作者: ARONSON / 6791次阅读 时间: 2012年10月17日
来源: 标签: Aronson



Stanford University


U. S. Army Leadership Human Research Unit, IhimRRO

IT is a frequent observation that persons who go through a great deal of trouble or pain to attain something tend to value it more highly than persons who attain the same thing with a minimum of effort. For example, one would expect persons who travel a great distance to see a motion picture to be more impressed with it than those who see the same picture at a neighborhood theater. By the same token, individuals who go through a severe initiation to gain admission to a club or organization should tend to think more highly of that organization than those who do not go through the severe initiation to gain admission.

Two questions are relevant here: 1. Is this "common observation" valid, that is, does it hold true when tested under cotrolled conditions? 2. If the observation is valid, how can it be accounted for? The relationship might be simply a result of differences in initial motivation. To take the case of initiations, persons who initially have a strong desire to join a particular club should be more willing to undergo unpleasantness to gain admission to it than persons who are low in initial interest. Therefore, a club that requires a severe initiation for admission should be joined only by those people with a strong desire to become members. On the other hand, a club that does not require a severe initiation should be joined by some individuals who like it very much, and by others who are relatively uninterested. Because of this self-selection, one would expect persons who are members of clubs with severe initiations to think more highly of their club, on the average, than members of clubs without severe initiations.

But is there something in the initiation itself that might account for this relationship? Is severity of initiation positively related to group preference when motivation for admission is held constant? Such a relationship is strongly implied by Festinger's (1957) theory of cognitive dissonance. The theory of cognitive dissonance predicts this relationship in the following manner. No matter how attractive a group is to a person it is rarely completely positive, i.e., usually there are some aspects of the group that the individual docs not like. If he has undergone an unpleasant initiation to gain admission to the group, his cognition that he has gone through an unpleasant experience for the sake of membership is dissonant with his cognition that there are things about the group that he does not like. He can reduce this dissonance in two ways. He can convince himself that the initiation was not very unpleasant, or he can exaggerate the positive characteristics of the group and minimize its negative aspects. With increasing severity of initiation it becomes more and more difficult to believe that the initiation was not very bad. Thus, a person who has gone through a painful initiation to become a member of a group should tend to reduce his dissonance by over estimating the attractiveness of the group. The specific hypothesis tested in the present study is that individuals who undergo an unpleasant initiation to become members of a group increase their liking for the group; that is, they find the group more attractive than do persons who become members without going through a severe initiation.


In designing the experiment it was necessary to have people join groups that were similar in every respect except for the severity of the initiation required for admission—and then to measure each individual's evaluation of the group. It was also necessary to randomize the initial motivation of subjects (5s) to gain admission to the various groups in order to eliminate systematic effects of differences in motivation. These requirements were met in the following manner: Volunteers were obtained to participate in group discussions. They were assigned randomly to one of three experimental conditions: A Severe initiation condition, a Mild initiation condition, and a Control condition. In the Severe condition, 5s were required to read some embarrassing material before joining the group; in the Mild condition the material they read in order to join the group was not very embarrassing; in the Control condition, Ss were not required to read any material before becoming group members. Each 5 listened to the same tape recording which was ostensibly an ongoing discussion by the members of the group that he had just joined. 5s then evaluated the discussion. The 5s were 63 college women. Thirty-three of them volunteered to participate in a scries of group discussions on the psychology of sex. The remaining 30, tested at a somewhat later date, were "captive volunteers" from a psychology course who elected to participate in the group discussions on the psychology of sex in preference to several other experiments. Since the results obtained from these two samples were very similar, they were combined in the analysis presented here.

Each 5 was individually scheduled to "meet with a group." When she arrived at the experimental room, she was told by the experimenter (E) that he was conducting several group discussions on the psychology of sex. E informed her that she was joining a group that had been meeting for several weeks and that she was taking the place of a girl who had to leave the group because of scheduling difficulties. E stated that the discussion had just begun and that she would join the other members of the group after he had explained the nature of the experiment to her. The purpose of the foregoing instructions was to confront 5 with an ongoing group and thus make plausible the recorded discussion to which she was to be exposed.

E then "explained" the purpose of the experiment. He said that he was interested in investigating the "dynamics of the group discussion process." Sex was chosen as the topic for the groups to discuss in order to provide interesting subject matter so that volunteers for the discussion groups could be obtained without much difficulty. E continued as follows:

But the fact that the discussions are concerned with sex has one major drawback. Although most people are interested in sex, they tend to be a little shy when it comes to discussing it. This is very bad from the point of view of the experiment; if one or two people in a group do not participate as much as they usually do in group discussions because they are embarrassed about sex, the picture we get of the group discussion process is distorted. Therefore, it is extremely important to arrange things so that the members of the discussion group can talk as freely and frankly as possible. We found that the major inhibiting factor in the discussions was the presence of the other people in the room. Somehow, it's easier to talk about embarrassing things if other people aren't staring at you. To get around this, we hit upon an idea which has proved very successful. Each member of the group is placed in a separate room, and the participants communicate through an intercom system using headphones and a microphone. In this way, we've helped people relax, and have succeeded in bringing about an increase in individual participation.

The foregoing explanation set the stage for the tape recording, which could now be presented to the 5 as a live discussion conducted by three people in separate rooms.

E then mentioned that, in spite of this precaution, occasionally some persons were still too embarrassed to engage in the discussions and had to be asked to withdraw from the discussion group. S was asked if she thought she could discuss sex freely. She invariably answered affirmatively. In the Control condition 5 was told, at this point, that she would be a member of the group.

In the other two conditions, E went on to say that it was difficult for him to ask people to leave the group once they had become members. Therefore, he had recently decided to screen new people before admitting them to the discussion groups. The screening device was described as an. "embarrassment test" which consists of reading aloud some sexually oriented material in the presence of E. S was told that E would make a clinical judgment of her degree of embarrassment, based upon hesitation, blushing, etc. and would determine whether or not she would be capable of participating in the discussion group. He stressed that she was not obligated to take this test, hut that she could not become a member unless she did. Only one S declined to take the test. She was excluded from the experiment. It was also emphasized, at this point, that the "embarrassment test" was a recent innovation and that the other members had joined the group before it was required for admission. These instructions were included in order to counteract any tendency to identify more strongly with the group as a result of feelings of having shared a common unpleasant experience. Such a process could conceivably bring about a greater preference for the discussion group on the part of 5s in the Severe condition, introducing ambiguity in the interpretation of the results.

In the Severe condition, the "embarrassment test" consisted of having 5s read aloud, from 3 X 5 cards, 12 obscene words, e.g., fuck, cock, and screw. 5s also read aloud two vivid descriptions of sexual activity from contemporary novels. In the Mild condition, 5s read aloud five words that were related to sex but not obscene, e.g., prostitute, virgin, and petting. In both the Severe and the Mild conditions, after each 5 finished reading the material, she was told that she had performed satisfactorily and was, therefore, a member of the group and could join the meeting that was now in progress.

It was of the utmost importance to prevent the 5 from attempting to participate in the discussion, for if she did, she would soon find that no one was responding to her statements and she would probably infer that the discussion was recorded. To insure their silence, all 5s were told that, in preparation for each meeting, the group reads an assignment which serves as the focal point of the discussion; for this meeting, the group read parts of the book, Sexual Behavior in Animals. After the 5 had indicated that she had never read this book, E told her that she would be at a disadvantage and vvouldj consequently, not be able to participate as fully in this discussion as she would had she done the reading. He continued, "Because the presence of a participant who isn't contributing optimally would result in an inaccurate picture of the dynamics of the group discussion process, it would be best if you wouldn't participate at all today, so that we may get an undistorted picture of the dynamics of the other three members of this group. Meanwhile, you can simply listen to the discussion, and get an idea of how the group operates. For the next meeting, you can do the reading and join in the discussion." 5s were invariably more than willing to comply with this suggestion. The above instructions not only prevented S from attempting to participate in the discussion but also served to orient her toward the actual content of discussion.

Under the guise of connecting the S's headphones and microphone, E went into the next room and turned on the tape recorder. He then returned to the experimental room, put on the headphones, picked up the microphone, and pretended to break into the discussion which supposedly was in progress. After holding a brief conversation with the "members of the group," he introduced the 5 to them. Then he handed the headphones to her. The tape was timed so that at the precise moment that S donned her headphones, the "group members" introduced themselves and then continued their discussion.

The use of a tape recording presented all 5s with an identical group experience. The recording was a discussion by three female undergraduates. It was deliberately designed to be as dull and banal as possible in order to maximize the dissonance of the 5s in the Severe condition. The participants spoke dryly and haltingly on secondary sex behavior in the lower animals, "inadvertently" contradicted themselves and one another, mumbled several non sequiiurs, started sentences that they never finished, hemmed, hawed, and in general conducted one of the most worthless and uninteresting discussions imaginable.

At the conclusion of the recording, E returned and explained that after each meeting every member of the group fills out a questionnaire expressing her reactions to the discussion. The questionnaire asked the 5 to rate the discussion and the group members of 14 different evaluative scales, e.g., dull-interesting, intelligent- unintelligent, by circling a number from 0 to 15. After completing the questionnaire, 5 made three additional ratings, orally, in response to questions from E. Nine of the scales concerned the 5's reactions to the discussion, while the other eight concerned her reactions to the participants.

At the close of the experiment, E engaged each 5 in conversation to determine whether or not she was suspicious of the procedure. Only one S entertained definite suspicions; her results were discarded. Finally, the true nature of the experiment was explained in detail. None of the 5s expressed any resentment or annoyance at having been misled. In fact, the majority were intrigued by the experiment and several returned at the end of the academic quarter to ascertain the results.


The sum of the ratings for the 17 different scales provides an index of each 5's liking for the discussion group. The means and SDs, for the three experimental conditions for this measure are presented in Table 1. Means and SDs are also presented in Table 1 separately for the eight scales which tapped the 5s' attitudes toward the discussion and the seven scales which tapped their attitudes toward the participants. The significance of the differences between the means for the different conditions were determined by t tests. The t values and significance levels are presented in Table 2.


Rating Scales

Experimental Conditions












Rating Scales

Differences Between Conditions

Mild-Severe Control-
Discussion[9] t <= 3.66
P < .001*
t = 2.62
P < .02
t = .29
Participants[8] t = 2.03
P < .OS
t = 1.97
P < .10
t = .15
Total[17] t = 3.32
P < .01
t = 2.33
P < .05
t = .49

Examination of Table 1 shows that 5s in the Severe condition rated both the discussion and the participants higher than did those in the Control and Mild conditions. The over-all difference between the ratings by 5s in the Severe condition and 5s in the Control condition reaches the .01 % level of significance. The over-all difference between the ratings by 5s in the Severe initiation condition and 5s in the Mild initiation condition reaches the .05 level. These differences cannot be explained by differences in initial motivation to become members of the group, since 5s (with varying degrees of motivation) were randomly assigned to the three experimental conditions. The differences in liking for the group must be considered a consequence of the unpleasant experience. The results clearly substantiate the hypothesis: persons who undergo a severe initiation to attain membership in a group increase their liking for the group. This hypothesis follows directly from Festingcr's theory of cognitive dissonance. According to the theory, 5s in the Severe initiation condition held the cognition that they had undergone a painful experience to become members of the discussion group. Then they listened to a dull, banal discussion. Negative cognitions about the discussion which they formed from listening to it were dissonant with the cognition that they had undergone a painful experience to gain membership in this group. The presence of dissonance leads to pressures to reduce it. 5s in this condition could reduce their dissonance either by denying the severity of the initiation or by distorting their cognitions concerning the group discussion in a positive direction. The initiation of the 5s in the Severe condition was apparently too painful for them to deny — hence, they reduced their dissonance by overestimating the attractiveness of the group.

There was no appreciable difference between the ratings made by 5s in the Control condition and those made by 5s in the Mild condition. It would seem that the Mild condition was so devoid of unpleasantness as to constitute little investment in the group. Hence, little dissonance was created. If any dissonance did occur in this situation it would be more realistic for the 5 to reduce it by minimizing the pain of the initiation, than by distorting her cognitions concerning the discussion. Thus, it is not an initiation per se that leads to increase in liking for a group. The initiation must be severe enough to constitute a genuine investment and to render it difficult to reduce dissonance by playing down the extent of the pain involved.

An examination of Table 1 shows that the rating scales concerning the discussion show greater differences between the conditions than the scales dealing with the evaluations of the participants in the discussion. There are at least two possible explanations for this result: (a) It may be easier for people to express negative criticism about an impersonal discussion than about the people involved. Thus, 5s in the Control and Mild conditions may have inflated their ratings of the participants to avoid making negative statements about fellow college students, (b) It is possible that 5s in the Severe condition had less need to distort their perception of the participants than of the discussion itself. The dissonance of the 5s in the Severe condition resulted from the actual discussion: they experienced dissonance between going through an unpleasant experience and taking part in worthless uninteresting discussions. The most direct way for them to reduce this dissonance would be to change their perceptions of the discussion in a positive direction. The participants in the discussion were peripheral to the cause of dissonance. If 5s in the Severe condition had less need to distort their perceptions of the participants than their perception of the discussion, their evaluations of the participants could be expected to be closer to the evaluations of the participants made by 5s in the Control and Mild conditions.


An experiment was conducted to test the hypothesis that persons who undergo an unpleasant initiation to become members of a group increase their liking for the group; that is, they find the group more attractive than do persons who become members without going through a severe initiation. This hypothesis was derived from Festinger's theory of cognitive dissonance.

College women who volunteered to participate in discussion groups were randomly assigned to one of three experimental conditions: A Severe initiation condition, a Mild initiation condition, and a Control condition. In the Severe condition, subjects were required to read some embarrassing material before joining the group; in the Mild condition the material they read in order to join the group SEVERITY OF INITIATION AND LIKING FOR A GROUP 181 was not very embarrassing; in the Control condition, subjects were not required to read any material before becoming group members. Each subject listened to a recording that appeared to be an ongoing discussion being conducted by the group which she had just joined. Afterwards, subjects filled out a questionnaire evaluating the discussion and the participants. The results clearly verified the hypothesis. Subjects who underwent a severe initiation perceived the group as being significantly more attractive than did those who underwent a mild initiation or no initiation. There was no appreciable difference between ratings by subjects who underwent a Mild initiation and those by subjects who underwent no initiation.


1 This research was partially supported by a grant from the National Science Foundation, administered by Leon Festingcr. The authors are grateful to Leon Festinger for his help and encouragement during the planning and execution of the study.


FESTINGER, L. A theory of cognitive dissonance. Evanston: Row, Peterson, 1957. Received June 9, 1958.

TAG: Aronson
«淺談「拼圖式」(Jigsaw)合作學習 78 埃里奥特·阿伦森 | Elliot Aronson
《78 埃里奥特·阿伦森 | Elliot Aronson》
《錯不在我?》第一章 認知失調:自我辯護的動力來源»