-
The Journalo/ SocialPsychology,1975,97, 195-208.
THE CATHARTIC VALUE OF SELF-EXPRESSION:
TESTING, CATHARSIS, DISSONANCE, ANDINTERFERENCE EXPLANATIONS*
1
University o] California, San Diego
ROBERT M. KAPLAN
SUMMARY
The effects of anger arousal, type of expression, and
communicationdestiny on anger and aggressive drive are examined.
One hundred tenstudents were exposed to an insulting or a
noninsulting communication.Subjects replied to the communication by
supporting it, opposing it, ortaking a neutral position. Some were
told that their replies would be readby the person who had written
the communication (target), and othersthat their responses would
not be shown to the target. Results show thatanger arousal produced
more hostility than the nonarousal and that angerarousal interacted
with type of expression. Angry subjects who had ex-pressed their
feelings became more hostile than subjects who had expressedthe
opposite of their feelings. Angry subjects who had taken a
neutralposition, however, were lowest in hostility. Experimental
effects attributableto other variables were nonsignificant. The
results are interpreted in termsof a cognitive interference
hypothesis.
A. INTRODUCTION
A notion widely accepted by psychologists, psychiatrists, and
lay peopleis that the expression of aggression, hostility, or rage
reduces the consequentprobability of the occurrence of aggressive
behavior (5, 9, 10, 11, 12, 19,28, 29). Such changes in aggressive
behavior or affect are usually labeled"catharsis" or a "cathartic
effect."
Many psychologists do not accept the existence of catharsis and,
in fact,
* Received in the Editorial Office, Provincetown, Massachusetts,
on May 22, 1974.Copyright, 1975, by The Journal Press.
a Some of the data presented herein also appeared in a doctoral
dissertation presentedto the University of California, Riverside by
the author. The helpful comments of RobertD. Singer, Roy D.
Goldman, and Arthur Bohart on an earlier draft of this paper
aregratefully acknowledged.
2 Requests for reprints should be addressed to the author at the
address shown at theend of this article.
-
1_6 _OURNAL OF SOCIAL PSYCHOLOGY
feel that the expression of aggression, anger, or hostility
leads to increases
rather than decreases in aggression (2, 21, 26). One theoretical
explanationof such views fits within the framework of the theory of
cognitive dissonance(13). Expressing hostility toward someone would
be dissonant with any cog-
nitions that the target may be a good or reasonable person or
with the aggres-sor's view of himself as a peaceful or friendly
individual. In order to reduce
the dissonance resulting from an attack and to justify the
attack on him
it may be necessary to denigrate the target of aggression.
Evidence tends to
show that disliked persons or groups are highly likely targets
for aggressivebehavior. Once denigrated, a person or group is more
likely to be attacked
than before. Aronson (2, p. 157) sums up the dissonance view
suggestingthat, "Violence does not reduce the tendency toward
violence: violence
breeds more violence."
Another possible interpretation of the catharsis and aggression
literature
is that conditions that force subjects to focus on anger enhance
the main-
tenance of high anger levels. According to this viewpoint,
symbolic expression
may only serve to alert the subject about his own hostility
level. The expres-
sion of anger by angry persons may result in maintenance of
aggressive drive.
However, conditions which produce cognitive interference with an
annoying
event would permit anger to dissipate.
Experimental studies have often served to confuse rather than
clarify thisdifference in theoretical outlook because researchers
have allowed considerable
variability in methods of anger arousal, modes of aggressive
expression, andchoice of dependent measures.
For the purposes of this discussion, aggression refers to
behavior which is
designed to result in harm to some person or his property (14);
anger con-notes an emotional state with autonomic correlates which
can serve to ener-
gize aggressive behavior. The terms hostility and aggressive
drive are used
interchangeably and refer to a negative attitude or feeling of
ill will about
people or events (6). In addition to these commonly used labels,
the term
hostile behavior is used to connote a composite of hostility and
aggression;it refers to the behavioral component of hostile
attitudes.
An experiment which considers anger (an emotional state),
aggressive
drive (an attitudinal state), and hostile behavior (a form of
aggression) has
been undertaken to clarify the role of self-expression in the
reduction of anger
and aggressive drive. Some of the variables manipulated in the
experiment
include (a) anger arousal; (b) type of expression; and (c)
communicationdestination.
-
ROBERT M. KAPLAN 197
1. Anger Arousal
The anger arousal phase in catharsis experimentation is of major
theoreti-cal and methodological import. Several literature
reviewers (e.g., 6) havesuggested that the presence or absence of
anger arousal may account for someof the discrepant results in
catharsis research. Buss (6) contends that ex-pressing aggression
will produce a cathartic effect for angry subjects. Ifsubjects are
not angry, expression of aggression may teach them to
behaveaggressively on subsequent occasions, thus producing an
increment due tolearning.
2. Type o/ Expression
Different theories would make different recommendations about
how hos-
tility could be reduced in angry subjects. Catharsis theories
predict thathonest, direct expressions would be most effective.
Dissonance theory holdsthat counterattitudinal role play would
reduce the most anger. The inter-ference viewpoint maintains that
any type of expression which interfereswith anger related
cognitions will reduce hostility.
3. Message Destination
Berkowitz (4, 5) among others (3, 10, 18) maintains that the
occurrenceof a cathartic effect depends upon the degree to which
the subject believeshis hostile behaviors will affect the person
toward whom they are directed.Thus, the destination of a message
may be a crucial variable in catharsisresearch.
Data on communication destination may be relevant to Collins'
revisionof dissonance theory. Collins (7) and Hoyt, Henley, and
Collins (20) haveproposed that an individual will only experience
dissonance when aversiveconsequences to himself or others result
from his attitudinal or counteratti-tudinal expression. Thus they
predict that the dissonance effect will occurwhen a self-expression
is presumed to reach its destination and to have anoxious
effect.
4. Summary o/ Theoretical Predictions
In the present experiment 55 subjects read an insulting
communicationand 55 subjects saw a noninsulting communication. All
subjects replied tothe communication, either supporting it,
opposing it, or taking a neutralposition. Some subjects were lead
to believe their replies would be shown tothe person who had
written the communication (target), while other sub-jects were told
their replies would not reach their target.
-
198 JOURNAL OF SOCIAL PSYCHOLOGY
Several theories predict different outcomes for the experiment.
For angrysubjects, the catharsis and the learning viewpoints
predict that direct attitu-dinal expression will be the most
effective treatment for reducing aggressivedrive. The dissonance
position predicts that counterattitudinal expressionwill be the
most effective means of reducing hostility and that direct
expres-sion will be the least effective treatment. The learning and
dissonance posi-tions both predict an increment in hostility for
nonangry counterexpressionsubjects. The catharsis viewpoint does
not make this prediction.
The interference position predicts that neutral activity will
lead to morehostility reduction than counterattitudinal expression,
since the latter wouldalso draw some attention to the anger
arousing incident. For nonangry sub-jects, counterattitudinal
expression should serve to increase self-perceptionsof hostility.
Direct attitudinal expression and neutral activities could
beexpected to have little effect.
B. METHOD
1. Subjects
The subjects were 50 male and 60 female introductory psychology
studentswho were enroIIed at San Bernardino Valley College, a
California communitycollege with students from a variety of ethnic
and social class backgrounds.
2. Independent Variables
The experiment was purported to be on communication. The
experimenter,who was introduced as an assistant to a researcher
interested in the scientific
study of journalism, explained that the exercise involved
reading and respond-ing to letters to the editor. The experiment
included several phases:
a. Anger Arousal. Anger was aroused in half of the subjects, and
notaroused in the remaining half. This was achieved by allowing
subjects toread a letter to the editor which they were told had
recently been printed ina large newspaper. The letter concerned a
proposition on the Californiaballot which would provide for funding
of California Community Colleges,an issue known to be of interest
to the students. For the Anger Arousalcondition, the letter urged a
vote against the proposition. In the argument,the letter writer
made several degrading remarks about community collegestudents. He
claimed they were stupid, irresponsible, and not worthy of
freeeducation. The letter for Non-Anger Arousal condition favored
the proposi-tion, adapting arguments from the campaign literature
to emphasize the needfor additional support for community colleges
and to focus on the need forexpanded funding of these
institutions.
-
ROBERT M. KAPLAN 199
b. Expression Type Manipulation. All subjects in the Direct and
Coun-terexpression groups were asked to reply to the letter they
had read in themanner specified by the instructions. There were
three versions of instruc-tions for the reply: Forty-four subjects
were instructed to reply to the lettersupporting the Position taken
by the letter writer. Angry subjects receivingthis instruction were
classified as the Counterexpression group. Nonangry
subjects assigned to this task were classified as the
Direct-expression group.
Another 44 subjects received a similar instruction asking them
to oppose theposition taken by the letter writer and to attack the
man and his ideas in
the reply. These subjects represented the Angry
Direct-expression group and
the Not Angry Counterexpression group. In addition, there was an
Angryand a Not Angry Neutral Expression group. The instruction to
these 22
subjects asked them to write an essay about the value of letters
to the editor
without mentioning either the letter they had read or its
author.
c. Communication Destiny Manipulation. A note on the page
followingthe space for the reply was used to manipulate
communication destiny. In
the Direct and Counterexpression conditions, the note thanked
subjects forwriting the reply. The remainder of the note was in one
of two forms. In the
To-target Condition, it was explained that the reply which had
been written
would be shown to the letter writer when he visited the college.
In the Not-to-target Condition it was emphasized that the reply
would only be used for
the purposes of the research and would not be shown to the
letter writer when
he made his campus visit. The note was not used for neutral
expression groups,since it would have had no meaning for them.
3. Dependent Measures
a. Manipulation Check. Just after the anger arousal phase of the
exper-
iment, several scales were administered so that the effect of
the manipulation
could be evaluated. These items were on a single page of the
experimentalbooklet which immediately followed the letter to the
editor. The first two
entries on this page were dummy items used to make the study
appear au-
thentic. The first asked if any of the letters concerning
Proposition 1 had
been read in the newspaper. The second item asked if the actual
letter used
for the study had been read previously. The next item was
included to deter-
mine the subject's orientation with respect to the letter. It
asked whether
the subject agreed with the comments made by the letter writer.
Another
item probed voting intention for Proposition 1.Following the
questions was a series of four semantic differential scales.
The poles of the scales, separated by a seven choice response
space, were as
-
200 JOURNAL OF SOCIAL PSYCHOLOGY
follows: .active-passive, relaxed-tense, angry-pleased, and
good-bad. Thesemantic differential scales were used to tap
affective changes which mayhave resulted from exposure to the
letter.
b. Attraction. As a measure of interpersonal attraction,
subjects wereasked to rate the target on a number of traits. The
traits were chosen fromAnderson's (1) list of traits which are
attributed to highly likable and highlydislikable persons. A cover
story explained that the letter writer had beenapproached and
interviewed, and that he had responded to items from wellknown
personality tests. The subject's task was to see how accurately
hecould fill in information on the basis of minimal exposure to the
letter writer.These appraisals were to be compared to the
personality test results. Thejudgments of each trait were made
along 13 point scales where 0 indicatedthe trait described the
writer well, 12 indicated the trait did not describehim well, and 6
was a neutral point. A similar story used in a recent stud5(27)
appeared to be taken at face value.
c. Affective Measure. The Abasement (aba) and Aggression (agg)
scalesof the Gough Adjective Checklist (16) followed the attraction
measures andwere used to evaluate the affective or anger state.
Instructions asked thesubject to mark each adjective which
described how he was feeling at thatmoment.
d. Behavioral Task. On the final page of the booklet, a note
explainedthat all of the people who had written letters to the
editor had been invitedto participate in a panel discussion at the
college during the week before theelection. The guests were to be
paid from a guest lecturer fund, but the exactamount had not been
decided upon. The note then explained that since thesubject was one
of the few people at the college who had been exposed tothe letter
writer's opinions, it would be appropriate for him to suggest
howmuch money the letter writer should receive for his visit.
Thirty dollars wasgiven as the usual fee, and the subject was led
to believe that his suggestionwould actually affect how much the
person would receive. The subject wasthen asked to select one of 11
values which ranged from $5 to $55 at fivedollar intervals. The
amount selected was taken as an index of hostile be-havior.
e. Cognitive Measure. Greenberg and Tannenbaum (17)
demonstratedthat angry subjects made more spelling and grammatical
errors while encod-ing than nonangry subjects. They suggested the
number of errors may be avalid index of cognitive stress. Each
subject's reply was read, and errors weretallied by two female
graduate students who had backgrounds in language.
-
ROBERTM. KAPLAN 201
C. RESULTS
1. Data Trans]ormations
Data for some of the dependent measures had to be transformed or
re-organized. Ratings of traits used as attraction measures were
factor analyzed(using varimax rotation of the principal components
matrix). The factoranalysis revealed that two factors accounted for
a substantial proportion ofthe variance. These factors each
represented clear conceptual dimensions andwere dubbed Trustworthy
and Annoyance. Subscales were calculated bysumming together ratings
for the four traits which loaded above a .50 criterionon each
factor. Trustworthy scores were calculated by summing together
rat-
ings for the traits: competent, dependable, helpful, and
sensible. Annoyancescores were obtained by summing together ratings
on the traits: obnoxious,narrow-minded, irritating, and
overcritical.
Spelling and grammatical errors were analyzed in terms of errors
perhundred words.
2. Manipulation Checks
In preparing the ficticious letters to the editor, two
assumptions weremade. These were as follows: (a) subjects would
disagree with the angeringletter and agree with the nonangering
letter; and (b) the angering letterwould make subjects more angry
than the nonangering letter.
To check on the former assumption, comparisons between subjects
who hadreceived the different letters were made for the question,
"Do you agree withthe comments made by the letter writer?" Four of
the 55 subjects exposed tothe angering letter claimed to agree with
the author's views and only threeof the 55 subjects who had read
the nonangering letter indicated disagreement(X2 -- 73.39, d] : 2,
p < .001).
To determine whether the angering letter produced feelings of
anger, re-sponses to the "angry-pleased" semantic differential
scale were analyzed. Theseven point scale was scored so that low
scores would indicate response towardthe "angry" pole and high
scores would reflect responses toward the "pleased"
pole. The mean for subjects exposed to the angering
communication (x -- 2.87)was considerably closer to the angry pole
than the mean for subjects who
were exposed to the nonangering communication (x--5.61).
Statisticalanalysis showed this difference to be significant beyond
the .0001 level[t (108)---10.66]. Subjects exposed to the angering
communication alsoreported feeling more tense It (108) -- 2.20; p
< .03], active [t (108) --
-
202 JOURNAL OF SOCIAL PSYCHOLOGY
--3.71; p < .001], and bad [t (108) = 3,54; p < .01] than
those exposedto the nonangering communication.
3. Pretreatment Comparisons
Although the angering manipulation occurred early in the
experiment, theother independent variables were introduced after
the set of manipulationchecks. Analysis of the four questionnaire
items and the four semantic dif-ferentials showed no differences
between subjects who were assigned to thevarious communication
destiny or expression type conditions. Similarly, therewere no
differences by sex, and all interactions were nonsignificant.
There-fore, it seems safe to postulate that there were no
differences between subjectswithin each anger arousal group before
the other manipulations were intro-duced.
4. Preliminary Analysis
The standard deviations and intercorrelafions of all dependent
variablesare presented in Table 1.
The data were first analyzed without neutral groups included but
withthe addition of sex of subjects as a variable. This permitted a
2 X 2 X 2 X 2(Anger X Communication Destination X Type of
Expression X Sex) multi-variate analysis of variance (MANOVA). The
results of this analysis revealeda highly significant effect for
the Anger manipulation [F (11,62)= 20.41;p _ .0001 ] and a
significant interaction between Anger and Expression Type[F
(11,62)--2.02; p _ .05]. All main effects and interactions
associatedwith Communication Destiny manipulation and the Sex of
subjects werenonsignificant.
In order to provide a clearer picture of the significant
effects, data werecollapsed across the two variables which failed
to provide statistically reliableresults. Since the Neutral
Expression groups contained half as many subjectsas the other
experimental groups (experimental groups having been combinedover
Communication Destiny were double in size), the cell sizes were
unequaland the design was nonorthogonal. In order to obtain
unbiased F contrasts,the contrast sequence reordering technique
(15) was employed. Each con-trast of interest was obtained by
subtraction from the between groups sum
of squares and cross-products matrix (SSCP) after all other
contrasts hadbeen subtracted. As a result of this process, each
multivariate F ratio isconservative and unbiased.
In the results of the 2 X 3 (Anger X Expression Type) analysis
the datashowed a significant effect for Anger Arousal [F (7/98) =
16.12; p < .0001]
-
TABLE 1
STANDARD DEVIATIONS AND INTERCORRELATIONS OF DEPENDENT VARIABLES
O_
Variables Standard deviations 1 2 3 # 5 6 7
1. Annoyance 7.80 1.00 --.54 .33 --.22 .24 --.15 .032.
Trustworthy 4.52 1.00 --.48 .36 --.23 .14 --.073. Behavioral task
2.68 1.00 --.40 .29 --.27 --.05 "4. Gough aggression 11.95 1.00
--.35 .29 --.015. Gough abasement 15.74 1.00 --.03 --.026. Spelling
errors 2.43 1.00 .497. Grammar errors 3.55
t',o
-
204 JOURNAL OF SOCIAL PSYCHOLOGY
and significant interaction between Anger and Expression Type [F
(14/196)-- 1.91; p < .03].
5. Discriminant Function Analysis
Multivariate interactions are difficult to interpret. It seems
that type of
expression affected hostility level differently for angry than
for nonangrysubjects, but the loci of the interaction were not
clear. To gain insight intothe meanings of the multivariate
interaction, a discriminant function analysis(8, 23) of groups
arrayed in a 1 X 6 design was employed. The results of thisanalysis
are shown in Table 2. There was a highly significant difference
between group centroids [Rao's approximation of F
(35,415)--3.49; p <.0001 ]. Only the first root of W-1A (where W
-1 = the inverse of the within
groups SSCP matrix and A--the between groups SSCP matrix) was
staffs-tically significant by conventional standards [X_ (35) =
11.32; p < .0001].
The meaning of the discriminant function can best be understood
by
examining the loading of dependent variables upon the function.
Theseloadings also appear in Table 2. Examination of Table 2
reveals that dif-ferences between groups can be largely accounted
for by differences on thetwo attraction measures and the behavioral
measure. This factor can be
regarded as general hostility or aggressive drive.
6. Group Centroids
Centroids were computed by multiplying scores on each dependent
vari-able by the corresponding raw discriminant function
coefficient and summingthe products across variables. The centroids
for the groups of interest aredisplayed in Table 3. High scores
indicate low hostility, and low scores indi-cate high hostility. As
expected, angry subjects scored higher on hostilitythan nonangry
subjects. The interaction between expression Type and Angercan be
seen in Table 3. In relation to Direct Expression subjects,
Counter-expression subjects were lower on hostility when angry, but
higher on hostilitywhen not angry. Multivariate simple effects
tests revealed the differencesamong Expression Types to be highly
significant among the angry subjects[F (14,196) = 3.27; p <
.0002] but nonsignificant among nonangry subjects
[F(14,196) = 1.43; p < .10]. Further analysis showed both the
AngryDirect IF (7.98) --3.94; p < .001] and the Angry Counter [F
(7.98) --
2.78; p < .01 ] groups to differ from the angry Neutral group
the Neutral
group showing the least hostility.
-
ROBERT M. KAPLAN 20S
TABLE 2MULTIVARIATE COMPARISON OF GROUPS ARRAYED IN A ONE-BY-SIX
DESIGN
Variables Univariate F p_ SDFC a
Annoyance 17.82 .0001 .71Trustworthy 12.23 .0001 --.35Behavioral
task 8.13 .0001 .42Gough aggression 3.45 .01 --.03Gough abasement
2.04 .07 --.05Spelling errors .84 NS .07Grammarerrors .92 NS
.01
Note: Multivariate F (35,415) -- 3.49; p < .0001.a SDFC --
Standardized Discriminant Function Coefficient.
D. DISCUSSION
The results of the experiment were consistent with the
interference hy-pothesis. Among angry subjects, those who engaged
in a neutral expressionbecame less hostile than those who
participated in either the Direct orCounterexpression groups. These
data suggest that activities which remindedangry persons of a
provocateur were less effective in reducing hostility
thanactivities which diverted attention away from the instigator.
Experimentsperformed under other circumstances with different
subject populations havesimilarly shown that distraction may lead
to significant reduction in aggres-sive drive. Mallick and
McCandles (26), for example, observed that workingproblems in
mathematics was more effective in reducing children's
aggressionthan playing aggressively. The interference viewpoint may
provide a viableexplanation for the confusing results of studies on
aggression mediated bytelevised violence. These experiments
frequently show that T.V. violence hasan aggression-activating
effect when subjects are angered, but no effect whensubjects are
not angered (24). It is suggested that violent acts on
televisionremind subjects of their own anger and keep their arousal
from dissipating.Nonviolent shows may serve to distract the angry
subjects and, therefore,may produce lower levels of arousal.
Clearly, the interference notion providesa more parsimonious
explanation of data from a variety of studies than doeseither the
catharsis or dissonance viewpoint. Zillman and Johnson (30)
haverecently come to a similar conclusion on the basis of some
experimental evi-dence. Thus, distraction rather than confrontation
may be a worthwhilemeans of avoiding outbreaks of hostility.
Neither catharsis nor the dissonance theory would have predicted
theresults of the present study. Catharsis theory would have
predicted that,among angry subjects, direct expression would have
produced the greatest
-
206 JOURNAL OF SOCIAL PSYCHOLOGY
TABLE 3GROL-P CENTROIDS ON LARGESTDESCRIIVIINANTFUNCTION
(Lower scores indicate greater hostility)
Arousal
Expressionform Angry Not angry
Direct .21a 3.10 eCounter .58a 2.39 eNeutral 1.59b 2.77 e
Note: Centroids with common subscript do not differ at the .01
level.
anger reduction. Our data show just the opposite. Dissonance
theory could•explain the outcome for the angry Direct expression
subjects, but would havedifficulty explaining why subjects taking a
neutral position became lesshostile than those taking a
counterattitudinal stand. Only the interferenceposition can account
for all of the experimental data.
Contrary to some theoretical positions (3, 5, 10, 20) subjects
who believedtheir essays were going to be shown to the target did
not differ from thosewho believed that the essay would not reach
the target (as reflected by scoreson several measures). It should
be noted, however, that the CommunicationDestiny manipulation was
relatively weak and that a more noticeable manip-ulation may have
been successful. The equivalence of the Angry and NotAngry groups
for the number of spelling and grammatical errors fails toreplicate
the finding reported by Greenberg and Tannenbaum (17). Subjectsin
the present experiment, however, wrote fewer words than those in
theGreenberg and Tannenbaum experiment, and the possibility still
remainsthat the effect would have occurred had our subjects written
more words.
Two limitations of the experiment should be mentioned. First,
the angerarousal manipulation was quite weak. Therefore, the
results may be specificto mild levels of anger. Second, the results
may also be quite specific to thedependent measures which were
employed. Both the author (22) and Kofiecni(25) have pointed out
that in catharsis and aggression research, differentdependent
measures will often show different effects in response to the
samemanipulation. If different measures had been used, the results
might havebeen different.
REFERENCES
1. ANDERSON,N. H. Likableness ratings of 555 personality trait
words. J. Personal. &Soc. Psychol., 1968, 9, 272-279.
2. ARONSON,E. The Social Animal. New York: Freedman, 1972.
3. BARON, R. A. Magnitude of victim's pain cues and level of
prior anger arousal as
-
ROBERTM. KAPLAN 207
determinants of adult aggressive behavior. J. Personal.
&Soc. Psychol., 1971, 17,236-243.
4. BERr:OWIrZ,L. Aggression:A Social-PsychologicalAnalysis.New
York: McGraw-Hill, 1962.
5. . Aggressive cues in aggressive behavior and hostility
catharsis. Psychot.Rev., 1964, 71, 104-122.
6. Buss, A. H. ThePsychology of Aggression. New York: Wiley,
1961.7. COLLr_S, B. E. Attribution theory analysis or forced
compliance. Proc. 77th Ann.
Conven. Amer. Psychol. Assoc., 1969, 4, 309-310.
8. COOLEY, W. W., & LOHNES, P. R. Multivariate Data
Analysis. New York: Wiley,1971.
9. DOr.r_ARD,J., DOOR, L. W., Mrr.r.Eg, N. E., MOWRER, O. H.,
& SEARS, R. R. Frustrationand Aggression. New Haven, Conn.:
Yale Univ. Press, 1939.
10. Doos, A. N. Catharsis and aggression: The effects of hurting
one's enemy. J. Exper.Res. Personal., 1970, 4, 291-296.
il. FESHBACH, S. The drive-reducing function of fantasy
behavior. J. Abn. & Soc.Psychol., 1955, 50, 3-11.
12. . The function of aggression and the regulation of
aggressive drive. Psychol.Rev., 1964, 71, 257-272.
13. FESTIVal.R, L. A Theory of Cognitive Dissonance. Evanston,
Ill.: Row Peterson,1957.
14. FRESD_AN, J. L., CARr.S_ITH, S. M., & SEARS, D. C.
Social Psychology. EnglewoodCliffs, N.J.: Prentice Hall, 1970.
15. GOLD_AN, R. D. Effects of a logical versus a mnemonic
learning strategy on per-formance in two undergraduate psychology
classes. J. Educ. Psychol., 1972, 63,347-352.
16. GOU6H, H. G., & H_._BRU_r, A. B. The Adjective Checklist
Manual. Palo Alto,Calif.: Consulting Psychologists Press, 1965.
17. GRESrCBERG,B. S., & TArCNErCBAU_,P. H. Communicator
performance under cognitivestress. Journalism Quart., 1962, 39,
1969-178.
18. HART_ANN, D. P. The influence of symbolically modeled
instrumental aggressionand pain cues on aggressive behavior. J.
Personal. & Soc. Psychol., 1969, 11,280-288.
19. HOrZANSON,J. E., & BURCESS, M. The effects of three
types of aggression on vascularprocesses. J. Abn. & Soc.
Psychol., 1962, 64, 446-449.
20. HoYT, M. F., HEr_LEY, M. D., & COLLI_S, B. E. Studies in
forced compliance: Theconfluence of choice and consequence of
attitude changes. Unpublished manuscript,University of California,
Los Angeles, 1971 (mimeograph).
21. KAHN, M. The psychology of catharsis. J. Personal. &
Soc. Psychol., 1966, 3, 278-286.
22. KAPr_A_r,R. M. The cathartic value of attitudinal and
counterattitudinal expression.Proc. 81st Ann. Conven. Amer.
Psychol. Assoc., 1973, 277-278.
23. KAPLAN, R. M., & GOLDMAN, R. N. Interracial perception
among Black, White, andMexican-American high school students. J.
Personal. & Soc. Psychol., 1973, 28,383-389.
24. KAPr.A_, R. M., & SI_rGER, R. D. Psychological effects
of televised fantasy: A re-view and methodological critique: J.
Soc. lss., 1976, in press.
25. KONE_NI, V. J. Experimental studies of human aggression: The
cathartic effect.Unpublished Doctoral dissertation, University of
Toronto, Toronto, Canada, 1973.
26. MALLICK, S. K., & MCCANDLES, B. R. A study of catharsis
and aggression. J. Per-sonal. & Soc. Psychol., 1966, 4,
591-596.
27. Rm_rE, R. J., & KApr.AN, R. M. The effect of message
incredulity upon the evalu-ation of the source of communication. J.
Soc. Psychol., 1972, 88, 255-266.
-
208 JOURNAL OF SOCIAL PSYCHOLOGY
28. ROSENBA_M, M. E., & DECHARMES, R. Direct and vicarious
reductions of hostility.I. Abn. & Soc. Psychol., 1960, 60,
105-1t!.
29. T_rmAUT, J. W., & Co_Es, J. The role of communication in
the reduction of inter-personal hostility. J. Abn. & Soc.
Psychol., 1952, 47, 770-777.
30. Z_LMAN, D., & Jo_r_SON, R. C. Motivated aggressiveness
perpetuated by exposureto aggressive films and reduced by exposure
to nonaggressive films. J. gxper. Res.Personal., 1973, 7,
261-276.
Psychological Clinic
San Diego State UniversitySan Diego, California 92182