensitive in detecting corresponding treatment effects on
counterargumentation but the differences are slight. Also the correla-
tion between counterargumentation and attitude change is highest with
the 3 minute RAI system, although this can.partially be attributed to

the fact that this measure is "longer".than the other."

93

A useful technique introduced in,this research concerned the pro-
cedures with respect to the time limits on the thought listing question.
While subjects were told that they would be given three minutes to
list their thoughts concerning the material they had been working with,
measures were also obtained after a 45 second interval. This was
accomplished unobtrusively inStudy l; the experimenter merely looked
over the subject's shoulder and noted the progress after 45 seconds.

In Study 2 (in which subjects were scheduled in groups) after the first
45 seconds, the subjects were asked to draw a line through the last~
word they had written and then to continue with the question. Since
the optimal time limit for this measure is yet to be determined, the
advantages of two observations per subject, per session. are obvious.

Contrary to what Osterhouse and.Brock (1970) reported for a one
minute.period, the 45 second interval was sensitive to many treatment
effects. It can also be seen by examining Tables 2 and 14 that the
45-second limit measures were not as sensitive to as many effects as '
the 3 minute measures were. The measures with the shorter time limits
did not detect differences in counterargumentation for treatment combina-
tions which ledto differences in attitude change, although the 3
minute measures did. The problem here is an unavoidable one in estab-
lishing the validity of a construct. If a measure does not indicate
differences where they theoretically should be, it may be the fault-of'
the theory or it may be the fault of the measure of the construct. In
this case, since there is not enough evidence to reject the hypothesis
that counterargumentation moderates the amount of attitude change
resulting from role-playing,.it is assumed that the 45-second measure
is less valid than the 3 minute measure. The evidence is not all one-

sided,.however.v A glance at Tables 2 and 14 reveals that in a majority

94

 

of the cases the shorter time-limit measure recorded effects at lower
probability levels than their longer counterparts. While the magnitude
of the differences is not great, the consistency of the direction should
not be overlooked. It is probably as Miller and Baron (1971) have
noted; the Optimal time limit will vary from situation to situation

and pilot work will be necessary in each case to determine satisfactory
limits.

Future research can easily adapt the technique employed here and
obtain two or more readings at whatever intervals seem appropriate.

The value of the evidence with respect to time limits is not that it'
allows the selection of one technique over another, but rather that

it provides a demonstration of a workable method for obtaining multiple
measures within a single session for a single~subject.‘

As a final methodological point. the category system used for
classifying subjects"thoughts deserves some comment.' In the present
research categories existed not only for counterargumentation, but also
for general disagreement. source, derogation, communication derogation.
supportive argumentation. ambivalent reactions, comments-related to
performance or task, and statements with unclear bearing on the attitude
position being studied. The differentiated set of categories for
defensiveness was created in order to more accurately monitor the types.
of responses that are involved in resisting persuasive appeals. Theoreti-
cal discussions in the literature have suggested that there are several
_different types of defensive reactions and that they may function as
alternative modes of resistance (Miller and Baron, 1971).

Several findings indicate.the potential utility of the expanded
category system.‘ It was observed in Study 1 that order of measurement

had a significant effect on total defensiveness but not on

95

counterargumentation. Moreover, in Study 2 the role manipulation
affected counterargumentation but not total defensiveness. In sum,
the two measures were differentially sensitive to the separate manipu-
lations. This lends some credence to the proposition that each type
of defensive reaction may be specific to certain sets of situations.
Communication derogation may be more common than counterargumentation
under some circumstances than others, for example. With the scoring
system develOped here it will also be possible to determine whether
the role-playing effect is due to the suppression of counterarguments,
the facilitation of supporting arguments, or some combination of the
tWO.‘

The present research has not taken full advantage of the potential
of the category system. For one thing, all arguments were assigned
the same weight in analyses (except for repeated arguments). But-is
the reaction of general disagreement equivalent to a single countere
argument? And, are all arguments of equal importance in supporting
or refuting a given position? The answer to both questions is probably
not. This seems to indicate that more precise estimates of the degree
and importance of counterarguing will result from a system which
assigns more weight to the thoughts that subjects feel are more important.
in-forming an opinion on the issue. Future efforts could be beneficially
directed toward this matter.

Evidence Concerning the Relationship Between
Counterargumentation.and Attitude Change

Attention is now turned to the major hypothesis of this_research.

An effort will be made to describe the evidence that can be brought t0~

bear on the proposition that role-playing is particularly effective in-

 

96
producing attitude change because it reduces the incidence of inter-
fering responses such as counterargumentation.

The strongest support for the above hypothesis was found in Study.
1 and involves the Sex-Delay interaction. Male subjects counterargued
less and changed their attitudes more when they performed under a .42
delay in auditory feedback than when there was no delay. Women counter-
argued more and changed their attitudes less when presented with the
.42 second delay than with no delay. While for both men and women
there is an inverse relationship between counterargumentation and
attitude change as predicted, the interaction was unexpected.‘ Women
counterargued more under conditions where they were expected to
counterargue less.

One possibility is that women in the .42 second delay condition
perceived the manipulation as the violation of a norm which states that
women should be accorded the right to speak without being interrupted.
Another possibility is that these women may have resented the attempt
by‘a male experimenter to make them sincerely say that males and
females should not-live together on the same floor in a residence,
hall. The net effect of either of these possibilities might have been
that women in the .42 second delay condition were more motivated to
counterargue than "no delay" females or either of the male groups.-

As discussed before, if the motivation to counterargue was a more
powerful determinant than the situational constraints, this may have
resulted in the higher level of counterargumentation for these .42»
second delay females. (Women with a higher verbal fluency may have
found it easier than men to overcome the obstacles posed by the
delivery of a speech.) If this line of speculation can be granted,
then the data provide good support for the proposed link between

counterargumentation and attitude change.

H.- .Hs-r 1...
l
1

 

 

97

The other effect in Study 1 is a little less easy to deal with
even though the covariation of counterargumentation and attitude change
is as predicted. It was observed in the first study that subjects
who indicated their attitudes befbre counterargumentation changed
their attitudes more, thought of more supportive arguments, and fewer
counterarguments-than did subjects who indicated their attitudes after
counterargumentation.

The most probable explanation is that the post-test, which came
immediately after the subject had completed his performance. cued the
subject that the experimenter was interested in whether his attitudes’
changed. In_accommodating this implicit request the subject could have
first expressed less positive feelings about coed living on the
semantic differential and then listed a relatively high number of
arguments supportive of the advocated policy. Subjects receiving the
counterargumentation measure first, presumably were under less pressure-
to show any change in attitude because the first instruction they
received sanctioned bOth pro and con-thoughts-about the issue. Thus
there was less need to list supportive arguments and.less reason to
indicate.more.acceptance of the policy concerning coed living.

In Study 2, "improvise" subjects changed their attitudes more than
"read" subjects but only under canditions in which instructions were‘
given to consider counterarguments. The improvise-read manipulation
had no differential impact when there were no instructions to counter-
argue. Supplementary data indicated that there was a marginal tendency
for those with instructions to be.more satisfied with their performance."
This suggests that the effect of instructions to counterargue may-have
been to put subjects at ease about the exploitative intentions of the

experimenter. From this starting point read subjects could have gone

98 J

on ahead with their assigned task of thinking about ideas contrary to
the ones they were being exposed to. The result was a reduction in

the tendency to accept the persuasive material for read subjects. The
improvise subjects, freed from concerns of being deceived, were probably
not as motivated to counterargue as improvise subjects not given
instructions to counterargue and as a result were more persuaded.

An argument very similar to the preceding one was offered by Janis
and Gilmore (1965) who suggested that a gain in attitude change would
not be as likely to result from role-playing if there were negative
incentives in the role-playing situation such as those provided by a
sponsor with exploitative or self-aggrandizing characteristics. Data
from research by Janis and Gilmore (1965) and.by Elms and Janis (1965)
on actively role-playing subjects supported the hypothesis about the.
effect of sponsor characteristics on attitude change. Perhaps the-
implications of these different bits of evidence is that the role-
playing effect occurs only when the sponsor or source is benign and
nonexploitative in nature.

The biggest problem with this explanation is that there were no
counterargumentation differences between read and improvise subjects
who were given instructions to counterargue--even though there were
significant differences between these groups in terms of communication
acceptance. What is more, of those subjects who were not given
instructions to counterargue, "read" subjects listed fewer countere
arguments than "improvise" subjects but did not have significantly
more favofiable attitudes. Taken\at face value the data do not support
the hypothesis that role-playing is especially effective because it
leads to a reduction in the incidence of counterargumentation and other

interfering responses.

 

99

A comment was made previously that a correction should be made
in.the scores of improvise subjects because they listed significantly
more total number of responses, more supportive arguments, and more,
counterarguments. However, when the score values for each type of
response are converted to percentages for eaCh group, the same pattern.
of differences arises as when absolute number of responses was con?
sidered. Again the evidence just does not seem to support the hypothe-
sized relationships.

The only other evidence bearing on the hypotheses is that with
regard to within-cell correlations. In Study 1 the correlation of
counterargumentation and total defensiveness, respectively, with
attitude change was .299 and .306. In Study 2 the corresponding
correlations were .304 and .346. Both these suggest that those.pe0ple
change their attitudes most who counterargue least. It is important
to note that these data and the data with respect to the experimental
manipulations do not unequivocally establish the fact that counter-
argumentation came first and that the attitudes followed as a consequence.
It is just as possible that peOple change their attitudes and then
develop arguments to be consistent with them.‘

In view of the equivocality about the time order and the diffi-
culty of manipulating counterargumentation with role—playing manipula-
tions, it is suggested that a more direct method of manipulating
counterargumentation is needed. DePolito3 has proposed one such method.
Essentially subjects would be presented with a list of arguments. After
each argument, the subject would be required to write either another

argument consistent with it or one opposed to it. In this manner the-

 

3Personal communication.

 

100

relative degree of counterargumentation could be manipulated. While
this technique would not rule out the possibility that anticipatory
attitude change would occur before the subject had a chance to counter-
argue, the tendency for this to occur could be assessed and accounted
for.

In summary, the basic findings of the two studies were that:
(a) There was little evidence to support the notion that there is less
counterargumentation for active participation role-players than for
passive exposure subjects.
(b) While the manipulations of Study 1 apparently created different
levels of effort under moderately high free choice circumstances, the
attitude change effects were not explainable in terms of an effort
justification hypothesis.
(c) Satisfaction with task performance and recall of counterattitudinal
arguments could not account for the attitude change patterns.
(d) On a post hoc basis, the data for the sex-delay interaction in
Study 1 was interpreted as supportive of the hypothesized mediating
role of counterargumentation in the role-playing effect. However, the
data for the role-instructions to.counterargue interaction in Study 2
was not consistent with the hypotheses.

Inna methodological vein, the research has demonstrated the:
(a) need for research on the modification of the instructions on
counterargumentation measures to reduce the demand for irrelevant and
inValid responses.
(b) need for the utilization of pre—post designs, since counterargumen-
tation is to a large extent a function of initial position.‘
(c) utility of a thought listing procedure which permits multiple time-

limit measures for each subject.

101
(d) utility of a scoring system which includes several types of

defensive reactions as well as supportive responses.

REFERENCES

REFERENCES

Aronson, E., Turner, J. A., and Carlsmith, J. M. Communicator credi-
bility and communication discrepancy as determinants of Opinion
change. Journal of’Abnormal and.Social Psychology, 1963, 61,
31.36 0

Ben, D. J. Self-perception: an alternative interpretation of cognitive
dissonance phenomena. Psychological Review, 1967, 14, 183-200.

Brehm, J. W.r A Theory of'Psychological Reactance. New York: .Academic
Press, 1966.

Brewer, M. B.. and Crano, W. D. Attitude change as a.function of
discrepancy and source of influence. Journal of'Social Psychology,
1968, 163.13-18.

Brock,.T. C. Communication discrepancy and intent to persuade as
determinants of counterargument production. Journal oj’Eaperi-
mental Social Psychology, 1967, g, 296-309.

Carlsmith, J. M., Collins. B. E., and 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
anonymous essay writing. Journal of’Personality andISocial
Psychology, l966.q§, l-13.

Cohen, A. R., Brehm. J. W., and Fleming, W. H. ‘Attitude change and
justification for compliance.“ Journal of’Abnormal and Social'
Psychology, 1958, §§, 276-278.

Collins, B. E. Social Psychology. Reading, Mass.: Addison-Wesley,
1970.

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

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

Elms, A. C.‘ The influence of_fantasy ability on attitude change through

role-playing.- JOurnal of'Personality and’Social Psychology.
1966, 3, 36-43.

102

103

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

Elms. A. C. and Janis, I. L. «Counter-norm attitudes induced by consonant
versus dissonant conditions of role-playing.‘ JOurnal of'Emperi-
mental Research in Personality. 1965, 1, 50-60.

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

Freedman. J. L., Carlsmith, J. M. and Sears, D. 0.‘ Social Psychology.
New Jersey: Prentice-Hall, 1970..

Freedman, J., and Sears. D. Warning, distraction and resistance to
influence. Journal of'Personality and Social Psychology, 1965,
1, 262-266.

Greenwald, A. G. The Opendmindedness of the counter-attitudinal role
player. Journal oj’Eaperimental social Psychology: 1969.;5,
375-388. ' :

Greenwald. A. G.- When does role playing produce attitude change?
Toward an answer.‘ Journal of'Personality and Sbcial Psychology.
1970, 16, 214—219.

Harvey,.0.,and Beverly, G. Some personality correlates-of concept
change through role playing. Journal of’Abnormal andISocial
Psychology, 1961. §§,*125-130.

Insko, C. A. Theories of'Attitude Change. New York: Appleton-
Century Crofts, 1967. .

Insko, C. A., and Schopler, J. Esperimental social Psychology. New
York: Academic Press, 1972.

Janis, I. L., and Gilmore, J. B. The influence of incentive conditions
on the success of role playing in modifying attitudes.‘ JOurnal
of'Personality and Social Psychology, 1965, 1, 17-27.’

Janis, I. L., and King, B. T. The influence of role-playing in opinion
change. Journal of'Abnormal and SOcial Psychology, 1954, 42,
211-218.

Janis, I. L., and Mann, L. Effectiveness of emotional role-playing in
modifying smoking habits and attitudes. Journal of’Esperimental
Research in Personality, 1965. 1, 84L90.

Jones, E. E., and Gerard, H. 3.: Fbundhtions of'Social Psychology.
New York: Wiley. 1967.

King, B. T., and Janis. 1., Comparison of-the effectiveness of improvised
vs. non-improvised role playing in producing opinion changes.‘
Human Relations, 1956, 2, 177-186.

104

Linder, D. E., Cooper, J., and Jones, E. E. Decision freedom as a
determinant of the role of incentive magnitude in attitude change.
Journal of’Personality and social Psychology. 1967, g, 245-254.

Linder, D. E., and Worchel. S. Opinion change as a result of effort-
fully drawing a counterattitudinal conclusion. JOurnal oj’Eaperi-
mental Social PsychOlogy, 1970, g, 432-448.

Little, D. M. Suspension of critical judgment: an approach to role-
playing induced attitude change. Unpublished master's thesis,
Michigan State University, 1971.

MtGuire, W. J. Inducing resistance to persuasion: Some contemporary.
approaches. In L. Berkowitz (Ed.), Adbances in Experimental
Social Psychology. Vol. 1. New York: Academic Press, 1964.
Pp. 191-229.»

McGuire, W.J. The nature of attitudes and attitude change. In G.
Lindzey and E. Aronson (Eds.), The Handbook of'Social Psychology.
Vbl. 3. Reading, Mass.: ‘Addison—Wesley, 1969.

Mann, L., and Janis, I. L. A follow-up study on the long-term effects
of emotional role playing. Journal of'Personality and Sbcial’
Psychology, 1968, 8, 339-342.

Miller, N., and Baron, R.S.- 0n measuring counterarguing.. Paper pre-
sented at American Psychological Association, Washington, D.C.,
September, 1971.

Osterhouse, R. A., and Brock, T. C. Distraction increases yielding to
propaganda by inhibiting counterarguing. Journal of'Personality
and Social Psychology, 1970, 15, 344-358. '

Scott, W. Attitude change through reward of verbal behavior. Journal
of Abnormal and Social Psychology, 1957, 2, 72-75.

Sears, D. 0., and Abeles, R. P.‘ Attitudes and opinions. In P.
Farnswroth (Ed.), Annual Review of'Psychology, 1969,.gg, 253-288.

Watts, W. A. Relative persistence of opinion change induced by active
compared to passive participation. Journal of'Personality and
Social Psychology, 1967. 5, 4-15.'

Weiss, R. F. Role playing and repetition effects on opinion strength.
Journal of’Social Psychology, 1971, 82, 29-35.

Zimbardo,.P. G. The effect of effort and improvisation on self-

persuasion produced by role-playing. Journal of’Esperimental
social Psychology, 1965,‘1, 103-120.

Zimbardo, P., and Ebbesen, E. B.. Influencing Attitudes and Changing
Behavior. Mass.: Addison-Wesley, 1969.

Emu-ah...
.

 

APPENDICES

APPENDIX A

INSTRUCTIONS FOR "READ" CONDITIONS (STUDY 1)

APPENDIX B

INSTRUCTIONS FOR "IMPROVISE" CONDITIONS (STUDY 1)

The research project scheduled for today is part of a program of
research systematically exploring interpersonal communication. What
we'll be looking at in the present study is most commonly called
improvisational speech. Basically, we are_trying to find out how
several different variables are related to the manner in which a person
presents a talk on a tOpic of current interest.‘ To put it another way,
our interest is in the things which affect and are affected by how a person
favorably expresses himself'when”on the spur of the moment he is asked to
speak on a controversial issue. ~This will become clearer as we proceed.

But before we get into that, there's another matter I'd like to
have you help me with. And that is, I‘m trying to find issues which
are interesting to use in studies like these. So I have a number of
issues which I'd like to have.you rate on some rating scales.

To get on with the specifics of this particular study, what we
would like you to do is present a talk--a sort of speech-discussion.

You will be provided with some materials on the issue of (in this case)
coeducational dormitory living. Your task will be to use these argu-
ments to present a speech in your natural argumentative style. Try to
be as sincere and convincing as possible in presenting the same point
of view as the one taken in the material you will be given entitled:

"A position paper: Coeducational living at Michigan State University."
Use all the arguments listed and feel free to add any other arguments

107

1' , Fri-damn.— .-
I 'L I
I. ..
4 I

108
or examples you want to as long as they are.consistent with the ones
that have been provided. Try to be as comprehensive as possible when
presenting the speech. You will be given three minutes to rehearse
the speech silently. Even though the circumstances are rather unique

and the task is perhaps a little difficult, try to imagine that you

actually are a person with beliefs like those expressed in the material-

on coeducational living.’ Don't be overly concerned by the modifications
of your voice by the recording equipment. (Delay subjects are to be
told: You will notice a slight time lag in your speech which might

be disturbing at first.) Remember, try to be as sincere and convincing
as possible. It may help to imagine you are speaking to a small group
of peOple who are eager to hear what you have to say about the issue.
After the three minute preparation period, I will signal you. Begin
and continue until finished talking despite any experimental variations

you may notice.

' ‘th-
- . L.
.L
A
.
u

APPENDIX B

INSTRUCTIONS FOR "IMPROVISE" CONDITIONS (STUDY 1)

APPENDIX C

FREQUENCY OF OCCURRENCE OF EACH RESPONSE CATEGORY (STUDY 1)

 

APPENDIX C

FREQUENCY OF OCCURRENCE OF EACH RESPONSE CATEGORY (STUDY 1)

Scoring System

 

 

Category 3 min. RAEa 45 sec. RAEb 3 min. RAIc 45 sec. RAId
Supportive 155 67 180 70
Arguments (38.46%) (42.952) (38.962) (42.68%)
Counter- 202 64 225 65
arguments (50.122) (41.032) (48.70%) (39.63%)
General 21 15 23 _ 18
Disagreement (5.21%) (9.62%) (4.97%) (10.981)
Communication 7 5 10 6
Derogation (1.74%) (3.20%) (2.17%) (3.66%)
Source 3 0 3 0
Derogation ( .74Z) ( .002) ( .652) ( .001)
Ambivalence 5 3 10 4
(1.24%) (1.92%) (2.172) (2.442)
Unclear 10 2 ll 1
(2.48%) (1.28%) (2.382) ( .611)
Tota1~ 403 156 462 164
(100100%) (100.00%)« (100.00%) (100.002)

Note 0 --N-80 s

aThreeminute time limit with repeated arguments excluded.

bForty-five second time limit with repeated arguments excluded.

cThree minute time limit with repeated arguments included.

dForty-five second time limit with repeated arguments included.

109

 

APPENDIX D

PRETEST MEASURE OF ATTITUDE TOWARD COEDUCATIONAL LIVING (STUDY 1)

APPENDIX D

PRETEST MEASURE OF ATTITUDE TOWARD COEDUCATIONAL LIVING (STUDY 1)

COEDUCATIONAL LIVING

Good

Mild
Beautiful
Sweet

Strong

Nicer

Fair
Large
Hot
Even

Important

How certain are you

Very Certain

ON THE SAME

O.

FLOOR WHERE YOU LIVE

 

 

 

 

 

 

of

Bad
Intense
Ugly
Sour
Weak
Awful
Unfair

Small

: Cold

Uneven

Unimportant

your opinion regarding this issue?

 

Very Uncertain

Circle the statement which best describes your position concerning

coeducational living on the same floor where you live.

1. I am very much in favor of it.

2. I am in favor of it.

3. I honestly have no opinion about the issue.

110

 

111
4. I an opposed to.it.

5. I am very much Opposed to it.

 

APPENDIX E

MATERIALS INCLUDED IN POSTEXPERIMENTAL QUESTIONNAIRE (STUDY 1)

 

APPENDIX E

MATERIALS INCLUDED IN POSTEXPERIMENTAL QUESTIONNAIRE (STUDY 1)

While reading over and presenting the material concerning coeducational
living at M.S.U. you probably had many different reactions to the
various ideas. List below any of these thoughts--either pro or con--
which may have occurred to you. State your own thoughts and ideas as,
briefly as possible--IGNORE SPELLING, GRAMMAR, AND PUNCTUATION. You
will have three minutes to write your ideas. Please stop writing
immediately and go on to the next page when asked to do so.

1. idea
2. idea
3. idea
4. idea
5. idea
6. idea
7. idea
8. idea
9. idea
10. idea
11. idea
12. idea.
13. idea
14. idea
15.‘ idea

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

112

113

COEDUCATIONAL LIVING ON THE SAME FLOOR WHERE YOU LIVE

Simple : : : : : : : Complex

 

 

 

 

 

 

 

 

 

Honest : : : . : : : Dishonest
Clean : : : : : : : : Dirty
Pleasant : . : : : : : : Unpleasant
Active : : .: .: : : : : Passive
Fragrant : : : : : : ,; : Foul
Kind : : : : : : : : Cruel
Fast‘ : : : : : : : : Slow
Soft : : : : : : : : Loud
Loose :~ : . : : : : : 'Tight
Important : : : : : : : : Unimportant

 

How certain are you of your opinion regarding this issue?

Very Certain : : : : : : : : Very Uncertain

 

1. How sincerely did you present the speech?

Not at all
Sincerely : : .; : : .: : : : : : Very Sincerely

2. How convincingly did you present the speech?

Not at all Very
ConvinCingly : : : : ~ : : : : : : : Convincingly

 

3. How logical were the arguments which you presented?

Not at all
Logical : : : : : : : : : : : Very Logical

a..-

4. How biased were the arguments which you presented?

Notrat‘all
Biased : : : : .: .: : .: : : : Very Biased

 

5. How difficult was it to present the speech?

Not at all

Difficult : : : : : ° : Very Difficult

114
6. How much effort did you put into presenting the speech?
None at all : : : : : : : : : : : A great deal
7. How satisfied were you with the way you fulfilled the task requirements?
Not at all : : : : .: : : : : : : Very much

8. How would you rate your performance relative to other subjects who
have taken part in this study?

Bottom 10% :'. : : : : : : -: : : : Top 10%
9. To what extent were you able to get involved in presenting the speech?
Not at all : : : : : : : :. : : : Very much

10. How knowledgeable about the issue were the people who wrote the
arguments which you read?

Not at all : : : : : : : : : : : Very

11. How trustworthy a source of information were the people who wrote.
the arguments which you read?

Very

Not at all : : : : : :

 

12. Briefly list the main thesis and supporting arguments of the material
which you were presented concerning coeducational living. Use the
back side on this page if necessary.

APPENDIX F

FREQUENCY OF OCCURRENCE OF THE MAJOR RESPONSE CATEGORIES
(STUDY 2)

APPENDIX F

FREQUENCY OF OCCURRENCE OF THE1MAJOR RESPONSE CATEGORIES

(STUDY 2)

Scoring System

 

Category 3 min. RAE 45 sec. RAE 3 min. RAI 45 sec. RAI
Supportive 656a 263b 744° 269Cl
Arguments (43.76%) (41.812) (42.932) (40.452)
Counter- 680 302 775 319
arguments (45.361) (48.01%) (44.722) (47.972)
Defensiveness
other than 66 42 82 45
Counter- (4.41%) (6.68%) (4.732) (6.771)
arguments
Neither
Supportive 97' 22 132 32
nor (6.472) (3.50%) (7.62%) (4.812)
Defensive

Total 1499 (100%) 629 (1002) 1733 (100%) 1 665 (1002)

aThree minute system with repeated arguments excluded.

bForty-five second system with repeated arguments excluded.

cThree minute system with repeated arguments included.

dForty-five second system with repeated arguments included.

115

APPENDIX G

MATERIALS INCLUDED IN POSTEXPERIMENTAL QUESTIONNAIRE (STUDY 2)

APPENDIX C

MATERIALS INCLUDED IN POSTEXPERIMENTAL QUESTIONNAIRE (STUDY 2)

While presenting the material concerning a tuition increase at M.S.U.,‘
you probably had many different reactions to the various ideas.- List:
below any of these thoughts--either pro or con--which may have occurred

to you. State your own thoughts and ideas as briefly as possible--

IGNORE SPELLING, GRAMMAR, AND PUNCTUATION. You will have three minutes

to write your ideas. When asked to do so, please stop writing immediately
and go to the next page.*

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1. idea

2. idea

3. idea f
4. idea 4*

5. idea

6. idea.

7. idea

8. idea

9. idea fig
10. idea,
11. ideagg
12. idea

13. idea
14. idea,
15. idea

 

 

*
The format and order of this question was varied according to
experimental treatment condition.

116

117

To help us understand your reactions to the task requirements, we would
like to have you answer several questions related to the tapic you were
working with. Circle the number under each statement which most directly
corresponds to your feelings. Work quickly but carefully.

1. A tuition increase at Michigan State University would cause more,

problems than it would solve.

 

1 2 3 4 5 6 7.84910

11 ,
Strongly Strongly
Disagree Neutral Agree

2. Out-of-state tuition should be the same as in-state tuition.

 

1 2 3 4 5 6 7 8 9 10

 

11
Strongly Strongly
Disagree Neutral Agree
3. I like the idea of a tuition increase at M.S.U.'
1 2 3 ' 4 5 6 7 8 9 1o 11
Strongly Strongly
Disagree Neutral Agree

4. A tuition increase would facilitate a few needed changes here at

 

 

M.S.U.'

1 2 3 4 5 ' 6 7 8 9 1o 11
Strongly Strongly
Disagree Neutral Agree
5. A tuition increase is not necessary at M.S.U.'

1 2 3 4 5 6 7 8 9 ' 1o 11
Strongly Strongly
Disagree Neutral Agree

6. Most students don't worry very much about the amount of tuition they

 

pay.

1 2 3 4 5 6 7 8 9 1o ' 11
Strongly Strongly
Disagree Neutral Agree

7. Even a moderate increase in tuition at M.S.U. would be unfair.

 

1 2 3 4 5 A 6 7 8 9 1o 11
Strongly' Strongly
Disagree Neutral Agree

118

8. I would be willing to write a letter to a paper such as the State

News advocating a-tuition increase at M.S.U.

 

 

 

 

 

1 2 3 4 5 6 7 8 9 10 11‘
Strongly Strongly
Disagree Neutral Agree
9. The taxpayers are responsible for too great a percentage of the

cost of educating a student.

1 2 3 4 5 6 7 8 .. 9 1o 11
Strongly Strongly
Disagree Neutral. Agree

10. It would be desirable for M.S.U. to raise tuition.
I 1 2 3 4 ‘ 5 6 7 8 , 9 1o 11
Strongly Strongly
Disagree Neutral Agree
11. A tuition increase at M.S.U. would not help cure any of the major
problems of the University.

1 2 3 4 5 6 7 8 9 10 111
Strongly Strongly.
Disagree Neutral Agree

12. Even the present cost of tuition is too high.

1 2 3 4 5 6 h 7 8 9 10 11
Strongly Strongly
Disagree Neutral Agree

DO NOT TURN THE PAGE UNTILaASKED TO CONTINUE

119
For each of the questions below, place an X in the space between the
dots that most closely corresponds to your feelings. If you wrote a
letter, evaluate that letter where asked; if you read a letter written
by someone else, evaluate that letter.

1. How sincerely was the letter written?

Not at all Very
Sincerely : : : : .: : : : : : : : Sincerely

2. How convincingly was the letter written?

Not at all Very
Convincingly : : : .: : : -: : -: : : Convincingly

 

3. How coherently was-the letter written?

Not at all Very
Coherently : 1: : ': a: : .; ': .: : : Coherently

 

4. How logical were the arguments presented in the letter?

Not at all
Logical : : : : : : : : : : : Very Logical

5. How biased were the arguments presented in the letter?

Not at all
Biased- : : : : : : : : : : : Very Biased

 

6. How difficult was it to write the letter?

Not at all
Difficult : : : : : : : : : : : Very Difficult

 

7. How much effort was put into writing the letter?

None at all : : : : : : ‘ : : : : : A great deal

 

8. How knowledgeable about the issue were the pEOplethO wrote the
list of‘arguments you read?

Not at all Very
Knowledgeable : : : : -: .: : : -: : -: Knowledgeable

9. How trustworthy a source of information were the peOple who wrote
the list of arguments which you read?

Not at all ‘ Very
Trustworthy :' : : : : : : : : : : Trustworthy

10. How satisfied were you with the way you fulfilled the task
requirements?

Not at all : : :

: : : Very much

120

11. How would you rate your task performance relative to other sub-
,jects who have taken part in this study?

Bottom 10% : : : : : : : : : : : TOp 102

12. To what extent were you able to get involved in the experimental
tasks?

Not at all : : : : : : : : " : : : Very much
13. How much did you feel pressured to change your opinions?

Not at all. : : : : : : : : : : : Very much
14. Briefly list the main thesis and supporting arguments of the

material you were presented concerning a tuition increase.~ Use
the back side of this page if necessary.

Iglluwm

ms