The attractiveness of an eliminated choice alternative

JOURNAL OF EXPERIMENTAL SOCIAL PSYCHOLOGY 2, 301-313 (1966)

The Attractiveness of an Eliminated Choice Alternative’

JACK W. BREHM, LLOYD K. STIRES, JOHN SENSENIG, AND JANET. SHABAN'

Duke University

A recent theory by Brehm implies that the elimination of a choice alternative will tend to increase the attractiveness of that alternative to the person who is about to make the choice. Two experiments, in which college students rated the attractiveness of four recordings of music in order to be able to choose one for themselves, found that the third most attractive record increased in attractiveness when it became unavailable as one of the choice alternatives. In order to rule out alternative explanations of this finding, the second experiment included a condition in which some subjects were given no prior freedom to choose which record they would receive. The results of this second experiment indicated that the tendency to see the eliminated record as more attractive occurred only when subjects had prior freedom to choose which record they would take. Those subjects who had no prior freedom to choose tended to see the eliminated record as having decreased in attractiveness.

It is apparent that an individual can better satisfy himself, gain benefits, avoid harm and dissatisfaction if he has freedom to do what he wants and do it when and how he wants. It is therefore plausible to assume that when a person has such freedom, he will be motivationally aroused by any threat to or reduction of it and he will consequently attempt to re-establish the threatened or eliminated freedom.

A recent theory by Brehm (in press) is an attempt to specify some conditions under which behavioral freedom is threatened or eliminated, some of the factors which control the magnitude of the resultant motivational arousal, and some-of the subjective and behavioral consequences of the motivation. According to this theory, an elimination or threat of elimination of the fredom to engage in a given behavior creates “psychological reactance” in the individual, that is, a motivational state directed toward restoration of the eliminated or threatened freedom.

Where the freedom which is threatened or eliminated is “having” a

‘This research was supported in part by National Science Foundation grant GS 364, and in part by National Institute of Mental Health grant MH-1122801, both under the direction of the senior author.

‘Now at the University of Texas. 301

302 BREHM, STIRES, SENSENIG, AND SHABAN

specific object such as a phonograph record, the individual’s desire to have the object should increase and, as a corollary, he should judge the object as having increased in attractiveness.

While the theory specifies several factors that affect the magnitude of reactance, only two of these are relevant to the present problem. Given that a freedom has been eliminated for an individual, the magnitude of reactance will be a direct function of how important that freedom is to him, and it will also be a direct function of the relative importance of that freedom compared to the importances of other freedoms of the moment. Thus, the greater the absolute and/or relative importance of a threatened or eliminated freedom, the more will the individual be moti- vated to re-establish that freedom.

When a situation allows a person to choose one of two or more alternatives it is implicit that the person has the freedom to have any one alternative. The importance of this freedom should be a direct function of the attractiveness of the alternative to the individual, given that the potential needs which may be satisfied by this alternative could not be satisfied by other available alternatives. It should be true, then, that if a person has not made up his mind about which alternative to choose, the elimination of an alternative will create reactance in him and the amount of reactance will vary directly with the absolute atkactive- ness of the eliminated alternative and with its attractiveness in com- parison with the other alternatives.

While the major hypothetical effect of reactance would be for the individual to attempt to restore his freedom to have the alternative, this response is precluded when the elimination is unequivocal. Never- theless, the motivation to have the eliminated alternative would theo- retically increase, and this increase should be detectable as an increase in the rated attractiveness of the alternative.

Two experiments are to be reported. The first was designed to test two propositions: (1) an eliminated choice alternative tends to become more attractive; and (2) the amount of increased attractiveness of the eliminated alternative is a direct function of the absolute attractiveness of the alternatives. The second experiment was designed to show that freedom to have the alternative is a necessary precondition for elimination of it to result in increased attractiveness.

EXPERIMENT I

Method

The general plan was to have people listen to and rate the attractiveness of 4 different records in order to obtain one as a gift, and then learn, prior to listening and rating a second time, that one of the records would be unavailable for their

ATTRACTIVENESS OF AN ELIMINATED ALTERNATIVE 303

choice of a gift. Changes in ratings of the records would reveal any tendency for the eliminated alternative to become more attractive.

Procedure

Students from a summer session introductory psychology course at Duke Uni- versity volunteered to take part in a two-session “Market Analysis of Phono- graph Records,” for which they were to receive experimental credit. Subjects were run individually, and when one arrived for his scheduled session he was seated at a table opposite a tape recorder. The experimenter introduced himself as a graduate student in psychology, and explained that, by virtue of its experimental facilities, the psychology department occasionally receives contracts to perform market research surveys of college populations for commercial organizations. This was represented as being helpful as a means of support for graduate students. The experimenter then explained that he was currently employed by a company which distributes phonograph records, and that the purpose of the study was to obtain judgments about a group of recorded selections which might be of interest to college students. The recordings were represented as selections which had previously been released on long-playing albums and which were now being considered for release as 45 rpm single records.

The subject was told that he was to listen to a tape containing four such selections and that his task was to indicate how well he liked the records. He was reminded that there would be another session on the following day, at which he would be asked to perform “a similar task,” the exact nature of which was not described.

Manipulation of importance. To create two degrees of importance of freedom, some subjects were led to believe that they would be able to choose one of four 45 rpm records (worth about $1 each), and other subjects were told they would be able to choose one of four LP albums (worth about $3 each). The procedure for the High Imp (Importance) condition will be described first.

After the introductory remarks described above, the subject was informed that the company wished each participant to have a complimentary record as a gift for his participation in the study. The experimenter explained that he was hopeful that these complimentary records would arrive on the following morning, and that, if so, the subject would receive a long-playing album at the conclusion of the second session. These albums were those which contained the selections which the subject would hear, and he would be free to choose whichever album he pre- ferred, making his judgment on the basis of the selections.

The subject then filled out a brief questionnaire dealing with such matters as his phonograph- and radio-listening habits, his record-purchasing habits, and his preferences among various types of music.

Next, he was given four rating scales and told that he was to listen to each of the four records and rate them with regard to “how well he liked them.” The scale consisted of a horizontal line of 101 dots, delineated into groups of ten by vertical slashes, and labeled at the left and right ends as “0” and “100,” respec- tively, with “50” at the center. It was explained that a rating at the left end of the scale was “unfavorable,” and a rating at the right, “favorable.” It was im- pressed on subjects that they were to express their own personal opinions about the records rather than how likely the records were to become best-sellers. Since subjects were supposed to give their first impressions of each of the selections, the tape was stopped after each piece to allow time for the ratings to be made. After

304 BFZEHM, STIBES, SENSENIG, AND SHABAN

completing the ratings, the subjects were scheduled for a second session and then excused.

The LOW Imp (Importance) condition differed from the High Imp in that subjects were told the selections were sides of 45 rpm records which were soon to be released to the general public and that they were to receive, as a gift, their choice of one of the 45 rpm selections they were to hear.

For both conditions at the beginning of the second session, the subject was told that he was to listen to the same four selections that he had heard on the preceding day, and was to rerate them. This was represented as an attempt to discover how listening to a record for the second time might affect one’s evaluation of it. The subject was told that his first impressions might change and that, if so, the experimenter was interested in finding this out. It was also noted that for the second rating, the subject would hear all four selections before making any evaluations, a procedure which would give him an opportunity to compare the selections. He was given pad and pencil to take notes if he desired.

The elimination of a choice alternative. The experimenter then explained that the complimentary records which he had promised the subject had indeed arrived that morning, but that the shipment, for some unknown reason, failed to include copies of one of the records. This “unavailable” record was clearly identified as that which the subject had rated third highest on the previous day. The third rated record was eliminated, rather than the first, to avoid frustrating the subject. Care was taken to avoid the implication that this record was unavailable because of greater popularity. I f questioned, the experimenter stated that the record merely had not been sent and that he had not yet had a chance to make any inquiry as to the reason.

Subjects then listened to the four selections and rated them again.3 Finally, the experimenter conducted a brief interview which included some

general questions about the records themselves and the possible effects of hearing them again, some questions regarding the elimination and whether or not this might have affected ratings, and some questions designed to ascertain what, if any, suspicions there might be with regard to the experiment. The purpose of the experiment was then revealed and further comments were entertained.

Design

In addition to the High and Low Importance conditions described above, control conditions were run with both LP and 45 rpm gift records in which the above described procedure was used, the only difference being that there was no elimination of choice alternatives at the second session. The control conditions provide an estimate of the change in attractiveness of the initial rank 3 record when it is not restricted.

Musical Selections

In line with the rationale of the study and also in order to avoid highly crystallized opinions about the four pieces of music, the selections, all of the folk music variety, were new and not well known or popular. While it would have been

*Subjects were also given an opportunity to listen to one record again before making the ratings in order to see which record would be chosen. This measure

proved insensitive, however, since only 17% of all subjects picked the third-rated record, most of them (72%) having picked the first-rated record.


desirable to have the four pieces about equally attractive, our search failed to come up with such a combination and the fmal set ranged from moderate to 10~ in attractiveness.

Subjects

Sixty-two students from a summer introductory psychology class at Duke Uni- versity served as subjects, 54 of them being females and most of these from the school of nursing. Subjects were assigned to conditions in the order in which they volunteered, the conditions being run in a counterbalanced design. Twice as many Elimination as Control sessions were run since it was expected that High and LOW Importance Control conditions would not differ and could be combined into a single Control condition. Of the 62 subjects run, 15 were eliminated from the analysis because they indicated in a post-experimental interview that they were suspicious of the procedure. Much of this suspicion was apparently due to the fact that some subjects had recently participated in another study which bore basic similarities to the present one.

Results

The Importance manipulation failed to have any effect on attractiveness ratings in either the Elimination or the Control conditions, a point which will be discussed later. For the following presentation the High and Low Importance conditions have been combined into single Control and Elimination conditions.

Table 1 presents the initial attraction scores in each condition of the rank 1, 2, 3, and 4 records. These scores vary rather widely from the rank 1 to the rank 4 record, despite our intent to select approximately

TABLE 1 INITIAL ATTRACTIVENESS RATINGS OF THE 4 RECORDS

Initial rank of record

1 2 3 4

Control (N = 14) 84.1” 73.6 50.9 36.3 Elimination (N = 33) 86.7 77.6 59.7 38.5

a 100 = high attractiveness.

equally attractive ones, and there is an unreliable tendency for the rank 3 record to be rated higher in the Elimination conditions than in the Controls. This latter difference means that if there is any tendency for the rank 3 record to be rated higher the second time than the first due to unreliability, the tendency should be greater in the Control condition than in the Elimination condition, an effect which would work against the hypothesis.

It was expected that elimination of the rank 3 record would arouse


reactance and a consequent increase in the attractiveness of that record. The mean changes in attractiveness ratings of the rank 3 record and of the other three records combined are shown in Table 2. These changes show that the mean increase in rating of the rank 3 record in the Elimination condition (6.33) is greater than the comparable figure in the Control condition (.32). That this difference is not due to a tendency to

TABLE 2 CHANGE IN ATTRACTIVENESS RATINGS

Rank 3 record Other records

Control Elimination

.32O 3.04 6.33 1.08

a Positive change indicates increased attractiveness.

raise the ratings of all records may be seen from the relatively small increase for the other three records. However, the difference between Elimination and Control conditions in rating changes of the rank 3 record yields a t of only 1.38, a value well short of statistical reliability. In large part this unreliable t value may be attributed to the relatively small number of subjects in the Control condition, since if we consider

TABLE 3 FREQUENCY OF INCREASED, DECREASED, AND UNCHANGED ATTRACTIVENESS RATINGS

OF THE RANK 3 RECORD

Attraction change

Control Elimination

Increase No change Decrease

6 1 7 23 3 7

the Elimination condition alone we find that the rise in attractiveness is significantly different from no change at beyond the 1% level (t = 2.89, df=32).4

A tabulation of persons showing increased attractiveness ratings, no change, or decreased ratings of the rank 3 record yields similar results, as may be seen in Table 3. While the Control subjects split about evenly between raising and lowering their ratings, the Elimination subjects showed a clear tendency to increase rather than to decrease their ratings. However, a Chi square for the difference between Elimination and Con- trol subjects, omitting those who did not change, is significant at only

‘All reported significance levels are two-tailed.


about the 11% level. And again considering only subjects in the Elimina- tion condition, their tendency to raise rather than lower their ratings is significant at beyond the lo/O level (x2 = 7.50).

Considering both mean change and frequency data, then, we find that subjects for whom there was no elimination of the rank 3 record showed no dominant tendency either to raise or lower their rating of that record. On the other hand, subjects who were told that the rank 3 record was unavailable showed a clearly significant tendency to raise their rating of that record. Nevertheless, apparently because of the relatively small number of subjects in the Control condition, the difference between Control and Elimination conditions fails to reach an acceptable level of statistical reliability.

As noted earlier, there was no evidence in support of the expected difference between High and Low Importance conditions. In the absence of a direct check on manipulated importance, we can only conjecture that the relatively low attractiveness of the rank 3 selection, coupled with its unfamiliarity, failed to convince subjects that they were getting a “good deal” when they had the opportunity to obtain the whole LP album. At best the subject could not be sure he was getting something he would like to hear on the whole album, and at worst he could predict that he would not like to hear the rest of what was on the album, since he was presumably hearing one of the better selections-one good enough for possible issuance as a 45 rpm record.

The obtained results are consistent with reactance theory, but they permit at least three further explanations. First, it could be that singling out the rank 3 record, that is, drawing attention to it, results in its being judged more carefully. Since it is possible that a more careful listening and consideration by the subject results in a higher attractiveness rating, such singling out may account for the experimental results. A parallel explanation would suppose that there is decreased attention given to the record because it has become unavailable and that this decreased attention somehow results in a higher rating. A third possible explanation is the proposition that people want what they cannot have, at least under the present experimental conditions. What it suggests is that when a person loses the possibility of having an alternative, that alternative becomes more attractive whether or not the loss also re- duces his freedom.

As it happens, it is possible to design a single control condition capable of ruling out all three alternative views. In this condition the subject is led to believe that he will receive one of the four records and that which he receives will be determined by chance rather than by his own choice. The rank 3 record is then eliminated in the same manner as be-


fore. Thus, attention is called to the record, the record is made unavailable, but the individual’s freedom is not reduced. If, in a condition of this design, there is no increase in the attractiveness of the restricted alternative, we may rule out these three possible explanations for the previously found increase and thereby more confidently attribute that effect to reactance.

EXPERIMENT II

The purposes of the second experiment were to (1) replicate the rise in the attractiveness of the eliminated alternative and (2) show that the rise occurs only when the elimination of the alternative is also an elimination of freedom. An experiment similar to the first one was therefore set up in which there were three conditions: Choice with no subsequent Elimination of one alternative; Choice with subsequent Elimina- tion of one alternative; and No Choice with subsequent Elimination of one alternative.

Method

The procedure was the same as before except for the following differences: (1) the experimenter was a female rather than a male; (2) the experiment was conducted during the regular school year and the subjects were all males; (3) approximately equally attractive and relatively well-known recordings were selected through pilot work; (4) questionnaires designed to obtain further in- formation about the relevant psychological processes were introduced at the ends of the first and second experimental sessions; and (5) subjects were given the explicit option of returning or not for the second session, with the understanding that they would receive a record but no additional credit if they did return.

As before, the first session involved giving the rationale of market research on records in order to ascertain if certain selections from LP albums should be issued on 45 rpm records. It was explained to the subject that he would rate some pieces of music, for which he would obtain experimental credit, and that he would also have the option of returning for a second similar session for which no experimental credit could be given but for which the sponsoring record company would give one of the four LP albums from which the four selections had been made. Experimental credit for the second session was dropped in order to increase the value of the choice alternatives to subjects.

At this point the Choice Manipulation was introduced by use of one of the following two sets of instructions: Choice: “That is, you may choose one of the four albums from which you hear selections today.” No Choice: “We will have a limited number of records so we’re going to give them out randomly. You would receive one of the four albums from which you hear selections today.” The subject was told he would be asked at the end of t.he session whether or not he would like to return for the second session. The listening and rating procedure was then explained, the short questionnaire on listening habits and musical preferences was administered, the record ratings were obtained, and a questionnaire, to be described below, was administered. Finally, the subject was asked if he would


like to return for the second session, and, if so, he was scheduled to return within a week.

The second session began with a description of the listening and rating procedure. The experimental conditions were then established with one of the following sets of instructions.

Choice-No Elimination. “As I told you, you will be able to choose one of the albums. You may do this at the end of the session.”

Choice-Elimination. “As I told you, you will be able to choose one of the albums. When the company sent us the albums, however, it neglected to include the album, that is, the one from which selection __ was taken. I am sorry about this, since we had, of course, intended for you to be able to choose any of the four. We decided to continue with the study anyway since time is getting short. Listen to the selection and rate it as carefully as the others.”

No Choice-Elimination: “As I told you, you will be able to have one of the albums. I told you we had a limited number and that we’re giving them out randomly. You will draw a numbered slip of paper from a box; each number corresponds to a selection number. So if your paper is numbered one, for example, you will receive the first album, the one with selection number one on it. When the record company sent us the albums, however, it neglected to include the - album, that is, the one from w.hich selection - was taken. I am sorry about this, since we had, of course, intended for you to be able to have any of the four. We decided to continue with the study anyway since time is getting short. Listen to the ___ selection and rate it ss carefully as the others.” In both Elimination conditions, the missing record was always the one initially rated third most attractive by the subject.

The listening and rating then took place, followed by the administration of two questionnaires. Finally, the subject was interviewed briefly about his impressions of the study; the true purpose of the study and deceptions were explained; and the subject was given a choice between having additional experimental credit and receiving a gift certificate for an LP album.

Questionnaires

First session. This questionnaire, administered after the initial ratings were completed, mainly contained items to support the rationale of the study, such as “Which of the four records would be most likely to become a best-seller’?” How- ever, one item, to be described later, was used to eliminate certain subjects from the analysis.

Second session. The first questionnaire administered in the second session again included some questions to support the rationale, but it also included some items relevant to the hypotheses. These asked subjects to indicate which records they liked better and which less after hearing them a second time and to rank the records in order of preference.

The second and last questionnaire of the second session was designed to explore suspicions and also awareness of reactions to the elimination of an alternative. Subjects were asked how suspicious they were before and during the experiment, whether or not they were disappointed or annoyed to learn that one record was unavailable and why, whether or not the attractiveness of the unavailable record changed and, if so, how and why. The precise form of these questions will be given with the report of relevant results.


Subjects and Design

Subjects were male volunteers from an Introductory Psychology course at Duke University. They were randomly assigned to the three conditions, Choice-No Elimination, Choice-Elimination, and No Choice-Elimination, except that near the end of the experiment an attempt was made to equate conditions’ Ns.

Of the ‘72 who took part in the first session, 16 chose not to return for the second. In addition, four subjects were discarded because they indicated strong suspicion about the procedure, one was discarded because his initial rating of three records was 100, and one was discarded because his initial ratings of the third and fourth records were markedly low. It WM also decided to eliminate eight subjects who indicated on the first session post-experimental questionnaire that the third- rated record was the one they would be least likely to buy. For these subjects the loss of the third record would have had relatively low importance. It should be noted, however, that the results for these subjects are similar to those for the main sample.

Results

The initial ratings of the records by rank, present’ed in Table 4, show that the records are indeed more nearly equal in attractiveness than were those of the previous experiment. Here the range is only about 30 points whereas before the range was more than 45. Additionally, the differences in initial scores of the rank 3 record are not great, nor is any between-condition difference anywhere near significance.

TABLE 4 INITIAL ATTRACTIVENESS RATINGS OF THE 4 RECORDS

Initial rank of record

1 2 3 4

Choice--No Elimination (N = 14) 85.8” 76.6 66.3 53.8 Choice-Elimination (N = 13) 86.8 78.2 69.2 57.4 No Choice-Elimination (N = 15) 90.9 81.9 73.0 64.3

0 100 = high attractiveness.

Table 5 presents the rating changes of the rank 3 record and the rlverage changes for the rank 1, 2, and 4 records combined. The rating changes of the rank 3 record for the Choice-No Elimination and for the Choice-Elimination conditions are quite similar to the changes obtained in Experiment I. The difference between the two conditions, however, is not, statistically reliable, an outcome which is not surprising since the corresponding difference in Experiment I, with a much larger N, also failed to be reliable. But, these two conditions replicate the Control and Elimination conditions of the first experiment, making it possible to combine the data from the second experiment with those of the first in


order to obtain a better estimate of the statistical reliability of the difference between the responses of subjects whose choice of the rank 3 record was eliminated and the responses of subjects whose choice of the rank 3 record was not eliminated. The difference between the combined increase in attractiveness of the Choice-Elimination conditions (5.74) ancl the Choice-No Elimination conditions (.20) is significant at the

TABLE 5 CHANGE IN ATTRACTIVENESS RATINGS

Rank 3 record Other records

Choice--No Elimination .07a Choice-Elimination 4.23 No Choice-Elimination -5.40

5 Positive change indicates increased attractiveness.

-.64 -.46 -.03

9% level (t = 1.74, df = 72). While this difference is only marginally reliable, the combined data indicate quite clearly that there is no tendency for a change in attractiveness ratings in the Control subjects, while for Choice-Elimination subjects there is a strong tendency for attractiveness ratings to increase relative to zero (t = 3.29, df = 45, p < .005).

TABLE 6 FREQUENCY OF INCREASED, DECREASED, AXD UNCHANGED ATTRACTIVENESS

RATINGS OF THE RANK 3 RECORD FOR THE Two EXPERIMENTS COMBIXED

Attraction change

Increase No change Decrease

Choice-No Elimination 12 3 13 Choice-Elimination 31 5 10

A tabulation for the combined experiments of those who increased, did not change, or decreased their rating of the rank 3 record is pre- sented in Table 6. The obtained difference between the Choice-Elimina- tion and Choice-No Elimination conditions, with those who did not change omitted, is significant at the 5% level (x2 = 4.07). These data too show no predominant tendency for No Elimination subjects to either raise or lower their ratings, while those in the Elimination condition qhow a strong tendency to raise their ratings (x2 = 9.76, p < .005).

In summary, the two experiments together indicate fairly clearly that when the rank 3 record is eliminated as one of the choice alternatives,


it tends to become more attractive. We may therefore turn to the second concern of the present experiment, namely, the question of whether or not prior freedom to choose the eliminated alternative is a necessary condition for the increase in attractiveness.

The mean change in the No Choice-Elimination condition was -5.40. This change is significantly different from the increase in the Choice-Elimination condition (4.23) at the 3% level (t = 2.43, df = 26). Quite clearly, then, prior freedom to choose the eliminated alternative accounts for the increased attractiveness. Equally clearly, neither the increased salience of the alternative due to the elimination nor the elimination per se leads to increased attractiveness. While the decreased attractiveness in the No Choice-Elimination condition is not reliably different from the slight increase in the Choice-No Elimination condition, it is very nearly a reliable decrease relative to zero change (t = 2.00, p < .07).

Questionnaire responses. Data from four questions designed to serve as additional measures of the attractiveness of the rank 3 record were generally consistent with the rating changes and will not be reported in detail. Of more interest are the questions concerning suspiciousness, reactions to the elimination, and awareness of the possibility that the elimination might affect attractiveness ratings. The responses of subjects indicated that, for some unknown reason, there was somewhat more suspiciousness about the purpose of the procedure in the Choice- Elimination condition than in either of the other two conditions. This suspiciousness seems to have worked against the hypothesis, however, since the changes in attractiveness ratings for the relatively suspicious subjects, although not different in mean value, were much more variable than those for nonsuspicious subjects (p of diff. < .Ol). The questions on the awareness of the connection between the elimination and change in rating indicated that most subjects, particularly in the No Choice- Elimination condition, doubted there was any effect. It is interesting to note that in the Choice-Elimination condition the subjects who gave relatively great increased attractiveness ratings to the eliminated record were those who checked the response “I looked upon the record more favorably today, since the fact that it was unavaliable made it seem more attractive,” or those who checked the response “I seriously doubt that the fact that this record was unavailable could have had any effect on my judgments of the record.” Subjects who indicated they were aware of the possibility their ratings might be affected but tried not to be influenced showed a slight average decrease in rating.

Finally, as a check to see if subjects felt any frustration in regard to the elimination of the rank 3 record, subjects were asked if they were


annoyed or disappointed in regard to the elimination, and, if so, for what reason. The reasons which could have been checked by the subject to indicate frustration were: “because I definitely wanted this record as my free gift” and lLbecause I thought I might have wanted this record as my free gift,.” Since neither item was checked by a single subject in either Elimination condition, it appears that use of the rank 3 record for the elimination had its intended effect of avoiding frustration.

DISCUSSION The combined data of these two experiments indicate that where the

third-ranked of four choice alternatives was eliminated, it tended to become more attractive. Although this increase in attractiveness was predicted as an effect of reactance, it could also have been due to the increased attention given the eliminated alternative or to a general tendency among people to want whatever they cannot have. Therefore, a condition was included in the second experiment in which there was no initial freedom to choose any of the four alternatives, but in which the third-ranked alternative was nevertheless eliminated as a possible gift item. The results of this condition show quite clearly that without prior freedom to choose it, an eliminated alternative shows no increase in attractiveness and, if anything, becomes less attractive. Thus, neither the increased attention given to the eliminated alternative nor the elimination per se could account for the increased attraction obtained in the condition with prior freedom to choose. It seems safe to conclude that the more a person believes himself free to have a given alternative, the more likely it is that the alternative will increase in attractiveness upon being eliminated.

In the first experiment the failure of the importance manipulation to affect the responses of subjects is probably due t,o a methodological failure rather than to a faulty theoretical formulation. As noted earlier, our best guess is that the relatively low attractiveness of the selection on the record which was eliminated meant to the subject that the whole record was probably not very attractive. It is nevertheless possible that the absolute attractiveness of choice alternatives does not affect the magnitude of reactance aroused by the elimination of one, and only further research can determine whether or not this is true.

REFERENCE BREHM, J. W. A theory of psychological reactance. New York: Academic Press,

in prem.

(Received January 12, 1966)

The attractiveness of an eliminated choice alternative

Documents