A reexamination of the familiarization hypothesis in group risk taking

A reexamination of the familiarization hypothesis in group risk taking

JOURNAL OF EXPERIMENTAL SOCIAL PSYCHOLOGY A Reexamination of the Hypothesis in Group 346-350 6, ( 1976) Familiarization Risk Taking1 ALL...

382KB Sizes 0 Downloads 9 Views

JOURNAL

OF

EXPERIMENTAL

SOCIAL

PSYCHOLOGY

A Reexamination

of the

Hypothesis

in Group

346-350

6,

( 1976)

Familiarization Risk

Taking1

ALLAN I. TEGER U&e&y

of Pennsyl.vania

DEAN State

RICHARD ST. JEAN University

A total separate taking results dividuals tunity replicate after a problems

PRUITT

G.

University of New

York at Buffalo

GORDONA.

AND

HAALAND

of New Hampshire

of 226 subjects at three universities participated in five studies which measured the effect on individual risk of increased familiarity with the problem situation. The of all five studies failed to replicate the finding that inwill take increased risk after they are given an opporto become familiar with the problem. These failures to cast doubt on the hypothesis that the increase in risk group discussion is due to increased familiarity with the rather than to the effect of group interaction.

Although it has been repeatedly demonstrated that groups take greater risk than individuals, the cause of this “risky shift” has yet to be established. Most theories which seek to explain this finding attribute the risky shift to phenomena associated with group interaction. Bateson (1966) has challenged these theories on the ground that the risky shift is a pseudogroup effect, which can be explained in terms of individual decision making without recourse to interpretations based on group phenomena. According to Bateson, the risky shift is due merely to the increased familiarity with the problem situation that results from group discussion. In order to study the effect of familiarization independent of discussion, Bateson asked his subjects first to make initial decisions on the choice dilemma questionnaire (Kogan & Wallach, 1964), then to study the questionnaire, and finally to make new decisions. He found an in‘A large portion of this research from the Office of Naval Research. critical comments on the manuscript.

was The

supported by Contract N00014-67-C-0190 authors are grateful to Nathan Kogan 346

for

crease in risk from the first to the second decisions. Flanders and Thistlethwaite (1967) obtained similar results. It is the purpose of this paper to present a series of five separate replications of the familiarization technique-varying the procedure from one study to another to give +Jle familiarization hypothesis the fairest possible test. METHOD

AND

FtESULTS

Subjects for the first replication of the Bateson study were 7 men and I6 women from an introductory psychology course at the University of Delaware. The experiment was conducted during a regular class meeting. The subjects were given a booklet containing those choice dilemma items on which Bateson’s subjects had shown the greatest increase in risk ( I, 6, 7) and those which normally show a decrease in risk following group discussion (5 and 12). After the subjects had made their initial decisions they were given a new booklet and a booklet of blank paper and told to “write down all the points for and against the possible decisions,” They were told to take approximately 5 min on each problem, although the time was not announced every 5 min as Bateson had done. When they had finished this familiarization procedure, they were asked to indicate their decisions once again in the new choice dilemma booklet. There were no significant changes in level of risk taking on either the problems on which

Bateson

which normally 37, n.s. ). Study

had show

found

increased

a decrease

in

risk risk

(D

following

=

- .I7, group

t = 62,

KS.)’

discussion

or on those

(b=

*44,

t =

2

The second replication of Bateson’s study was conducted during a meeting of an introductory psychology class at the State University of New York at Buffalo with 16 men and 6 women serving as subjects. The procedure and choice dilemma Items were exactly the same as used in the original Bateson study. Once aga.in the familiarization procedure did not lead to a change in risk preferences (B = -2’7, t = .47, ns. ). sttdy

3

The third study was a replication of the critical condition from the Flanders and Thistlethwaite study. The procedure used by these authors was similar to that used by Bateson except that the familiarization routine was justified as a preparation for a group discussion, the entire 12 items of the choice dilemma were administered. and the subjects completed Eysenck’s extraversion scale after the pretest. Also, the familiarization procedure was less structured than in the Bateson study as the sub-jects were not reqzlired to make lists during their study of the problems Study 3 included an attempt to assess the effect of the administration of the extraversion scale after the pretest. Ii was hypothesized that the scaIe might have suggested that ‘Positive numbers of the total change the five studies.)

indicate over all

a change toward items. (Note that

greater risk. the number

ij refers of items

to the mean varies among

348

TEGER

ET

AL.

the experimenter regarded risk as a desirable trait-thus encouraging the subjects to increase their level of risk on the posttest. Twenty-three men and 16 women in an introductory psychology class at the State University of New York at Buffalo served as subjects during a regular class meeting. The instructions and procedure followed exactly those used in Flanders and Thistlethwaite’s “familiarization” condition except that one-half of the subjects received a filler scale instead of the extraversion scale following the pretest. The filler scale involved semantic differential ratings of risk-neutral words like “love” and “walk.” The results showed no risky shift for those subjects who had completed the extraversion

scale

(fi

=

.21,

t = .15,

n.s.)

and

tendency

toward

(5 = -2.90, t = 1.87, p < JO) for those who had completed difference between these two conditions was not statistically p < .20). (All t values are two-tailed. )

a cautious

the filler significant

shift

scale. The (t = 1.39,

Study 4 Although the original investigators had used a small laboratory for their research, our first three replications occurred in a classroom. In order to be certain that there was nothing about the use of a classroom which suppressed the risky shift, we ran a standard group discussion condition in the classroom. All members of the class were given a pretest, and then their chairs were rearranged to form 10 groups of from 3 to 5 members each. The groups discussed the items and arrived at new decisions. This procedure yielded a risky shift of the same magnitude that is usually produced by discussion in the laboratory (8 = 65). As a further check on the validity of our classroom results, we also ran a replication of the familiarization condition in a laboratory. Thirty-seven women from an introductory psychology course at the State University of New York at Buffalo were formed into 11 groups of 3 subjects and 2 groups of 2 subjects each. The groups were run according to the procedure of Study 3, except that one group was run at a time in a small laboratory with partitions set up between each subject. This was an exact replication of the Flanders and Thistlethwaite procedure except that the subjects were female instead of male. [Wallach, Kogan, and Bern (1962) have shown that sex differences do not affect the risky shift.] As a check on the possibility of experimenter bias, seven of the groups were run by an experimenter who had been told to expect a risky shift, while the other six were run by an experimenter told to expect a cautious shift. The mean shift which resulted was slightly in the cautious direction for both the experimenter

expecting

a risky

a cautious shift ( 8 = - .06). (t = .81, n.s.), and the combined

(B=

-.14,

Study

5

t=

shift

(B =

-.23)

The two groups mean shift from

and

the

experimenter

did not differ both conditions

expecting

from one another was nonsignificant

1.33, n.s.).

The Bateson procedure differed from that of Flanders and Thistlethwaite in that the subjects in the latter study expected a group discussion to follow the familiarization procedure. This factor of anticipated discussion is examined in study 5. Forty-five men and women from an introductory psychology course at the University of New Hampshire were formed into groups of five members each and assigned to the following three conditions: (a) Familiarization with anticipated

FAMILIARIZATION

IN

GROUP

RISK-

349

TAKEVG

discussion (FA)-This condition again replicates the critical condition of Flanders and Thistlethwaite except that the extraversion scale was not used and the posttest was administered after the familiarization was completed on each item, rather than waiting until all items had been completed. (b ) Familiarization (F )-Subjects followed the same procedure as in the above condition except that no mention was made of a forthcoming discussion. (c ) Discussion--Subjects received a. pretest followed by a discussion and a posttest. The choice dilemma items consisted of numbers 4, 6, 7, 9, 10, and II (a11 of which show consistent risky shift in group discussion), item 12, the medical doctor item from Rabow (1966), two items from Nordhoy ( 1962), and two items written by the author? (all of which show consistent cautious shifts following group discussion ). Sixty additional subjects were run with the same procedure but with items number 1, 4, 8, 11, the two items from Nordhoy, and the two by the authors. All groups were run in a small laboratory. As the results of the two parts of the study were identical, the data were combined for analysis. The risky items showed no significant changes in either the F or the tively). (.E = produced

FA conditions The cautious -.97,

t = 1.95, a risky

shift

(n = items n.s.; on

-1.53, t = 1.53, n.s.; 5 = --.08, t = .Ol, n.s,, respecalso showed no significant changes in either condition B = the

-.94, risky

t = 1.56, items

(b

ns., =

respectively).

3.75,

e = 4.10,

The

discussion

p < .QI)

but

no

significant change on the cautious items (5 = - 1.28, t = 1.28, as. ). The discussion condition showed greater change toward risk on the risk items than did either the F or the FA conditions (t = 4.65, p < Bl; t = 3.49, p < .Ol, respectively). There were no other significant differences between conditions. DISCUSSION

The shift toward risk following a group discussion is a reliable phenomenon, having been replicated in dozens of institutions by many investigators. If familiarization is the central co-mponent of the risky shift, as Bateson and Flanders and Thistlethwaite claim, familiarization alone should be sufficient to produce a reliable shift. Our present results do not support this claim. We are left with the question of how to account for the findings of Bateson and Flanders and Thistlethwaite. Bateson’s research took place in England. A possible explanation for his results has been suggested Kogan (personal communication) who notes that the English school system may? to a greater extent than the American system, encourage students ,to think carefully about a problem before making a decision. The response style which this produces would appear quite compatible with Bateson’s theory. Subjects may have been initially hesitant to accept risk since they had had little time to contemplate the alternatives. This reluctance may have disappeared following a period of study which they had been taught was necessary for wise decision making. 3 The two of childbirth

items follow the and job choice.

usual

choice

dilemma

format

and

deal

with

the

topics

TEGER

350

ET AL.

Flanders and Thistlethwaite’s research took place in the United States; hence their results are not subject to the above explanation. However, their data do not show the usual pattern of risky shift across problems. Although the pattern of risky shift across problems produced by group discussion is quite standard, the pattern produced by familiarization deviates from that pattern.4 This may indicate that entirely different determinants were operating in their familiarization condition than operate in the group discussion condition. Although a familiarization procedure has, in some instances, produced an increase in risk, it appears that the effect is too inconsistent to be acceptable as the basis of the risky shift. A group experience of some ,sort seems to be required in order to produce a consistent risky shift, and thus we must conclude that the risky shift is a true group phenomenon. REFERENCES

BATESON, N. Familiarization, group discussion, and risk-taking. Joumzl of Experimental Social Psychology, 1966, 2, 119-129. EYSENCK, H. J. The inheritance of extroversion-introversion. Acta Psychologica, 1956, 12, 95-110. FLANDERS, J. P., & THISTLETHWAITE, D. L. Effects of familiarization and group discussion upon risk taking. Journal of Personality and Social Psychology, 1967, 5, 91-97. KOGAN, N., 8r WALLACH, M. A. Risk taking: A study in cognition and personality. New York: Holt, Rinehart, and Winston, 1964. NORDHOY, F. Group interaction in decision-making under risk. Unpublished master’s thesis, Massachusetts Institute of Technology, School of Industrial Management, 1962. RABOW, J., FOWLER, F. J., JR., BRADFORD, D. L., HOFELLER, M. A., & SHIBUYA, Y. The role of social norms and leadership in ‘risk taking. Sociometry, 1966, 29,

16-27. TE'GER, A. I., & PRUITT, D. G. Components of group risk taking. Journal of Experimental Social Psychology, 1967, 3, 189-205. WALLACH, M. A., KOGAN, N., & BEM, D. J. Diffusion of responsibility and level of risk taking in groups. Journal of Abnormal and Social Psychology, 1962, 68, 263-274. (Received

March

5, 1969)

4 For example, the correlation of mean shift across problems between the data from Teger and Pruitt ( 1967) and Wallach, Kogan, and Bern ( 1962) is .83 (IV = 12, p < .Ol ). The correlation of the pattern produced by familiarization with that produced by discussion in the two above studies is .29 and .OO, respectively (N = 12, n.s. for both correlations). (As Flanders and Thisthlethwaite were unable to provide us with the raw data for these comparisions, the data from that study are based on the t values for the 12 items which are assumed to be roughly monotonically related to the mean risky shift.)