8 4 v u A r \ u - h ' O -. v c on r --r‘ r _ v. 0' u ...: ‘ a “(N‘- . SUG’J’S‘T' 6 c 1’? 3“! 0 D , "9W ---._-. <.‘v--'v' . THE GROUP RISKY SHIFT EFFECT AS A FUNCTION OF AFFECTWE BONDS, ACTUAL CONSEQUENCES, AND EXTENT OF RESPONSIBILITY mass or e egree . . .9 . o . - . . . -- . ‘ . T h . " ' 3’ ‘ -..---l‘ I ' ' .,.._ " ‘1": l . l - .. .v .4. ’ ' ‘9’. ‘2'; -’;:, I. _. v- "9 '- . . ~ ... ,.'. .1.» ‘ ~ ’ _ -_ ,_- - --9°. 5'23; L'-,.:,"'." — - I: I . - . t‘_ ” ' ' . _' 1:. -"—- o .‘o . :o.':- ;9-‘.'.‘:;.'. ' .I...."‘:-. .?:-'- 0;;‘1"- " - ' . .— I I ‘- f: :‘ .’- - I" .0 I":“.‘D.‘.'. ' . ;'."‘.-' -.- '.‘.IV a . , - . "' , _ - _' - — . . - .. “ . .‘ . 1 . I _ ,0 -‘-.~°'L9 . »r.-.~..-- ":1- - 2...- ._.~‘..,,.:, A o - - - - O - . '. >4-‘ -.“ o - ‘ - of . . .. o‘- 0' ‘ - '0' ~ ‘ ' . , - —‘00—o- r-fi . - ' ’1 - ' ' ‘ o - , >90 . _ - - -. 9 _ - a . ‘ can ' ' . - .. ...-1" .v—v~- - I .. . . v. - . a _ '.-0.V*~ "V‘ O c-” p "' ' . .. ’- ' - - v f. ' 9- " V ‘ c I.- . . ’-‘-" v - ’ ‘ ._‘ ..'_‘ q-o- .O'--’-O- «fir ' ‘ . , 4-..o' ‘ ‘ " _ ‘ ,_‘-o-.¢.'(:'-* - "‘ ' “ ‘.‘ 'T . -. -~.s¢-. ~v'oo ’ ,1 . --' A... * -- - " .' .."'v‘.-l-."‘ . c - . h§d H‘ 1" . ' _ A ,— ’.-~.‘ __'(4v "' f_ "“.._\.__."-'o~ vs.g".oscfi~‘ovl’ ' " _‘, ‘ . A..- av ”'7 0" - -“- C... u , ,. -¢-o.-r-" . .w u~\h”.". ‘34} t “ "I 4 n.;‘.‘.’.’£8:£(..«~' ‘ m. ml',49p~“M . .3 «.2 '1 5’ BINDING av w-HMG & WNS' ‘BUUK BINDERY INC. LIBRARY BINDERS ’ "vfiiiiiFGfiT. MICHIGAN r ‘ A _=t n ABSTRACT THE GROUP RISKY SHIFT EFFECT AS A FUNCTION OF AFFECTIVE BONDS, ACTUAL CONSEQUENCES, AND EXTENT OF RESPONSIBILITY By David L. Runyan Previous research has demonstrated the ubiquity of the risky shift effect in group decision making in comparison to individual decision making. Several hypotheses have been advanced to explain this phenomenon. The purpose of the present research is to test the validity of the diffusion of responsibility hypothesis proposed by Wallach and Kogan (1965). The diffusion of responsibility hypothesis emphasizes the role of affective bonds between group members that permit group members to share feelings of responsibility concerning the final decision. Shared feelings of responsibility reduce all group members' fear of failure from a possibly incorrect decision, and this leads to increased risk taking. This research examined the diffusion of responsibility hypothesis from three frames of reference. Affective bonds and David L. Runyan diffusion of responsibility were hypothesized to increase risk taking following group decision concerning a realistic gambling task designed to be analogous to the Kogan and Wallach Choice Dilemma Question- naires (1964). Previous research on the diffusion of responsibility hypothesis has been equivocal (Kogan & Wallach, 1967a, b; Teger & Pruitt, 1967). From one perspective the diffusion of responsibility hypoth- esis was examined by creating conditions in which the group decision was either binding or merely advisory in regard to an existing outside party. It was predicted that fear of failure would be greatly reduced if the group knew theirdecision to be merely advisory and therefore they would reach riskier decisions than those groups which were faced with a decision that was binding on the outside party. In an attempt to examine the diffusion of responsibility/ affective bonds hypothesis groups of mutual friends or groups of strangers were used to create conditions with either high or low levels of pre -existing affective bonds between group members. It was predicted that groups with high levels of affective bonds would exhibit greater risk taking than groups with low levels of affective bonds. Third, if diffusion of responsibility is critical for increased risk taking in groups, it seems more realistic to create a situation David L. Runyan with actual gains and losses that can be experienced by the members of the groups. It is difficult to generalize from hypothetical situations to those more directly analogous to real life. Therefore a realistic gambling situation was used that was designed to be analogous to the Kogan and Wallach Choice Dilemma Questionnaires that are the basis of 80 percent of all the research undertaken on the risky Shift phenomenon (Baron, Dion, 8: Miller, 1970). The present situation examined the effect of gambling with money for a real or hypothetical other. It was predicted that shifts would be smaller where the decisions have actual consequences. Forty -eight groups of 3- 5 subjects were assigned to 1 of 8 different conditions. All groups were instructed to reach consensus " Groups were on a suitable gamble involving an "outside person. composed of either friends or strangers, decisions were either binding or advisory, and the outside person was either real or hypothetical. Results supported two of the hypotheses: friends, and also groups reaching advisory decisions were significantly riskier than their counterparts. Real and hypothetical situations did not differ with respect to risk level, suggesting that generaliza - tion from hypothetical Choice Dilemma problems to more consequential situations is not as potentially invalid as previously believed. THE GROUP RISKY SHIFT EFFECT AS A FUNCTION OF AFFECTIVE BONDS, ACTUAL CONSEQUENCES, AND EXTENT OF RESPONSIBILITY By David L. Runyan A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Psychology 1971 ACKNOWLEDGMENTS I am particularly grateful to Dr. Lawrence Messé , my advisor and friend, for without his encouragement, patience, and wisdom this project could not have been undertaken. I would also like to thank Drs. Jeanne Gullahorn and William Crano for their time and insightful comments. Finally I would like to acknowledge Clint Tarkoe for his conscientious assis- tance in running subjects. This research was supported by the Cooperation/Conflict Research Group of the Computer Institute for Social Science Research. ii TABLE OF CONTENTS LIST OF TABLES . CHAPTER I. SUMMARY OF THE LITERATURE . Diffusion of Responsibility Cultural Value Hypothesis Familiarization . Persuasion . . Enhancement Models II. CRITICAL ISSUES IN RISKY SHIFT LITERATURE Problem and Hypotheses III. METHOD Subjects Choice -Dilemma Situation Design and Procedure IV. RESULTS Individual Decisions Group Decisions V . DISCUSSION . Conclusion iii Page NHCDODOO HI-I‘ 13 14 21 21 21 23 27 27 28 31 35 REFERENCES . . . APPENDIX: INSTRUCTIONS Real and Binding Instructions Real and Not Binding Instructions Hypothetical and Binding Instructions Hypothetical and Not Binding Instructions . iv Page 39 44 44 48 52 55 LIST OF TABLES Table 1. Probabilities of Success and Corresponding Rewards 2. Summary of the Analysis of Variance of Individual Choice Decisions 3. Summary of the Analysis of Variance of Risky Shift Scores Page 22 28 29 CHAPTER II CRITICAL ISSUES IN THE RISKY SHIFT LITERATURE In their recent review of the risky shift literature, Baron, Dion, and Miller (1970) offer some methodological criticism of this research area. They stated their criticisms under the headings of 1) theoretical chauvinism; 2) single testing sites; and 3) problems with hypothesis testing. They begin by suggesting that researchers pay exclusive attention to their own hypotheses with little regard or awareness that the five theoretical models discussed here are not logically incompatible with one another. Increased affective bonds may indeed lead to diffusion of responsibility and increased risk taking. How — ever, greater affective bonds might also facilitate information exchange thereby reducing uncertainty about the problem and increasing the salience of the relevant cultural values. Another problem with risky shift research is the almost exclusive use of the Kogan and Wallach Choice Dilemma questionnaires. 13 14 This standard procedure limits the generalizability of the obtained effects. Also, it is possible that decisions concerning hypothetical situations with no actual consequences is a more realistic explana - tion for the high frequency of risky shifts. A third major problem with the research in this area is the degree of difficulty encountered in constructing strong and direct tests of the hypotheses. For example, affective bonds have been manipulated by varying whether or not subjects are allowed to engage in group discussion. The affective bonds generated in a short dis- cussion-among a group of strangers are not likely to be strong enough to provide a viable test of this hypothesis. A better procedure would be to use groups of friends or families that have strong affective bonds already established. Problem and Hypotheses The phenomenon that a person' s willingness to take risks increases when decisions are made as a member of a group was first demonstrated by Stoner (1961). Since then, Stoner' s findings have been replicated for a number of subject populations (Lamm 8: Kogan, 1969; Marquis, 1962; Wallach, Kogan, & Bem, 1962). However, the "risky shift" has not been unconditionally accepted as an important concept in social psychology. For example, in their 15 critical review of the literature, Dion, Baron, and Miller (1970) present a number of theoretical and methodological issues that must be resolved before a valid evaluation of the utility of the construct canbe made. The present research addressed itself to three points raised by these authors. One point of controversy mentioned by Dion, Baron, and Miller (197 0) is the validity of the diffusion of responsibility hypothe- sis (Wallach, Kogan, & Bem, 1964). This hypothesis, as originally formulated, postulates that group discussion and consensus allows group members to share feelings of responsibility concerning the final decision. Responsibility for failure resulting from an incorrect decision is shared by all members of the group rather than by one particular member, and this leads to increased risk ~taking. Research results on the diffusion of responsibility hypothesis have been equivocal. Wallach, Kogan, and Bem (1964) produced results supporting the diffusion of responsibility hypothesis; Marquis (1962) and Teger and Pruitt (1967) obtained results that demonstrated that differing degrees of responsibility diffusion were irrelevant in determining group risk -taking. In these latter studies, however, it is uncertain whether the experimental manipulations of responsibility diffusion were effective enough to create the desired psychological state in the minds of the subjects. 1.6 Since the results of past studies were ambiguous, it seemed reasonable to examine the hypothesis from a somewhat different perspective. Therefore, the present research created conditions in which the group decision was either binding or merely advisory in regard to an outside party. It was predicted that fear of failure would be greatly reduced if the group knew their decision was only advisory and therefore they would reach riskier decisions than groups making binding decisions. A more recent modification of the diffusion of responsibility hypothesis emphasizes the role of affective bonds in groups which permit members to diffuse responsibility for risky decisions (Kogan 8: Wallach, 1967a, b; Wallach 8: Kogan, 1965). However, as Dion, Baron, and Miller (1970) indicate, research results on the affective bonds hypothesis have also been equivocal (e. g. , Kogan 8: Wallach, 1967d; Teger 8: Pruitt, 1967; Rettig, Johnson, 8: Turoff, 1967; Dion, Miller, 8: Magnan, 1969). These results, taken together, suggest that affective bonds, created only as a result of a discussion of risk relevant problems, is unlikely to be strong enough to provide an adequate test of the hypothesis. The studies cited all attempted to manipulate affective bonds solely within the confines of the experi- mental situation. A potentially more effective procedure would be to use groups of friends or co -workers with strong collective affective bonds already established. 17 Chandler and Rabow (1969), using a procedure that examined groups with pre —established levels of affective bonds, had family members (both parents and at least one child over fourteen) reach a decision on the usual Choice Dilemma Questionnaires (Kogan 8: Wallach, 1964). The groups were asked to give advice to a person who had emotional significance to the members of the family. Instead of giving advice to a fictitious Mr. X, they were to advise their father. As hypothesized, groups of families, advising family members, emphasized group norms of caution, and made more con- servative pretest individual decisions and group decisions than did control groups of strangers. A high level of affective bonds should pre -exist in groups composed of particular family members. How- ever, it seems obvious now that this unique group should have an unusual predilection for caution in making decisions, especially when they. involve giving advice to the head of the family. This predilection for caution should not be so predominant in other groups with high levels of affective bonds such as groups of friends, co -workers, or members of a team. The present research used groups of friends in comparison with groups of strangers to examine the affect of high and low levels of affective bonds on group risk taking. It was predicted that high levels of affective bondswould lead to greater risk taking than low levels of affective bonds. 18 Finally, Baron, Dion, and Miller (1970) report that 80 percent of all the research on the risky shift phenomenon used the hypothetical Choice Dilemma Questionnaires created by Kogan and Wallach (1964). If diffusion of responsibility is critical for increased risk taking in groups, it seems more realistic to create situations with actual consequences in terms of gains or losses that are actually felt or perceived by the group members. It is difficult to generalize results obtained from hypothetical situations to those from a situation more directly analogous to real life. Some research on the risky shift effect has used situations with actual consequences; however, to date none of these studies worked with the Choice Dilemma Questionnaires. Wallach, Kogan, and Bern (1964) gave subjects increasing monetary reward for correct answers to College Board Entrance Examination problems of com- mensurate difficulty, and still obtained a risky shift. However, in studies involving other experimental situations, Flanders (1970) and Clement and Sullivan (1969) demonstrated a conservative orienta- tion with actual consequences. A few researchers have used pure gambling situations with actual payoffs. Some results have shown no shift, or else a shift to caution (Lonegran 8: McClintock, 1961; Zajonc et a1. , 1968). Others have found risky shifts (Zajonc et al. , 1968; Zajonc et a1. , 1969). 19 The present research used a stimulus situation that had actual consequences yet was directly analogous to the widely -used Choice Dilemma Questionnaire. The situation used was Similar to that employed by Zaleska and Kogan (1971) who examined the effect on risk of gambling for oneself versus gambling for another person. The present situation, on the other hand, examined the effect of gambling for a real or hypothetical other. In this way, the effect on risk of high (real other) versus low (hypothetical other) conse- quences was examined directly, and since choices always affected a party outside the group, the situation was analogous to the Choice Dilemmas. In summary, the present research examined the effect of differences in (a) degree of responsibility (binding versus advisory decisions), (b) level of affective bonds (friends versus strangers), and (c) degree of consequences (real versus hypothetical decisions) on changes in risk -taking behavior as a function of group discussion. If the diffusion of responsibility hypothesis and its affective bonds corollary are valid, greater risky shifts should occur when decisions are advisory and group members are friends. Moreover, if, as Dion, Baron, and Miller (1970) imply, the risky shift occurs primarily when decisions are hypothetical, there should be smaller shifts when decisions have actual consequences. On the other hand, if persons 20 can "role play" effectively, their hypothetical decisions should accurately reflect their "actual" behavior and there should be no difference in shift between real and hypothetical choices. CHAPTER III ME THOD Subj e ctS One hundred and seventy -three students at Michigan State University were randomly assigned to 48 homogeneous groups of 3 to 5 male or female students. For half the groups (strangers), subjects were randomly selected from a group of 500 respondents to an advertisement in the school newspaper asking for volunteers to participate in "motivational research" for pay. Groups of friends were recruited by contacting one member of the subject pool and requesting him to bring 2 to 4 mutual friends of the same sex to the experimental session. Choice Dilemma Situation A single modified Choice Dilemma problem was used as the group task. Each group was asked to reach a decision on whether or not to gamble with the money earned by another person for writing an essay. If the group decided to gamble with his money, they were 21 22 to select one of six possible different gambles. Each gamble had a different probability of success associated with a monetary reward commensurate with the probability of success of the gamble. The expected value of all the gambles was approximately the same to eliminate rationality as a factor in the decision. Table 1 was given to all subjects. Table 1 Probabilities of Success and Corresponding Rewards Probability of Success .1 .3 .5 .7 .9 1.0(nobet) Reward $20.00 $7.00 $4.00 $3.00 $2.25 $2.00 The actual outcome of the gamble was decided by drawing a slip of paper from a box. According to the group' s decision, the correct portion of win slips and loss slips were placed in the box and the outside worker, for whom the gamble was relevant, was asked to choose a slip. If he chose a win slip, he was paid accord- ingly; if not, he was given $1. 00. All §_s were paid $2. 00. If the group decided not to gamble, the "outside worker" also received $2.00. 23 Design and Procedure Groups were run in one of eight different conditions. Each group'was composed of either friends or strangers, making either a binding or advisory decision, for either a real or fictitious outside worker. The instructions the experimenter gave to the groups were not affected by the friends or strangers groups manipulations. The experimenter gave the same instructions regardless of whether he was talking to a group of friends or a group of strangers. However, instructions varied depending on whether the decision was binding or non -binding or real or hypothetical. Four sets of instructions were read verbatim to groups in the appropriate conditions. This was done tocontrol for errors of instrumentation. (See Appendix for a complete set of instructions.) The basic procedure was to have the group enter the room together. In the room, they were shown, through a window to a second room, another subject (actually a confederate) seated at a table with his back to them. The group was seated around a large table in every other chair and each individual was given a pencil and acopy of Table 1. In all real conditions requiring an outside worker the group'was informed that he was also a member of the study who had been required to arrive a half hour early. After the experiment 24 was explained to him he was set to work writing an essay giving pro and con arguments for the legalization of marijuana. The purpose of this procedure was to show that the outside worker had earned his $2 pay just as these subjects would. Therefore, it had real value and its loss would be meaningful to him should the gamble fail. This is analogous to the Choice Dilemma problems. The confederate or "outside worker" was not used in the hypothetical conditions. Once the group was seated, subjects in the hypothetical conditions were simply asked to "imagine" that there was one more person participating in the study who had arrived half an hour earlier and been set to work writing an essay giving pro and con arguments concerning the possible legalization of marijuana. After an explanation of Table 1 each person in all conditions was instructed to decide individually the minimum probability of success he'would require before advising the outside worker to gamble with his pay for a more rewarding alternative. Once each person reached a decision, he was asked to record it privately on his copy of Table 1 without discussion. When all members had completed this task, they were asked to reach a group consensus on the same problem. They'were told that the previous individual decisions were preliminary steps to acquaint them with the problem. After consensus was reached and recorded, the outside worker was brought in and the final decision was implemented. 25 In real conditions the outside worker actually drew one of the ten slips out of the box after the correct percentage of win and loss slips were placed in the box. Depending on the success of his draw he was paid accordingly and then left. The group was detained a few minutes to allow the outside worker to depart, and then they were also paid and allowed to leave. The pretext for detention was to obtain a final post test individual decision. However, these data were not considered because the final post test decision was confounded by whether the gamble turned out successfully or not. In the hypothetical groups, once a consensus opinion had been reached and recorded the groups were dismissed. The outside worker was merely imaginary in the hypothetical conditions; there- fore these groups never witnessed the consequences of their decision. If the group was merely to advise the outside worker (either real or imaginary) on whether to gamble or not, this was emphasized in the instructions so that all subjects understood the advisory nature of their decision. If the decision was binding on the outside worker, this was also made clear to each member of the group. In the real advisory conditions, the outside worker was brought in and the advisory decision'was explained to him, but he was allowed to make his own decisions. In practice, the confederate chose "no bet" in the real 26 advisory condition. He was then paid and allowed to leave. Again the group was detained for a few minutes to allow him to depart. CHAPTER IV RESULTS Iliiiividual Decisions A 2 (affective bonds) X 2 (consequences) X 2 (responsibility) unweighted means analysis of variance of the individual decisions was performed and is summarized in Table 2. Preliminary inspection of the data indicated that there were no differences as a function of sex; therefore, this variable was dis- regarded in subsequent analyses. Also, as Table 2 indicates, groups within conditions were included as a nested factor in the analysis. However, since a preliminary test indicated that the effect of this variable was not significant, a pooled error term was used for sub- sequent tests. This analysis revealed a significant difference in the predilection for risk taking between subjects in the binding decision and advisory decision conditions. Subjects who made binding deci- sions were less risky (X = . 67) than were subjects whose decisions were advisory (. 51). 27 Summary of Analysis of Variance of Individual Choice Decisions 28 Table 2 Source E MS 3 Affective Bonds (A) 1 .03 . 83 Consequences (B) 1 .02 .33 Responsibility (C) 1 . 71 11. 83* A X B 1 .01 . 17 A X C 1 .00 .00 B X C 1 .00 .00 A X B X C 1 . 03 . 50 Pooled Error 165 .06 Groups/ABC 40 . 06 l. 20 Subjects/Groups 125 .05 *£ < . 01 Grouj Decisions Since subjects in the binding decision condition were sig- nificantly less risky in their original, individual choices than were subjects in the nonbinding, advice condition, the amount of shift that groups could manifest was different as a function of differences in this factor. To compensate for this difference in potential shift, 29 scores were derived that indicated the amount of shift between individual and group decision, relative to the amount of shift possible. A shift score was the difference between an individual decision and the group decision divided by the difference between the individual decision and the most extreme score of the scale in the direction of theshift. These shift scores were subjected to a second 2X 2 X 2 unweighted means analysis of variance, summarized in Table 3. Table 3 Summary of the Analysis of Variance of Risky Shift Scores Source if MS 3 Affective Bonds (A) l . 1. 03 6. 44* Consequences (B) 1 .07 .44 Responsibility (C) 1 . 68 4. 25* A X B 1 . 12 . 75 A X C 1 . 00 . 00 B X C , 1 . 00 . 00 A X B X C 1 . 03 . 19 Groups/ABC 40 . 16 *_p< .05 30 Since a preliminary test revealed that the effect of the groups within conditions nested factor was significant (p < . 05), it was used as the error term for subsequent tests. The analysis revealed that there were significant main effects for both the affective bonds and degree of responsibility factors. As predicted by the diffusion of responsibility hypothesis, subjects had greater shift scores in the advisory decision condition (X = . 49) than the binding decision condition (. 09). As predicted by the affective bonds hypothesis, there was a greater shift in groups of friends (. 51) than in groups of strangers (. 07). However, there was no significant difference as a function of real (X = . 37) versus hypothetical (.21) consequences. Thus, the present study yielded no evidence that the risky shift is tempered when subjects' decisions have real consequences than when they are merely the result of "role playing. " CHAPTER V DISCUSSION The results provide strong support for the diffusion of responsiblity-affective bonds hypothesis. Groups of friends with previously established affective bonds made riskier group decisions than did strangers. Thus, it could be that as Kogan and Wallach (1967a, b) speculate, affective bonds increase group risk taking by expanding feelings of shared responsibility among group members. Close ties and feelings of empathy between group members could help them to approach the decisional task as a unit, and all members become equally responsible for the outcome, whether it is a success or a failure. These results do not provide direct support for the affec- tive bonds hypothesis as formulated by Wallach and Kogan (1965). According to this statement of the hypothesis, it is the stronger affective bonds developed between group members durig discussion of the risk relevant tasks that leads to feelings of shared responsi- bility. But, as stated in the introduction, affective bonds created in 31 32 the experimental situation alone are unlikely to be strong enough to provide an adequate test of the affective bonds hypothesis. The present study attempted to look critically at the basic foundation of this hypothesis, i. e. , groups with high affective bonds generate riskier decisions than groups with low affective bonds. Also providing support for the diffusion of responsibility hypothesis were the results indicating that advisory decisions created more risk taking than binding decisions. A binding decision increased the responsibility of the group for the fate of the other person. An advisory decision effectively reduces personal and group responsibility, as blame for an erroneous decision is shared not only by the advisory group, but ultimately withthe advisee himself. Baron, Dion, and Miller (1969) reported that 80 percent of all research on the risky shift phenomenon used Kogan and Wallach' s hypothetical Choice Dilemma Questionnaire for the group task. They argued that in this situation group-decisions have no actual con- sequences and, therefore, the risky shift cannot be generalized to real-life situations. The Choice Dilemma Questionnaires have groups reach an advisory decision concerning a hypothetical problem con- fronting a fictitious person. Group members are never confronted withthe consequences of their decisions. 33 The present study attempted to create a realistic situation (in the Aronson and Carlsmith [1969] sense of "experimental realism") analogous to that provided by the Choice Dilemma Questionnaires used in most risk taking research. The results show that groups confronted with a situation with actual consequences were just as risky as groups who responded to a situation that was directly analogous to the hypothetical Choice Dilemma problems. These data indicate that the‘speculation regarding the invalidity of generalizing from hypothetical to real life situations was not supported. It might be argued that the amounts of money presented as the stimulus to subjects in the real choice condition were trivial, and, therefore, all conditions were inconsequential. However, research on the norm of equity (e.g. , Lane 8: Messé, 1970) indicates that subjects who participate in research to make money do take such amounts seriously and are concerned about adequate payment. Therefore, it is likely that the decisions presented to subjects in the real -choice condition'were meaningful and did present them with actual dilemmas. This should not imply that the real choices in the present research would be as consequential as real decisions concern- ing careers or graduate education, etc. , the stimuli comprising the Choice Dilemma Questionnaire. However, the results of the present study do indicate that, within limits, subjects can role play very well 34 what their actual behavior would be, and there is no evidence that they cannot do so for even more important decisions. Although the results of the present research are consistent with the diffusion of responsibility -affective bonds hypothesis for increased group risk -taking, it should be noted that they are not incompatible with other explanations of the risky shift. For example, these results are not inconsistent with the cultural value theory of risk -taking (Brown, 1965). Perhaps friendship groups increase the saliency of the cultural value for risk taking. Saliency, is increased by social comparison with peers, who are significant others. Making binding decisions for others without knowledge of their values might increase a cultural value for caution and thereby reduce risk -taking. The significant difference in the binding/ advisory conditions in the preliminary, individual decision analysis might also be explained by cultural value theory. These data show original differences in pre- disposition to the value of risk ~taking. Moreover, the findings of the present research provide no insight into the degree to which the affective bonds generated in group discussion alone are sufficient to account for the risky shift noticed in groups of strangers. Teger and Pruitt (1967) and Blank (1968) obtained risky shifts using a balloting and an "information exchange" condition without verbal interaction between group members 35 who were strangers. It is difficult to assimilate these results with the diffusion of responsibility -affective bonds hypothesis. The present study merely provides evidence that affective bonds between group members lead to a greater degree of risk taking in group decisions; it does not provide direct evidence that group discussion is a necessary condition for this result to occur. Finally, the basic origin of the risky shift phenomenon is still unknown. It is impossible to state, from the results of this study or any other, the necessary. and sufficient conditions to account for the risky shift in group decisions. Many theoretical positions have some explanatory power and most are not mutually exclusive. In any event, the present research indicates that the group risky shift effect is augmented by high levels of affective bonds and is reduced by binding decisions, findings that are most congruent with the dif- fusion of responsibility explanation. Conclusion The risky shift has not been unconditionally accepted as an important concept in social psychology. Dion, Baron, and Miller (1970) present a number of methodological and theoretical issues that must be resolved. One of these deals with the reality and utility of the construct; another concerns the myriad overlapping 36 hypotheses used to explain the risky shift. Both of these problems could be more easily dealt with if the phenomenon was examined in real -life situations rather than using artificial groups resolving academic prlblems in a laboratory setting. The present study suggests that real groups of friends increases the risky shift and that the more important the group' s decision, 1. e. , the effects of that decision are binding on another person, the smaller the risky shift. Although the manipulation in this study designed to examine the effects of hypothetical vs. actual consequences produced no differences, these results must be dealt with. It is a reasonable argument that even amounts of money as great as $20 are not as significant as many dilemmas facing groups in real life or even those described in the hypothetical Choice'Dilemma Questionnaire. The cultural value theory would predict that problems of varying gravity would elicit differences in shift from cautious to risky. The diffusion of responsibility/ affective bonds hypotheses could be further strengthened if these results were replicated or magnified by experimentation with other groups having high levels of affective bonds between group members. The present research dealt with personal friendship cliques. Future research might utilize groups composed of business associates from one firm or one 37 department, members of an athletic team, a social club, a military squad, relatives, etc. Perhaps a pattern of group interaction would appear'which would lead the way to an understanding of the best method to magnify or reduce the risky shift in group decision. As these groups actually exist outside of the laboratory setting, any data concerning the-risky shift should have greater validity than data collected from "artificial" groups. Similar results from these studies should firmly establish the diffusion of responsibility/ affective bonds hypothesis as an explanation for the group risky shift effect. The importance of having all groups, dilemmas, decisions, and consequences close to real life cannot be stressed enough. Group dilemmas and decisions should involve situations that are meaningful to the particular groups in question. Military personnel should be given a military dilemma; college students should be asked to choose between the difficult but prestigious university and the easier but less famous institution. To construct a thoroughexamina- tion of the depth and breadth of the risky shift these groups can also be asked to make decisions concerning dilemmas outside their own fields of specialty and these canbe compared to decisions involving more familiar'situations. This procedure could be used to assess more accurately the relevance of the familiarization hypothesis to 38 the strength of the risky shift effect. The importance of leadership could also be more readily determined using real groups and mean- ingful situations. Another question that has never been systematically examined concerns any differences between a group making a decision for-itself or one of its members and a group making a decision for another group or individual. Perhaps role playing is easy, but per- haps a meaningful dilemma wouldlead a person or group to do something different than he claims he would do or would want another party to do. Again, all groups, dilemmas, decisions, and conse- quences should approach real life as nearly as possible if the origin of the risky shift is to be discovered. REFERENCES Allport, F. H. Social psychology. Boston: Houghton Mifflin, 1924. Aronson, E. , 8: Carlsmith, J. M. Experimentation in social psy- chology. In G. Lindzey 8: E. Aronson (Eds. ). The handbook of social psychology (2nd edition) 4. Reading, Mass. : Addison-Wesley, 1969. Baron, R. S., Dion, K. L., and Baron, P. Group norms, elicited values, and risk taking. Unpublished manuscript. Uni- versity of Minnesota: Minneapolis, Minn. , 1968. Bateson, N. Familiarization group discussion and risk taking. Journal of Experimental and Social Psychology, 1966, _2_(2), 1 19 -12 9. Bem, D. J. , Wallach, M. A. , 8: Kogan, N. Group decision making under risk of aversive consequences. Journal of Personality a_nd Social Psycholog_y, 1965, l, 453 -460. Blank, A. Effects of group and individual conditions on chance behavior. Journal of Personality and Social Psycholfl, 1968, 8, 294-298. Brown, R. Social psychology. New York: The Free Press, 1965. Burns, J. F. An extremity -variance model of risk -taking. Unpub - lished doctoral dissertation. School of Industrial Manage- ment, Massachusetts Institute of Technology, 1967. Burnstein, E. Decision making and problem solving in groups. Unpublished manuscript. University of Michigan: Ann Arbor, Michigan, 1967. 39 40 Chandler, S. , 8: Rabow, J. Ethnicity acquaintance as variables in risk taking. Journal of Social Psychology, 1969, _7_'_7_, 221 — 229. Clausen, Geraldine. Risk taking in small groups. Unpublished doctoral dissertation. University of Michigan, 1965. Clement, D. , 8: Sullivan, D. W. No risky shift effect with real groups and real risks. Psychonomic Science, 1970, _l_8_(4), 243 -2 45. Dion, K. L., Baron, R. S., Miller, N. Why do groups make riskier decisions than individuals? In L. Berkowitz (Ed. ). Advances in experimental social psychology, 1970, 5. New York: Academic Press. _ — Dion, K. L. , 8: Miller, N. The risky shift: True or pseudo group effect? Unpublished manuscript. University of Minnesota, Minneapolis, Minn., 1969. Dion, K. L. , Miller, N. , 8: Magnan. Group cohesiveness and social responsibility as determinants of the risky shift. Unpub- lished manuscript. University of Minnesota, Minneapolis, Minn. , 1969. Flanders, J. P. Does the risky shift generalize to a task with demonstrably non —trivial decision consequences? Proceed- ings of the American Psychological Association Convention, Miami Beach, Sept. , 1970. J Flanders, J. P., 8: Thistlewaite, D. L. Effects of familiarization and group discussion upon risk taking. Journal of Personality and Social Psycholog_y_, 1967, 5(1), 91-97. Gibb, J. R. The effects of group size and threat reduction upon creativity in a problem -solving situation. American Psychologist, 1951, 6, 324 (Abstract). Hoyt, G. C. , and Stoner, J. A. F. Leadership and group decisions involving risk. Journal of Experimental and Social Psy- chology, 1968, 1, 275-285. Kelley, H. H. , 8: Thibaut, J. W. Group problem solving. In G. Lindzey 8: E. Aronson (Eds. ). Handbook of social psy- chology (revised edition), Vol. 4. Cambridge, Mass.: Addison-Wesley, 1968. P. 81. 41 Kogan, N. , 8: Dorse, W. Effects of anticipated delegate status on level of risk taking in small decision making groups. Acta Psychologica, 1969. Kogan, N. , 8: Wallach, M. A. Risk taking: A study in cognition and personality. New York: Holt, 1964. Kogan, N. , 8: Wallach, M. A. Effects of physical separation of group members upon group risk taking. Human Relations, 1967a, 20, 41 -48. Kogan, N. , 8: Wallach, M. A. Group risk taking as a function of members' anxiety and defensiveness levels. Journal of Personality, 1967b, _3_5_, 57—63. Kogan, N. , 8: Wallach, M. A. The risky -shift phenomenon in small decision making groups: A test of the information exchange hypothesis. Journal of Experimental Social Psychology, 1967d, 3, 75-85. ‘— Krech, D., Crutchfield, R. S., 8: Bullechey, E. L. Individual in society. New York: McGraw-Hill, 1962. Lamm, H. Will an observer advise high risk taking after hearing a discussion of the decision problem? Journal of Personality and Social Psychologl, 1967, 6, 467 -471. Lane, I. M. , 8: Messé, L. A. The distribution of insufficient, sufficient, and oversufficient rewards: A clarification of equity theory. Journal of Personality and Social Psy- chology, 1971, in press. Levinger, 8: Schneider. Test of the risk as a value hypothesis. Journal of Personality and Social Psychology, 1961, _1_1(2), 165—169. Lonegran, S. G. , 8: McClintock, C. G. Effects of group member- ship on risk taking behavior. Psychological Reports, 1961, _8, 447 -455. Maderas, G. R. , 8: Bern, D. J. Risk and conservatism in group decision making. Journal of Experimental Social Psychology, 1968, _4_, 350-366. 42 Marquis, D. G. Individual responsibility and group decisions involving risk. Industrial Management Review, 1962, 3, 8-23. Nordhoy, F. Group interaction in decision making under risk. Unpublished Master' 8 Thesis, School of Industrial Manage - ment, M. I. T. , 1962. Rabow, J. , et al. The role of social norms and leadership in risk taking. Sociometry, 1966, 2_9, 16-27. Rettig, S., Johnson, F. A., 8: Turoff, S. J. Group responsibility, affiliation, and ethical risk taking. American Psychological Association Convention, Washington, D.C. , Sept. , 1967. Riecken, H. W., 8: Honans, G. C. Psychological aspects of social structure. In G. Lindzey (Ed. ). Handbook of social psychology, Vol. 2. Cambridge: Addison-Wesley, 1954. Rim, Y. Risk taking and need for achievement. Acta Psycholggica, 1963, _Zi, 108-115. Schacter, S. Deviation, rejection, and communication. Journal of Abnormal and Social Psycholog_y_, 1951, g, 190-207. Sherif, M. Group influences on the formation of norms and attitudes. In Newcomb, T. M. , 8: Hartley, E. L. (Eds.). Readings in social psychology (1st ed. ). New York: Holt, 1947. Steiner, personal communication, cited in Dion, K. L. , Baron, R. S. , 8: Miller, N. Why do groups make riskier decisions than individuals? Unpublished manuscript. University of Minne- sota, Minneapolis, Minn., 1969. Stoner, J. A. F. A comparison of individual and group decisions including risk. Unpublished Master's Thesis, School of Industrial Management, M. I. T. , 1961. Teger, A. I. , 8: Pruitt, D. G. Components of group risk taking. Journal of Experimental Social Psychology, 1967, _3, 189- 205. Wallach, M. A., Kogan, N., 8: Bem, D. J. Group influence on individual risk taking. Journal of Abnormal Social Psy- chology, 1962, _6_5_, 75-86. 43 Wallach, M. A., Kogan, N., 8: Bem, D. J. Diffusion of responsibility and level of risk taking in groups. Journal of Abnormal and Social Psychology, 1964, _6_8, 263 -274. Wallach, M. A. , Kogan, N. , 8: Burt, R. Can group members recognize the effects of group discussion upon risk taking? Journal of Experimental Social Psycholog_y, 1965, l, 379- 395. Wallach, M. A. , 8: Kogan, N. The roles of information discussion and consensus in group risk taking. Journal of Experi- mental Social Psychology, 1965, _1_, 1-19. Wallach, M. A., 8: Wing, C. W. Is risk a value. Journal of Personality and Social Psycholog, 1968, _9, 101-106. Whyte, W. H. The organization man. New York: Simon 8: Schuster, 1956. Zajonc, R. B., Wotosin, R. J., Wotosin, M. A., 8: Sherman, S. J. Individual and group risk taking in a two choice situation. Journal of Experimental Social Psychology, 1968, _4_, 89- 107. Zajonc, R. B., Wotosin, R. J., Wotosin, M. A., 8: Sherman, S. J. Group risk taking in a two choice situation: Replication, extension and a model. Journal of Experimental Social Psychology, 1969, _5_, 127-140. Zajonc, R. B., Wotosin, R. J., Wotosin, M. A., 8: Loh, W. P. Social facilitation and immitation in group risk taking. Journal of Experimental Social Psycholog_y_, 1970, 6, 26-46. Zaleska, M. , 8: Kogan, N. Level of risk selected by individuals and groups when deciding for self and for others. Sociometry, 1971, a, 198-213. APPENDIX INSTRUCTIONS APPENDIX INS TRUC TIONS R_eal and Bindigg Instructions "Now, as we enter the room, please look through the window on the left and notice the person who is working at the desk opposite the window with his back to you. " (Enter room 400. Make sure SS see outside worker.) (Seat them around conference table --every other chair. ) (Hand each S a copy of Probabilities and Reward Table.) ”The fellow you saw behind the window is also part of this study. For the last half hour he has been busy writing pro and con arguments on the possible, legalization of marijuana. Like your- selves, he volunteered to participate in motivational research for pay. He was told he would earn $2. 00 for about an hour of work. When he arrived, about a half hour ago, I explained this study to him just as I am about to explain it to you now. Then I set him to work writing pro and con arguments on the legalization of marijuana, the task he is still working on now. 44 45 ”What I want each of you to do right now is to decide individually, without consulting each other, the minimum probability of success you would require before you would gamble with the $2. 00 this person expects to earn for the more rewarding alternative. If you want, you can decide not to gamble at all, and he will receive his $2. 00 at the end of the hour. However, if you decide to gamble and win, he will be paid substantially more money for his work, depending on the amount of money and probability of success you choose to gamble for. If you choose to gamble, and lose, he will only be paid $1. 00 for his hour of work. As you can see from the table, the smaller the probability of success of the gamble, the more money he will be paid if the gamble is successful. For example, a gamble that has only 3 chances in 10 of being successful will pay him $7. 00 if it is successful. A gamble that has 9 chances out of 10 of being successful will pay him $2. 25 if it is successful. If you choose "no bet, " he is paid $2. 00. If you gamble and fail, he receives only $1. 00. "No matter-what the outcome of the gamble, each of you will receive the $2. 00 you agreed to work for. Now, will you take a few moments and write your choice on the table I handed you. Please make your decision by yourself without consulting each other and please think carefully about the choices before you write 46 down your decision. I will now give you a few moments to decide. " an: an: * =9: "Has everyone finished? OK, the decisions that you made individually were merely preliminary steps to get you to think about the various factors involved, and the possible consequences of each choice. Now, I would like you to talk to each other and to decide togetherwhether or not you want to gamble with his money, and if so, the minimum probability of success you would require before choosing to gamble for the more rewarding alternative. Take as much time as you need to reach a group decision. Once a decision is reached, I will go into the other room and tell him his writing time is up, ask him to give me his written pro and con arguments on the legalization of marijuana, and then bring him into this room and explain your decision to him. Beside this box I have eighteen slips of paper, nine marked win and nine marked lose. Depending on the reward you choose to gamble for, I will put the correct per- centage of win slips and correct percentage of lose slips inside the box. If he picks a win slip he will be paid the appropriate amount; if he loses he will receive $1. 00. If you decide not to gamble, he will be told his time is up, brought in, and paid $2. 00. Take all the time you need to come to a decision. " 47 (Once decision is made put slips in box.) (Go get outside worker.) (Ask him for argument paper. ) (Explain the group' 3 choice.) (Have outside worker choose and pay him accordingly.) (To outside worker) " Thank you for your participation; please put your name and address here to acknowledge receipt of your money, and then you are free to go. " (To group) "Finally, I would like each one of you to make a second individual decision concerning the probability of success you would choose for this person to gamble for. It does not have to be the same decision you made as a group, or the same as you initially chose yourself. Please do not discuss your decisions. When you have decided, write your decision at the bottom of your Probability and Reward Table. I will give you a few moments to make your decision. " **** ”Thank you very much for your cooperation. Please do not discuss this study with your friends as I can only use people who are ignorant of the content of my study. Do you all agree not to discuss this study except among yourselves? OK, sign your name and address here to acknowledge that you have received your money. " 48 Real and Not Binding Instructions "Now, as we enter the room, please look through the window on the left and notice the person who is working at the desk opposite the window with his back to you. " (Enter room 400. Make sure §S see outside worker.) (Seat them around conference table--every other chair.) (Hand each _S_ a copy of Probabilities and Reward Table.) "The fellow you saw behind the window is also a part of the study. For the last half hour he has been busy writing pro and con arguments on the possible legalization of marijuana. Like yourselves, he volunteered to participate in motivational research for pay. He was told he would earn $2. 00 for about an hour of work. When he arrived, about a half hour ago, I explained this study to him just as I am about to explain it to you now. Then I set him to work writing pro and con arguments on the legalization of marijuana, the task he is still working on now. "What I want each of you to do right now is to decide indi- vidually, without consulting each other, the minimum probability of success you would require before you would gamble with the $2.00 this person expects to earn for the more rewarding alternative. If you want, you can decide not to gamble at all, and he will receive his &. 00 at the end of the hour. However, if you decide to gamble 49 and win, he will be paid substantially more money for his work, depending on the amount of money and probability of successxiyou choose to gamble for. If you choose to gamble, and lose, he will only be paid $1. 00 for his hour of work. As you can see from the table, the smaller the probability of success of the gamble, the more money he will be paid if the gamble is successful. For example, a gamble that has only 3 chances in 10 of being successful will pay him $7.00 if it is successful. A gamble that has 9 chances out of 10 of being successful will pay him $2. 25 if it is successful. If you choose ”no bet, ” he is paid $2.00. If you gamble and fail, he receives only $1. 00. ”No matter what the outcome of the gamble, each of you will receive the $2.00 you agreed to work for. However, the decision you come to will not be binding on him; it will only be advisory. He will be informed of your decision, of your advice to him, but then he can choose to act on your advice, or else he can make his own, entirely different decision. Now, will you take a few moments and write your choice on the table I handed you. Please make your decision by yourself without consulting each other and please think carefully about the choices before you write down your decision. I will now give you a few moments to decide. " **** 50 "Has everyone finished? OK, the decisions that you made individually were merely preliminary steps to get you to think about the various factors involved, and the possible consequences of each choice. Now, I want you to talk to each other and decide together whether or not you want to advise him to gamble or not, and if so, the minimum probability of success you would require before advising him to gamble. Take as much time as you need to reach a group decision. Once a decision is reached, I will go into the other room and tell him his writing time is up, ask him to give me his written pro and con arguments on the legalization of marijuana, and then bring him into this room and explain your decision to him. Beside this box I have eighteen slips of paper, nine marked win and nine marked lose. After explaining to him the results of your advisory decision he will be given a moment to make up his mind whether to follow your advice or not. Depending on the reward he chooses to gamble for, I will put the correct percentage of win slips and correct percentage of lose slips inside the box. If he picks a win slip, he will be paid the appropriate amount; if he loses he will receive $1.00. If he decides not to gamble, he will be paid his $2. 00, and he will be free to leave. Take all the time you need to come to a decision. " 51 (Go get outside worker.) (Ask him for argument paper. ) (Explain the group' s choice.) (Have outside worker choose and pay him accordingly.) (To outside worker) "Thank you for your participation; please put your name and address here to acknowledge receipt of your money, and then you are free to go. " (To group) "Finally, I wouldlike each one of you to make a second individual decision concerning the probability of success you would choose to advise this person to gamble for. It does not have to be the same decision you made as a group, or the same as you initially chose yourself. Please do not discuss your decisions. When you have decided, write your decision at the bottom of your Probability and Reward Table. I will give you a few moments to make your decision. " =4: * * >4: "Thank you very much for your cooperation. Please do not discuss this study with your friends as I can only use people who are ignorant of the content of my study. Do you all agree not to discuss this study except among yourselves? OK, sign your name and address here to acknowledge that you have received your money. " 52 Hypothetical and Binding Instructions (Subjects enter room 400.) (Seat them every other chair. ) (Hand out Probabilities and Reward Table.) "Now, what I would like each one of you to do is to pretend, to imagine, that there is one more person participating in this study. Although you have never met him, and he has never met you, imagine that he is sitting and working right now in the room next door. Even though he is a participant in this study he arrived a half an hour before you did. At that time, I explained this study to him just as I am about to explain it to you, and then I sat him down to work on writing pro and con arguments on the possible legalization of mari- juana. "This person I'm asking you to imagine volunteered to be a participant in motivational research for pay, just like yourselves. He was originally told that he would earn $2.00 for about an hour of work. But what I want each one of you to do is to imagine, to pretend, that you have the actual control over the amount of money that this person can earn. Imagine that you are in a position to gamble‘with the money this person expects to earn if you so choose. I want each one of you todecide individually, without consulting each other, the minimum probability of success that you would require before you 53 would gamble with the $2.00 this person expects to earn, for the more rewarding alternative. If you want, you can decide not to gamble at all and he would receive his $2. 00 at the end of the hour. However, if you decided to gamble and won, he would be paid substantially more money for his work, depending on the amount of money and probability of success you choose to gamble for. If you choose to gamble and lost, he‘would be paid only $1.00 for his hour of work. "As you can see from the table, the smaller the probability of success of the gamble, the more money he would be paid if the gamble is successful. For example, a gamble that has only 3 chances in 10 of being successful would pay him $7.00 if it was successful. A gamble that has 9 chances out of 10 of being successful would pay him $2.25 if it was successful. If you chose "no bet, " he would be paid $2. 00. If you chose to gamble and lost, he would receive only $1.00 for his hour of work. No matter what your choice for him is, each of you will receive the $2.00 you agreed to work for. "Now, will you take a few moments and write your choice on the table I handed you. Please make your decision by yourself without consulting each other, and please think carefully about your choices before you write down a decision. I will now give you a few moments to decide. " **** 54 "Has everyone finished? OK, the decisions that you made individually were merely preliminary steps to get you to think about the various factors involved, and the possible consequences of each choice. Now, I would like you to talk to each other and to decide together whether or not you would want to gamble with this person' 5 money, and if so, the minimum probability of success that you would require before choosing to gamble for the more rewarding alternative. Take as much time as you need to reach a group decision. " =t< * * * "Finally, I would like each one of you to make a second individual decision concerning the probability of success you would require before gambling for the money the person expected to earn. It does not have to be the same decision you made as a group or the same as you initially chose yourself. Please do not discuss your decisions. When you have decided, write your decision at the bottom of your probability and reward table. I will give you a few moments to make your decision. " * * * a: "Thank you very much for your cooperation. Please do not discuss this study with your friends as I can only use people who are ignorant of the content of this study. Do you all agree not to discuss this study except among yourselves? OK, sign your name and address here to acknowledge that you have received your money. " 55 Hypothetical and Not Binding Instructions (Subjects enter room 400.) (Seat them every other chair.) (Hand out Probabilities and Reward Table.) "Now, what I would like each one of you to do is to pretend, to imagine, that there is one more person participating in this study, Although you have never met him, and he has never met you, imagine that he is sitting and working right now in the room next door. Even though he is a participant in this study he arrived a half an hour before you did. At that time, I explained this study to him, just as. I am about to explain it to you, and then I sat him down to work on writing pro and con arguments on the possible legalization of mari- juana. "This person I' m asking you to imagine volunteered to be a participant in motivational research for pay, just like yourselves. He was originally told that he would earn $2. 00 for about an hour of work. But, what I want each one of you to do is to imagine, to pretend, that you are in a position to advise this person whether or not to gamble with his salary, and if you decide to advise him to gamble, to advise him on the minimum probability of success you would require before you would advise him to gamble with his $2. 00 for the more rewarding alternative. Of course, he would not have to 56 accept your advice, whatever it is, but pretend that you are in a position to at least give him some advice that he can listen to and think about. As you can see from the table I handed out to you, the smaller the probability of success of the gamble, the greater the reward if it is successful. For example, a gamble that has only 3 chances in 10 of being successful will pay him $7.00 if it is suc- cessful. A gamble that has 9 chances out of 10 of being successful will pay him $2.25 if it is successful. If you choose to advise "no bet, " it means you are suggesting that he not gamble and just keep his $2. 00. If he gambles and fails he will receive only $1.00. No matter what you choose to advise him, you will receive the $2. 00 you agreed to work for. ”Now, will you take a few moments and write your decision on the table I handed you. Please make your decision by yourself, without consulting each other and please think carefully about the choices before you write down your decision. I will now give you a few moments to decide. " an: :4: an: * "Has everyone finished? OK, the decisions that you made individually-were merely preliminary steps to get you to think about the various factors involved, and the possible consequences of each 57 choice. Now, I would like you to talk to each other and to decide together whether or not you want to advise him to gamble with his money or not, and if so, the minimum probability of success you would require before recommending to him that he gamble for the more rewarding alternative. Take as much time as you need to reach a group decision. " * * * * "Finally, I would like each one of you to make a second individual decision concerning the probability of success you would advise this person to gamble for, if you would advise him to gamble at all. It does not have to be the same decision you made as a group or that you made initially by yourself. Please do not discuss your decisions. When you have decided, write your decision at the bottom of your probability and reward table. I will give you a few moments to make your decision. " * * * * "Thank you very much for your cooperation. Please do not discuss this study with your friends as I can only use people who are ignorant of the content of my study. Do you all agree not to discuss this study except among yourselves? OK, sign your name and address here to acknowledge that you received your money. ” In”trimming“ulmrutjjuwlyjlrI:I