an attempt to predict the performance of three-man teams from attitude measures

12
This article was downloaded by: [University of Glasgow] On: 22 December 2014, At: 16:25 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK The Journal of Social Psychology Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/vsoc20 An Attempt to Predict the Performance of Three- Man Teams from Attitude Measures Loren J. Chapman a & Donald T. Campbell a a Departments of Psychology , University of Chicago and Northwestern University , USA Published online: 30 Jun 2010. To cite this article: Loren J. Chapman & Donald T. Campbell (1957) An Attempt to Predict the Performance of Three-Man Teams from Attitude Measures, The Journal of Social Psychology, 46:2, 277-286, DOI: 10.1080/00224545.1957.9714327 To link to this article: http://dx.doi.org/10.1080/00224545.1957.9714327 PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content.

Upload: donald-t

Post on 17-Apr-2017

213 views

Category:

Documents


0 download

TRANSCRIPT

This article was downloaded by: [University of Glasgow]On: 22 December 2014, At: 16:25Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number:1072954 Registered office: Mortimer House, 37-41 Mortimer Street,London W1T 3JH, UK

The Journal of SocialPsychologyPublication details, including instructions forauthors and subscription information:http://www.tandfonline.com/loi/vsoc20

An Attempt to Predict thePerformance of Three-Man Teams from AttitudeMeasuresLoren J. Chapman a & Donald T. Campbell aa Departments of Psychology , University ofChicago and Northwestern University , USAPublished online: 30 Jun 2010.

To cite this article: Loren J. Chapman & Donald T. Campbell (1957) An Attempt toPredict the Performance of Three-Man Teams from Attitude Measures, The Journalof Social Psychology, 46:2, 277-286, DOI: 10.1080/00224545.1957.9714327

To link to this article: http://dx.doi.org/10.1080/00224545.1957.9714327

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of allthe information (the “Content”) contained in the publications on ourplatform. However, Taylor & Francis, our agents, and our licensorsmake no representations or warranties whatsoever as to the accuracy,completeness, or suitability for any purpose of the Content. Any opinionsand views expressed in this publication are the opinions and views ofthe authors, and are not the views of or endorsed by Taylor & Francis.The accuracy of the Content should not be relied upon and should beindependently verified with primary sources of information. Taylor andFrancis shall not be liable for any losses, actions, claims, proceedings,demands, costs, expenses, damages, and other liabilities whatsoeveror howsoever caused arising directly or indirectly in connection with, inrelation to or arising out of the use of the Content.

This article may be used for research, teaching, and private studypurposes. Any substantial or systematic reproduction, redistribution,reselling, loan, sub-licensing, systematic supply, or distribution in anyform to anyone is expressly forbidden. Terms & Conditions of accessand use can be found at http://www.tandfonline.com/page/terms-and-conditions

Dow

nloa

ded

by [

Uni

vers

ity o

f G

lasg

ow]

at 1

6:25

22

Dec

embe

r 20

14

T h e Journal of Social Prychology, 1957, 40, 277-286.

AN ATTEMPT TO PREDICT THE PERFORMANCE OF

MEASURES*

Deparirnenir of PsychoIogy, Unioerrity of Chicago and Notthwuertern University

THREE-MAN TEAMS FROM ATTITUDE

LOREN J. CHAPMAN AND DONALD T. CAMP BELL^

A. PROBLEM T h e present paper reports an attempt to predict performance on tasks

requiring cooperative effort by means of a group administered paper and pencil attitude test. Such predictions are of two sorts: (a) main effects, i.e., simple relationships between test scores and performance of teams in which S serves; and ( b ) interaction effects, e.g., a relationship between the combinations of test scorers on a team and the team’s performance.

B. METHOD 1. Attitude Scales

All of the attitude scales utilized were incorporated into a single ques- tionnaire. T w o of the scales were central in the experiment in that ex- perimental groups were selected from the extreme scorers. T h e other scales in the questionnaire were used in analyzing results through correlational procedures alone.

T h e first scale used in selecting experimental groups is a modification of the F-Scale of Adorno, et al . ( l ) , designed to measure authoritarian per- sonality trends. All the items of the original F-Scale are worded so that agreement with the item contributes to a high F score. Sixteen such items are used as a separate scale in this report, and are indicated as “,+.I’ For example:

In

People can be divided into two distinct classes: the weak and the

No sane, normal, decent person could ever think of hurting a close strong.

friend or relative.

the present test, 16 additional “F-” items were used, worded so that

*Received in the Editorial Office on January 17, 1956. 1This study was supported in part by the United States Air Force under contract

No. A F 18(600)-170 monitored by the Crew Research Laboratory, Air Force Per- sonnel and Training Research Center, Randolph Air Force Base, Randolph Field, Texas.

277

Dow

nloa

ded

by [

Uni

vers

ity o

f G

lasg

ow]

at 1

6:25

22

Dec

embe

r 20

14

278 J O U R N A L OF SOCIAL PSYCHOLOGY

disagreement with the item contributes toward a high F score. These items are modifications of original F-Scale items, and have been included in an effort to remove the “response set” variance which Cronbach (4, 5 ) has shown to arise from having all items worded in the same direction. Ex- amples of F- items are:

Getting rid of all the immoral, feebleminded and crooked people would

It would be a good thing if more young people kept their youthful not actually solve many of the country’s problems.

rebellious ideas as they grew older.

F o r experimental subject selection, the total F score was used. In cor- relational analyses, the two halves are treated separately.

T h e items were answered on a five-level scale, i.e., strongly agree, agree, undecided, disagree, and strongly disagree. T h e undecided category was scored so as to give, as nearly as possible, an even dichotomy between the agree and the disagree.

T h e second scale used in selecting experimental groups was the Superior- Subordinate Scale (SS), designed to measure the direction of identification in situations involving a conflict between a superior and a subordinate. A high score reflects a tendency to take a point of view of the superior. T h e development of this scale is reported elsewhere ( 3 ) . T h e form used here contained 20 Likert type items and 17 forced choice items. T h e following are examples of the items used. The answer which contributes toward a high score on the scale is indicated as positive.

T h e schools today are in danger of being too: A. soft with the students. B. strict with the students (positive). “The advantage of any extra privileges that go with high job status

is more than outweighed by the burden of increased responsibilities. (Agree ia positive.)

T h e other scales in the questionnaire are Alienation, Identification with Discipline, and Cooperation. A high score on Alienation ( A l ) reflects a tendency toward psycho-social isolation with a cynicism about the worth- whileness of conventional values and about the sincerity of people. I t con tains 28 items. I ts development is reported elsewhere ( 3 ) . Examples of the items on the Alienation Scale are:

I’ve run into quite a few fellow students whose personality or charac- (Disagree is

If a person uses underhanded methods in deal ing with you, you a r e

ter was so fine I felt it could serve as a model for me. positive.)

justified in using the same methods in return. (Agree is positive.)

Dow

nloa

ded

by [

Uni

vers

ity o

f G

lasg

ow]

at 1

6:25

22

Dec

embe

r 20

14

LOREN J. CHAPMAN AND DONALD T. CAMPBELL 279

A high score on the Identification W i t h Discipline Scale (ID) reflects a tendency to readily accept discipline and to model one’s behavior after those who administer it. This scale has 14 items. Its development has been reported elsewhere ( 3 ) . Typical items are :

When I get a lower grade on a midterm than I deserve, I lose inter-

Usually when I’ve been severely scolded or punished, I see the jus- est in the course.

tice of it and a m glad for it.

(Disagree is positive.)

(Agree is positive.)

T h e only scale which was created especially for this study was the 38- A high score on this scale is supposed t o item Cooperation Scale ( C o o p ) .

TABLE 1 INTERCORRELATIONS A N D KUDER-RICHARDSON REI.IABILITIES OF THE SCALES*

Tota l F+ F- F ss A1 I D C d p

F+ ( 5 3 ) F- -.03 (.40) Total F .74 .63 (.43) ss . I7 .06 2 1 (.69) Al .07 -28 -.08 .oil ( .58) I D .17 .18 2 6 .25 -.25 ( 5 6 ) C d p -.03 .16 .07 -20 -.33 .24 (.62)

Those intercorrelations which are significant at the .01 level a r e italicized.

reflect feeling of belonging to groups, and a preference for doing things with other people rather than by one’s self. Examples of these items are:

(Disagree is positive.)

* ( N = 142.) T h e reliability values a re in parentheses.

Most parties a re rather dull affairs. I do best on my homework when I work with others ra ther than by

myself. (Agree is positive.)

T h e 146-item questionnaire was administered to 142 male students in an introductory psychology course a t Northwestern University. Table I shows the Kuder-Richardson reliabilities and intercorrelations of the six scales for this population.

2. The Tasks

Tasks were desired which would reflect the achievement of the group as a whole, rather than either the sum of the contributions of the individual members or merely the contribution of the most effective member.

F o r the verbal task a pilot study was done on three tasks which seemed to be possibly useful. These were a n anagrams game, a poker hand game, and a game of naming objects that belong to categories. T h i s latter was

Dow

nloa

ded

by [

Uni

vers

ity o

f G

lasg

ow]

at 1

6:25

22

Dec

embe

r 20

14

280 JOURNAL OF SOCIAL PSYCHOLOGY

found most suitable and was used in the experiment proper. This game begins with the leader of a team choosing a category from a list of six cate- gories, naming a letter, and asking one of his teammates to name an object in the category that begins with the letter. For the second question, the leader again chooses a category from the list and the other teammate answers with the name of an object in it that begins with the same letter that the last answer ended with. T h e leader continues to choose categories and his two teammates give the answers for a 3S-minute trial. T h e categories have point values, ranging from 1 to 4, which are empirically derived meas- ures of their relative difficulty. For each answer, the team receives the num- ber of points assigned to the category. If a member cannot answer, he may pass, but his team loses two points. When a member passes, the other team- mate is given the same letter, but the leader may, i f he wishes, choose a new category. T h e object of the game is to get the maximum number of points in the 3%-ninute trial. Thus, it is seen that each response reflects the activity of all three teammates, since it is dependent on the one teammate providing an easy letter, on the leader giving a good category, and on the third teammate giving a correct response. After each trial, but before the scores were announced, each S rated his two teammates on a 7-point scale for desirability as a co-worker. S was required to use whole numbers only and, on any one trial, to assign different ratings to his two teammates.

T h e second task was a modification of the Ball-and-Spiral Apparatus ( 5 ) . I t consisted of a spiraled ramp, made from a split garden hose, that winds around a wooden tripod. T h e tripod sits on a triangular wooden base from which extend three handles. T h e three Ss lift the apparatus by the handles and tilt it in order to make a :A inch ball bearing roll up the ramp. T h e slant of the ramp is such that as the ball goes higher, it becomes progressively more difficult to keep it on the ramp. There are 10 turns to the ramp, and a progressively higher point value is assigned to each turn. When the ball falls off, E records the score at the highest turn passed, and the team starts over with a new ball bearing. T h e object of the game is to get the highest possible sum of scores in the 3S-minute trial. As with the categories game, all Ss rate their teammates after each trial, and then the performance scores of the two teams are announced. Prior to the analysis proper, it was neces- sary to correct the ball-and-spiral game scores for differences in difficulty between the two pieces of apparatus. T h e corrected scores only were used in the analysis of the data.

Dow

nloa

ded

by [

Uni

vers

ity o

f G

lasg

ow]

at 1

6:25

22

Dec

embe

r 20

14

LOREN J. CHAPMAN A N D DONALD T. CAMPBELL 28 1

3 . Exper imenta l D e s i g n

For each of the experimental sessions, six Ss were assembled, three being from the upper 25 per cent of a test scale, and three from the lower 25 per cent. Each group of six was divided into two three-person teams which were required to compete in the performance of two different tasks. For each of the teams, one person was designated as leader. During an experimental session of two hours, the six persons were recombined into different teams 18 times, nine times for the first task, and the same pattern of nine combina- tions again for the second task. T h e possible variations in combining the high and low scoring Ss (here called H and L ) and in assigning H and L leaders permitted six types of teams. These are, with leader indicated by italics, H-H-L, L-L-HI H-H-L, L-L-H, H-H-H, and L-L-L. T h e same pairings of types of teams were always used, so that in the nine trials of each game in a given session, there were three in which a team of all H Ss competed with a team of all L Ss, three trials with an H-H-L team, com- peting with an L-L-H team, and three trials with an H-H-L team, com- peting with an L-L-H team. A Latin square design was used so that among three sessions or groups of Ss, given nine trials each on a game, each type of team appeared once in each of the nine trial positions. Within every group of Ss, each H S appeared in each type of team as often as every other H S, and each was leader one time for each type. Similarly, each L S ap- peared in each as often as every other L S, and was leader once in each. In all, seven sessions were run, involving 42 persons. There were two groups chosen from the extreme scores on the SS-Scale, three from the extremes of the F-Scale, and two groups that consisted of Ss who were extreme, in the same direction, on both the F- and SS-Scales. These represented all the sessions that it was possible to assemble from the tested population.

C. RESULTS 1. Analysis by G r o u p s

T h e data were examined to determine whether different types of teams differ in effectiveness of group performance on the games.

Since there were nine trials per session, with each of six types of teams appearing three times in each session, practice effects could be equated for types of teams only through counterbalanced sets of three sessions. Since there were four groups of extreme scorers on the SS-Scale, and five groups on the F-Scale, analyses that are limited to such sets of three sessions must, of course, omit some of the sessions. Table 2 presents the means of the six types of teams for such balanced sets of three sessions, one set being based

Dow

nloa

ded

by [

Uni

vers

ity o

f G

lasg

ow]

at 1

6:25

22

Dec

embe

r 20

14

282 J O U R N A L OF SOCIAL PSYCHOLOGY

on three groups of high and low SS scorers, and one set on high and low F scorers. It is difficult to apply a statistical test of differences to these data because only three scores contribute to each mean. W e are not helped by the fact that each of the three scores was in turn derived from three replica- tions of that type of team in the same session, because these replications a re not independent since they in part used the same Ss.

TABLE 2 MEAN PERFORMANCE SCORES WECEIVED BY DIFFERENT TYPES OF TEAMS I N THREE,

COUNTERBALANCED GROUPS, A N D THE SUM OF RANKS OF 2 SCORES FOR ALL GROUPS

Game Type of team*

Scale H H L HHL HHH LLH LLH LLL

Ball-and-spiral SS Mean Perf. Sum of R**

Ball-and-spiral F Mean Perf. Sum of R**

Category SS Mean Perf. Sum of R**

Category F Mean Perf. Sum of R**

61 50 63 54 5 3 46 7 14 13 15 1s 20

46 41 42 46 44 5 3 13 19.5 14.5 19 22 17 12 9 16 19 13 13 16 12 13 11 22 10 13 6 12 12 9 13 20 16 18.5 15 21 14.5

*The leader is in each case italicized. **High performance is indicated hy low rank.

T o make a test of significance more applicable to the data from all the sessions, and not just to the two possible combinations of counter-balanced sets of sessions, an attempt was made to remove position effects. Lumping together, for a given performance measure, all types of teams in all sessions, the means and standard deviations of team performances were computed for each ordinal trial. Since, for each trial, two teams competed, the num- ber of teams a t each trial was 14. Each score was converted into a stand- a rd score for its ordinal position. In these refined scores, effects of prac- tice were eliminated, and the effectiveness of each assembly type within each session could thus be compared. Each of the assembly types occurred three times within each seesion. T h e sums of these three occurrences have been ranked within each session, and the sums of these ranks are presented in Table 2. (Low rank in Table 2 means high performance.) T h e coeffi- cient of concordance has been used to state the degree of session to session similarity in the ranking of the six types of teams. For the SS groups, these values, corrected for continuity, are .34 for the category game and .31 for the ball-and-spiral game. For the F groups, the values are .08 and .12. None of these values reach significance a t the .05 level. Of the two scales, SS seems to be the more promising, though the hierarchy in excellence of

Dow

nloa

ded

by [

Uni

vers

ity o

f G

lasg

ow]

at 1

6:25

22

Dec

embe

r 20

14

LOREN J . CHAPMAN A N D DONALD T. CAMPBELL 283

types of teams almost reverses for the two tasks, and no easily interpretable pattern is apparent.

2. Analyses by Individuals F o r the analysis of the relation between subtest score of the individual Ss

and their game performance, three performance measures were computed. Individual scores on each game were computed. Individual scores on each game were computed by adding together the scores of all the teams on which S served. I n addition, a combination performance score was computed in such a way that each game contributed cqual variance.

T h e r e was no significant correlation between performance measures on the two games. W h e n computed by individuals, the r was -.06, N = 42. W h e n computed on the basis of groups, taking each specific leader-follower combination as a separtae item, the r was .18, N = 126. ( T h i s N of 126 is not appropriate for evaluating the significance of the r because of the pres- ence of the same individuals in many groups.)

F o r both the F- and SS-Scales, high and low scorers were compared on each of these three response measures by means of Student’s t . None of the t ratios attained significance a t the .05 level.

Using the entire 42 Ss, correlations were computed between the scores on each of the six test scales and each of the response measures. ( I t should be noted that these correlations would be inflated in the case of the F- and SS-Scales if there were a true correlation, because the Ss were chosen on the basis of their scoring on the extremes of these two scales.) None of the re- sultant r values are statistically significant. T h e two highest values, - 2 3 between the SS-Scale and category game score, and .27 between SS and ball- and-spiral score, are in opposing direztions, a fact that discourages speculation on their meaning. T h e corresponding values for the F+ items are -.06 and .06.

Further analyses were done on the relationship between ratings and game performance. T h e total ratings received by each S from his co-workers correlate .33 with performance score for the ball-and-spiral game, and 2 6 for the category game. T h e value for the ball-and-spiral game is significant a t the .05 level, while the other falls short of significance. T h e ratings on the two games correlate .31, which is significant a t the .05 level.

F o r both games, the ratings received by high scoring Ss on both the P- and SS-Scales were compared with the ratings received by low scoring Ss. T h e mean ratings and the probabilities that true differences exist are shown in Table 3. High rating means desirable as a teammate. As seen there, none of the differences attain significance, although there are trends toward

Dow

nloa

ded

by [

Uni

vers

ity o

f G

lasg

ow]

at 1

6:25

22

Dec

embe

r 20

14

284 JOURN.4L OF SOCIAL PSYCHOLOGY

significance, with low scores on the F- and SS-Scales tending to be preferred as teammates.

As with the performance scores, the ratings were correlated with scores on the test scales. T h e correlations are all nonsignificant. T h e values are somewhat higher than for the performance measures, and some are high enough to be interesting. T h e ,F+ Scale correlates -,24 with the ratings, in agreement with the findings of Bass et al. ( 2 ) in leaderless group dis- cussion. T h e Coiip Scale, the rationale of which is most applicable to group effort, yielded the highest value, .26. This might encourage further explora- tion of the use of this scale. T h e SS-Scale gave a value of -.05, and the values for the other scales were similarly low.

TABLE 3 MEAN RATINGS RECEIVED BY HIGH A N D LOW TEST SCALE SCORERS A N D PROBABILITY OF A

TRUE DIFFERENCE EXISTING F-Scale SS-Scale

High Low Low High

Category game Mean rating 4.78 5.06 4.85 5.13 P of a diff. .30 > P > 20 .2O>P > .10

Ball-and-Spiral game Mean rating 4.83 5.21 5.00 5.08

Total Mean rating 4.80 5.13 4.92 5.11 P of a diff. .20> P > .10 .20> P > .10 P of a diff. .20 > P > .10 .70> P > .60

T h e data were further analyzed to determine whether high scoring Ss on the F- and SS-Scales gave more desirable ratings to fellow Ss who scored high than they gave to low scoring Ss. This prediction seems to follow froin the general theory of authoritarianism. No significant differences were found for the F-Scale. While a significant difference appeared in the case of the SS-Scale, the difference appeared to result not from preference of high scorers by high scorers, but rather from a greater over-all generosity on the part of the high SS scorers. Therefore, a more direct correlational analysis of gener- osity of ratings was done. Correlations were computed between scores on the test scales and total ratings given to other Ss. T h e only value that came close to significance was .27 for the SS-Scale. T h e direction of the correlation shows the inore superior oriented to be more generous in their ratings of their fellows.

T h e ratings assigned to leaders and nonleaders were compared by means of a direct difference t test in order to determine whether leaders are given higher ratings than nonleaders. No difference was found, the t ratio being .51. T h e mean rating assigned to a S when he was leader correlated with the ratings he received when he was not leader with an r of .76, which is

Dow

nloa

ded

by [

Uni

vers

ity o

f G

lasg

ow]

at 1

6:25

22

Dec

embe

r 20

14

LOREN J. CHAPMAN AND DONALD T. CAMPBELL 285

significant a t the .01 level and indicates considerable consistency in the ratings. Nevertheless, a further analysis was done to determine if any preference for leaders, in assigning ratings, might be related to score on the test scales. T o do this, there was computed a preference score for each S, consisting of the mean rating that he gave all leaders minus the mean rating he gave all nonleaders. Comparison, by Student’s t, of high and low F- and SS-Scale scorers on this preference score revealed no differences. Also, the correlations of this leader preference score with score on the test scales yielded no significant values.

D. SUMMARY AND CONCLUSIONS Persons selected on the basis of attitude test scores were combined in vari-

ous fashions in three-man teams in the performance of a motor and a verbal task. Results were analyzed in terms of types of team and in terms of indi- vidual performance.

1. No significant superiority of any particular type of team was found. Trends indicated by coefficients of concordance of .34 and .31 show a tend- ency for groups led and staffed by persons with superior-oriented attitudes to be more effective on the motor coordination task, less effective on the verbal task. Similar trends were found in correlating individual test scores and performance measures, where values of .27 and -.23 were found. T h e I.’-Scale measure of authoritarian personality trends produced no trends ap- proaching significance, nor did exploratory scales of identification with dis- cipline, alienation, or cooperation.

There was little relation between a team’s performance on the verbal task and its performance on the motor task. This casts doubt on any generalized cooperation-coordination-integration factor as a general attri- bute of groups, a t least for briefly assembled ones.

Ratings of a person’s desirability as a teammate correlate significantly with the success of the teams of which he is a member. These ratings also show significant consistency from task to task.

Ratings received by persons fail to correlate significantly with atti- tude test scores. T h e trend indicated by a correlation of -.23 with the F-Scale tends to confirm previous research showing high F people to be less popular. A value of .26 with the Coop Scale is in the expected direc- tion and approaches significance.

T h e favorableness of ratings given by persons with superior-oriented attitudes tends to be higher than that of persons less superior-oriented, the over-all correlation being .27. No meaningful values were obtained for

T h e main outcomes are as follows:

2.

3.

4.

5.

Dow

nloa

ded

by [

Uni

vers

ity o

f G

lasg

ow]

at 1

6:25

22

Dec

embe

r 20

14

286 J O U R N A L OF SOCIAL PSYCHOLOGY

other scales. No differences were found when persons of different attitudes were compared on their tendency to give leaders higher ratings than non- leaders.

6. A n effort was made to revise the F-Scale by the inclusion of items which the high F person would disagree with. T h i s effort was totally un- successful, even though the new items represented rewordings of traditional F-Scale items. T h i s indicates the extreme dependence of the usual F-Scale scores upon response-set variance. F o r purposes of comparability, the re- sults reported in this summary have been in terms of the usual F-Scale items only.

REFERENCES 1. ADORNO, T. W., FRENKEL-BRUNSWIK, E., LEVINSON, D. J., & SANFORD, R. N.

2. BASS, B. M., MCGEHEE, C. R., HAWKINS, W. C., YOUNG, P. C., & GEBEL, A. S. J. Abn.

Assessing attitudes toward Draf t research

4. CRONBACH, L. J . Response sets and test validity. Educ. PJychol. MCUJ. , 1946,

5 . - . Further evidence on response sets and test design. EJuc. Psycho/.

6. FRENCH, J. R. P. Disruption and cohesion of groups. J . Abn. &f SOC. Psychof.,

T h e Authoritarian Personality. New York: Harper , 1950.

Personality variables related to leaderless group discussion behavior. U SOC. PJyChOI., 1953, 48, 120-128.

3. CAMPBELL, D. T., BURWEN, L. S., & CHAPMAN, J. P. superiors and subordinates through direct attitude statements. report submitted for monitor’s approval.

6, 475-494.

M ~ w J . , 1950, 10, 3-31.

1941, 36, 361-377.

D e p a r t rn e n t of P s y c h o l o g y U n i v e r s i t y of C h i c a g o C h i c a g o 37, I l l ino i s

D e p a r t m e n t of P s y c h o l o g y N o r t h w e s t e r n U n i v e r s i t y

E v a n s t o n , I l l ino i s

Dow

nloa

ded

by [

Uni

vers

ity o

f G

lasg

ow]

at 1

6:25

22

Dec

embe

r 20

14