LIBRA RY 
 
 OF THE 
 
 U N IVERSITY 
 
 OF ILLINOIS 
 
 370 
 
 no. 6-\2 
 
The person charging this material is re- 
 sponsible for its return on or before the 
 Latest Date stamped below. 
 
 Theft, mutilation and underlining of books 
 are reasons for disciplinary action and may 
 result in dismissal from the University. 
 
 UNIVERSITY OF ILLINOIS LIBRARY AT URBANA-CHAMPAIGN 
 
 di:c :$i! 19" 
 
 L161— O-1096 
 
Digitized by the Internet Archive 
 
 in 2012 with funding from 
 
 University of Illinois Urbana-Champaign 
 
 http://www.archive.org/details/influenceofleade09fiel 
 
UGt 
 
 no .9 
 
 GROUP EFFECTIVENESS RESEARCH LABORATORY 
 DEPARTMENT OF PSYCHOLOGY 
 UNIVERSITY OF ILLINOIS 
 URBANA. ILLINOIS 
 
 THE INFLUENCE OF LEADER-KEYMAN RELATIONS 
 ON COMBAT CREW EFFECTIVENESS 
 
 Fred E. Fiedler 
 Technical Report No. 9 
 
 Study performed under Contract N6-ori-07135 
 with the Office of Naval Research 
 
 Project on 
 Social Perception and Group Effectiveness 
 
 June, 1954 
 
GROUP EFFECTIVENESS RESEARCH LABORATORY 
 DEPARTMENT OF PSYCHOLOGY 
 UNIVERSITY OF ILLINOIS 
 URBANA, ILLINOIS 
 
 THE INFLUENCE OF LEADER -KEY MAN RELATIONS 
 ON COMBAT CREW EFFECTIVENESS 
 
 Fred E. Fiedler 
 Technical Report No. 9 
 
 Study performed under Contract N6-ori-07135 
 with the Office of Naval Research 
 
 Project on 
 Social Perception and Group Effectiveness 
 
 June, 1954 
 
 THE LIBRARY OF THE 
 
 SEP 27 1954 
 
 UNIVERSITY OF ILLINOIS 
 
. 
 
 . . 
 
 I 
 
 . 
 
 
7o 
 
 THE INFLUENCE OF LEADER -KEY MAN RELATIONS 
 ON COMBAT CREW EFFECTIVENESS 
 
 Fred j^. ii-c.-er 
 University of Illinois 
 
 Two studies dealing with the prediction of t3am effective- 
 ness are here reported; These studies investigated the relationship 
 between the formal leader's interpersonal perception and his team's 
 operating efficiency. The samples under investigation consisted of 
 B-29 Bomber crews and Army Tank crews. The former served 
 primarily for exploratory purposes while the tank crews were used 
 as a validation sample. The research has been based on the working 
 assumption that relevant interpersonal relations can be inferred from 
 standard sociometric procedures and from interpersonal perception 
 tests measuring Assumed Similarity (aS) which will be described 
 below. 
 
 Background 
 
 The research program of which the present studies are a part 
 aims to discover some of the psychological principles underlying 
 group productivity. We shall summarize briefly the earlier work which 
 led to the present investigations. 
 
 Measurement of Assumed Similarity . We have measured inter- 
 personal perception with the score, " Assumed Similarity between 
 Opposites ," or ASo. This score is obtained when we ask a subject (S) 
 to fill out a personality questionnaire under two different sets: to 
 predict (a) the responses of a person with whom he ran work very well. 
 

 
2 
 and (b) the responses of the person with whom he finds it most 
 difficult to cooperate. The difference score D (2) indicates the 
 similarity subject (S) perceives between his most and least pre- 
 ferred work companions. A person who perceives his most and 
 least preferred work companions as similar has high Assumed Simi- 
 larity (ASo); a person who perceives marked differences between them 
 has low ASo. 
 
 The split-half reliability of ASo scores ranges between .85 
 and .95 for a 60 item questionnaire (3). 
 
 Despite numerous attempts, we have not found ASo scores 
 to correlate consistently with traits measured by other personality 
 inventories. Assumed Similarity appears to measure some aspects 
 of emotional distance. A person who perceives others to be similar 
 (has high ASo) seems to desire closer emotional relations than a person 
 who perceives others as dissimilar. Scattered findings and interview 
 data further suggest that the person with high ASo is essentially accep- 
 tant. pliable, and receptive and is inclined to be uncritical of others. 
 A person having low ASo appears to be a more critical, reserved, and 
 analytic individual who rejects those with whom he cannot work (5). 
 The Relations of ASo to Effectiveness of High School Basketball Teams 
 
 We first utilized ASo in two studies of informal teams (4, 5). 
 A frankly exploratory study investigated 14 high school basketball 
 teams at the beginning of the season. These fama consisted of from 
 9-18 men each. ASo scores and sociometric preference ratings, by 
 which we could identify the team's "most preferred co-worker/' were 
 obtained from each S. Team effectiveness was defined as the proportion 
 of games the team had won by midseason. 
 
 As shown in Table 1. we found a correlation of -.63 between AS o 
 

 
 ■ 
 
 
 
 ' 
 
 
 
 ■ ' 
 
 
 
 
 ■ 
 
 • 
 
 ■ ■ 
 
 
 
 
 
 
 ■ 
 
 
 
3 
 of the team's most preferred co-worker and the criterion. In other 
 words, teams v»hich chose a low ASo person tended to be more 
 effective than those preferring a high ASo co-worker. 
 
 Seven good and five poor teams, tested at the end of the 
 season, were used as validation sample. A significant correlation 
 between ASo and the criterion (-.58) corroborated our original find- 
 ing (5). 
 
 Relation of ASo t o Effectiveness of Student Surveying Parties 
 
 The results on basketball teams were cross validated in a 
 second study of informal teams. Members of 11 surveying parties 
 participated in this study. These 3 to 4 man teams consisted of 
 civil engineering students taking a required field course in surveying. 
 
 We again obtained ASo scores and sociometric preference 
 ratings from each S. The criterion consisted of instructor ratings 
 of team accuracy. 
 
 The hypothesis was supported that ASo of the team's most 
 preferred co-worker (i.e., its informal leader) is negatively corre- 
 lated with team effectiveness (Table 1) (5). The findings were inter- 
 preted to indicate that effective teams chose informal lea'de'rs who 
 perceive many differences between their co-workers^ The 
 present report extends the earlier research to formally organized 
 . teams. 
 
 Work with Formal Teams ; B-29 Bomber Crews* 
 Hypotheses 
 
 The team effectiveness of informal teams was correlated 
 with the ASo score of only the team's most preferred co-worker, 
 i.e., its informal leader. For this reason we believed psychological 
 
 The groups used in this study also provided data for a project 
 sponsored by the Crew Research Laboratory of AFPTRC. We are 
 indebted to Dr. Robert L. French, Director of Research, CRL, and 
 Dr. Thornton B. Roby, Chief, Crew Assembly Section, for their 
 cooperation. 
 

distance (as measured by ASo) to be an attitude related to effective 
 leadership. We therefore hypothesized that the aSo score of formal 
 leaders would also be negatively related to crew effectiveness 
 criteria. 
 
 Table 1 
 Correlations of Teams' Most prefe^redCc-workers* 
 
 ASo Score with Criteria of Effectiveness 
 Sample Correlation N P 
 
 High School Basketball 
 teams tested pre -season 
 
 High School Basketball 
 teams tested toward 
 end of season 
 
 Student Surveying Teams 
 
 -.63 
 
 -.58 
 
 ** 
 
 14 .05 
 
 12 .05 
 
 -.51 22 .02 
 
 * Spearman Rho 
 ** Point bis e rial, significance tested 
 
 with student's t. (one tailed) 
 *** Pearson r. (one tailed) 
 
 However, several studies have pointed out that groups may 
 have informal as well as formal leaders (1, 11, 12). In cases where 
 the formal leader is not also the informal leader, his influence over 
 the group is considerably weakened. Hence, a special case of the 
 first hypothesis stated that the ASo score of the accepted leader 
 (i.e., who is informal leader or most preferred crew member) would *>« 
 correlated with effectiveness criteria. 
 

 
Procedure 
 
 Sample. Seventy B-29 crews, from four different training 
 
 classes, were available for this study. Each class contained 
 
 17 or 18 B-29 crews. 
 
 Crew Structure . A B-29 crew normally consists of 11 men. 
 Five are officers: The Aircraft Commander (AC), Pilot (P), 
 Navigator (N), Radar (or Video) Observer (VO), and the Bombar- 
 dier (B). Six are enlisted crew members: the Flight Engineer, 
 Radio Operator, and four gunners. 
 
 Assumed Similarity Tests . ASjo scores were obtained from 
 all crew members who were available for testing. The test con- 
 sisted of 80 items, such as the following: "I am often bored with 
 people", and "It annoys me to leave a task unfinished." 
 
 The test instructions asked S to predict the responses of 
 the man in the Air Force with whom he had worked best, and the 
 responses of the man with whom he had worked least well. The 
 test was given during two sessions four weeks apart. Reliability 
 for ASo in the 80 item test was obtained by the Guttman Split- 
 half formula and is .86 for one class (N = 178). 
 
 Sociometric Questionnaires . Sociometric preference 
 ratings were derived from a routinely administered "Position 
 Description Form", which asks Ss to indicate the persons on their 
 crew whom they would definitely prefer, moderately, or least 
 prefer tor five crew activities. These activities are, (a) organizing 
 a crew party, (b) going on leave, (c) going on a dangerous mission, 
 (d) loading special cargo, and (e) returning with the crew from be- 
 hind enemy lines. 
 

 
 ; ..' i : ■" 
 
 
 • 
 
 
 
 
 .- 
 
 
 
6 
 Responses to these five questions were highly intercor re- 
 lated; a pooled preference score was therefore computed which 
 indicated the extent to which a crew member chose various others 
 on his crew. Because we were particularly interested in relative 
 sociometric choices, the preference ratings were converted to 
 ranks within the crew. 
 
 Criterion Scores . 
 
 Two criteria of bomber crew effectiveness were originally 
 used. Both were developed by the Crew Research Laboratory, 
 Randolph Air Force Base. 
 
 Radar Bomb Scores - Circular Error Average (RBS) . This 
 measure is an error score indicating how far off the target a par- 
 ticular bomb would theoretically have fallen. The score is computed 
 as the average for 10 missions. According to the Crew Research 
 Laboratory the reliability of RBS is .4 to .6. 
 
 Control Time Error (CTE) . This is also a theoretical score 
 which indicates the number of minutes by which a plane would be 
 too early or too late at a certain predetermined point of meeting. 
 According to the Crew Research Laboratory, the Navigator (N) 
 accounts for most of the variance of this score. The reliability 
 of CTE is approximately .5 to .6. The 
 
 correlation between RBS and CTE as computed by the Crew Re- 
 search Laboratory is -.16 (N = 100). RBS and CTE are thus indepen- 
 dent criterion scores. 
 
 Tests of A Priori Hypotheses 
 
 One hypothesis of the study stated that ASo of the Aircraft 
 
 Dr. T. B. Roby, Personal communication. 
 

 
 
 • 
 
7 
 Commander (ASo r ) would be negatively correlated with criteria 
 of crew effectiveness. This hypothesis was not supported. (r*-. 24) 
 
 A subsidiary hypothesis stated that ASo _. would be nega- 
 tively correlated with crew effectiveness criteria in crews in which 
 the Aircraft Commander was most preferred crew member (AC = 
 MPC). This hypothesis was also not supported. In fact, we found a 
 positive correlation of .62 (N = 22, PX.01) between Control Time 
 Error and the ASo score of accepted (most preferred) Aircraft 
 Commanders. 
 
 Relations of ASo to Radar Bombing 
 
 General Considerations . Since the a priori hypotheses had 
 yielded negative results we used the bomber crew data for further 
 exploratory work. Hypotheses derived from these data were then 
 tested on a sample of Army Tank crews. 
 
 In our exploratory work on Air Force crews we decided to 
 work with the radar bombing criterion since it reflects one of the 
 most important functions of B-29 crews. Radar bombing is also 
 considered to be more nearly a crew product in contrast with Con- 
 trol Time Error which depends almost entirely on the Navigator. 
 
 Sociometric relations . Our studies of informal groups had 
 considered the sociometric relationship within the crew, since we 
 correlated only the ASo score of the informal leader (the most 
 preferred co-worker) with effectiveness criteria. Although the 
 hypothesis that the ASo of the "accepted" Aircraft Commander 
 would correlate with Radar Bomb Scores was not supported we 
 
 explored further whether ASo of the Aircraft Commander might 
 
 Dr. T. B. Roby, Personal communication 
 
8 
 correlate with radar bombing under still more clearly defined 
 sociometricr conditions. 
 
 In particular we considered the accepted leader's relations 
 with his keymen, i.e., the specialists on his crew who are most 
 directly concerned with criterion relevant tasks in radar bombing. 
 These keymen are the Radar Operator (VO) and the Navigator (N), 
 both of whom have radar equipment and perform related tasks 
 during RBS runs. 
 
 Results . Tables 2, 3, and 4 present the results of this ex- 
 ploratory work. As can be seen, high negative correlations were 
 found in two related sociometric conditions. In the first of these 
 the Aircraft Commander is Most Preferred Crew Member and 
 endorses one or both of his keymen (AC = MPC --> VO and/or N), 
 (Tables 2 and 3). 
 
 Where the accepted Aircraft Commander does not choose 
 his keyman,ASo and Radar Bomb Scores tend to be positively corre- 
 lated. A second related method selects crews in which Aircraft 
 Commander .and keymen mutually choose each other. Here again 
 ASo.^, correlates negatively with the criterion. (Table 4) To 
 judge from the magnitude of the correlations this is apparently a 
 less powerful method but it has the advantage of utilizing a some- 
 what greater number of crews. It is based on the assumption that 
 
 Sociometric endorsement was arbitrarily defined by the ranks 
 
 1 to 2 for high choice ( --£■ ), 2.5 to 3 for medium choice ( ), and 
 
 1.9 - 10 for rejection ( -/> ). These breaking points are in part based 
 on the desire to divide the groups into three equal sub-samples; how- 
 eve fc, it does not seem reasonable to assume that a person tied with 
 two others for first choice, or chosen as second most liked, is some- 
 one toward whom the AC is indifferent. A compromise was here ma<l» 
 by considering the second rank as still indicative of high prcfeienre, 
 even though this makes the highly liked group «cm*ewtiat larger than 
 the other two sub-groups. 
 
' 
 
 
 " 
 
 
 
 
9 
 
 Table 2 
 
 Correlation of the Aircraft Commander's ASo Score with Radar Bombing 
 
 Criterion under Selected Sociometric Relations between 
 
 the Aircraft Commander and the Keymen 
 
 Sociometric Condition 
 
 Rho 
 
 N 
 
 AC a MFC --*■ VO/N 
 
 -- VO/N 
 
 -/» vo/n 
 
 AC f MPC --> VO/N 
 
 — VO/N 
 
 -?» VO/N 
 
 -.81 
 
 i 
 U 
 
 .13 
 
 7 
 
 .42 
 
 7 
 
 -.03 
 
 18 
 
 -.80 
 
 5 
 
 -.67 
 
 7 
 
 (.01) 
 
 AC = Aircraft Commander --> High sociometric choice 
 
 VO = Radar Observer -- Neutral sociometric choice 
 
 MPC = Most Preferred Crew Member */> Low sociometric choice 
 
 As this study explored many hypotheses, tests of significance 
 ere not interpretable. 
 
 better 
 a leader will be/able to influence his keymen whence has a mutually 
 
 good relationship with them. 
 
 Relation of ASo to Control Time Error . It will be recalled 
 that we found no correlation between the ASo score of the Aircraft 
 Commander (AC g ) and Control Time Error (CTE) (.16, N - 51) 
 but that the correlation of ASo _ with CTE was .62 (N = 22) in 
 crews which accepted their leader. This relation is contrary to the 
 results of all previous studies, and we therefore compared the opera- 
 tions involved in CTE with those related to radar bombing. 
 
 Interviews with former B-29 personnel, as well as perusal 
 
 of various manuals, suggested* some basic differences in the Aircraft 
 As we indicated above, RBS and CTE are uncorrelated scores. 
 
10 
 Table 3 
 
 Correlation of the Aircraft Commander's ASo Score with 
 Radar Bombing under Different Sociometric Conditions 
 
 Sociometric 
 
 K =1 Radar 
 
 Operator- 
 
 K=Navi 
 
 gator 
 
 
 Condition 
 
 Rho 
 
 N 
 
 P 
 
 Rho 
 
 N 
 
 P 
 
 AC = MPC -■* K 
 
 -.76 
 
 8 
 
 (.05*) 
 
 -.93 
 
 9 
 
 * 
 (.01) 
 
 -- K 
 
 -.05 
 
 8 
 
 — 
 
 .43 
 
 10 
 
 ■— 
 
 -/:>K 
 
 -.05 
 
 9 
 
 — 
 
 .89 
 
 6 
 
 (.05) 
 
 AC i MFC ~? K 
 
 .03 
 
 £ 
 
 — 
 
 .05 
 
 8 
 
 — 
 
 --K 
 
 -.13 
 
 7 
 
 — 
 
 .36 
 
 10 
 
 — 
 
 /*K 
 
 -.56 
 
 10 
 
 * 
 
 -.11 
 
 12 
 
 
 AC = Aircraft Commander -* Positive sociometric cnoice 
 
 MPC = Most Preferred Crew Member j* Negative sociometric choice 
 K = Keyman -- Neutral sociometric choice 
 
 See Footnote to Table 2 
 
 Commander's relationship with his Navigator as against that with 
 other crew members. The aC -Navigator relationship seems to be 
 unique in approaching that of an advisee to an expert advisor. The 
 Navigator is charged with collecting and integrating various naviga- 
 tional data and he then advises the Aircraft Commander on the approp- 
 riate course heading. While the AC ordinarily is supposed to follow 
 his Navigator's directions during the operations relevant to CTE, 
 Aircraft Commanders vary in their willingness or ability to do so, 
 especially since the AC is allowed considerable latitude in using his 
 judgment, and may even disregard the Navigator's advice if he 
 so desires. 
 
11 
 
 Table 4 
 
 Correlation of the Aircraft Commander's ASo with the Radar 
 Bombing Criterion in Crews in which the AC 
 Reciprocates his Keymen's Choices 
 
 Sociometric 
 Condition 
 
 Rho 
 
 N 
 
 P 
 
 AC «-» VO and/or N 
 AC </* VO and/or N 
 
 -.48 
 .05 
 
 22 
 
 27 
 
 * 
 (.05) 
 
 * 
 
 See footnote to Table 2 
 
 
 
 
 On the basis of these considerations we decided that 
 Control Time Error represents a special case which will need to 
 be further explored when appropriate groups are available. It 
 leads to the hypothesis that high ASo on the part of accepted leaders 
 will facilitate . decision making or problem solving by groups in 
 which the leader is dependent for information and advice on his 
 subordinates. Our data suggest that CTE requires a different type 
 of leadership than Radar Bombing. 
 
 Discussion of exploratory analyses 
 
 Our rationale was concurrently developed with, and in part 
 guided, this exploratory work. It has led to a consideration of three 
 factors which affect the performance of formal teams. 
 
 ^ generalized interperso nal attitude toward co-workers . 
 This attitude is presumably reflected by the interpersonal perception 
 score ASo. If the leader's ASo is related to team effectiveness, this 
 generalized interpersonal attitude must in some way influence others 
 on the crew. The variable thus becomes operative only und«»T -con- 
 ditions in which the leader can in fact exert positive and negative 
 influence. 
 
12 
 The leader's relations to his keymen . A formally organized 
 team, such as a bomber crew, is composed of specialists. Most ob- 
 jective criteria of team effectiveness are primarily a function of a 
 limited number of these specialists. The leader must then influence 
 these specialists or keymen in order to affect the corresponding per- 
 formance criterion. For example, to affect crew effectiveness in 
 radar bombing, the Aircraft Commander's attitude must influence the 
 Radar Operator and/or the Navigator who have radar equipment, not 
 some other crew members, such as the Gunners, who have very little 
 influence in radar bombing, but who may be keymen for a different 
 criterion. 
 
 The leader's status. The leader's attitude, here measured by 
 ASo, will strongly affect the keyman's behavior only under certain 
 conditions of group structure, namely when the leader has power in 
 the group which allows him to apply formal or informal sanctions to 
 influence his keymen. 
 
 We may note in passing that the condition L = MPC ---»K, 
 (i.e., accepted leader endorses his keyman) closely parallels the 
 group structure in informal teams. It will be recalled that ASo of 
 the leader correlated negatively with group effectiveness in informal 
 teams. The formal and informal leadership is, of course, identical 
 where the formal leader is also most preferred crew member (18). 
 In common with most informal teams, our informal groups did not 
 consist of specialists and any team member could thus be a keyman. 
 As Whyte has pointed out, this is usually the person whom the in- 
 formal group's leader chooses and respects most highly (19). 
 
 Promising relations were also found for crews in which 
 the Aircraft Commander was most preferred crew member (MPC) 
 
13 
 and rejected his keymen (AC = MPC -/* K). Since these relations 
 did not reach the 5% level in the exploratory study, the formal hypo- 
 theses Cor the validation study were not based on them. We shall, 
 however, return to these relations in a forthcoming section of this 
 paper. 
 
 Ancillary Validation Study : Visua l Bombing Accuracy 
 
 The exploratory work with the radar bombing criterion led 
 to the main hypothesis that ASo of the formal leader will be nega- 
 tively related to the criterion under specified sociometric relation*: 
 Namely, where (a) the leader is most preferred crew member and 
 endorses the keyman, or (b) the leader and keyman endorse each 
 other. 
 
 The present analysis was made after the major validation 
 study on tank crews. Iteis presented at this point for the sake of sim- 
 plifying the report. 
 
 Criterion . One criterion of visual bombing accuracy which 
 
 is independent of RBS (* RBS #c V c s * 08, N = 100 ^ is the P ercent oi 
 satisfactory camera bombing runs. The reliability of this criterion 
 has been reported as ranging from zero to .54, for ordinary missions 
 (14). Because it is currently not considered satisfactory by the Crew 
 Research Laboratory it was not part of our original study. 
 
 The Percent Satisfactory Visual Camera Runs (%SVC) criterion 
 is unsatisfactory from a statistical point of view since approximately 
 50% of the crews obtained 100% SVC. The distribution is therefore 
 highly skewed and the correlations with this criterion can only be 
 interpreted as roughly indicating the direction of the relationship. 
 
14 
 Procedure , The analysis utilized the same sample of B-29 
 crews as reported above. The relationship of the Aircraft Comman- 
 der's ASo to the visual bombing criterion was investigated for crews 
 in which (a) the accepted Aircraft Commander endorses the Bombar- 
 dier, and (b) the Aircraft Commander and Bombardier endorse each 
 other. 
 
 Results . The results, while not significant, were consistent 
 in direction with those found using the radar bombing criterion. 
 These relations thus represent further corroborative evidence when 
 considered in the context of the entire investigation. (See Table 9) 
 
 Main Validation Study : Tank Crews 
 
 The two related hypotheses which emerged from the bomber 
 crew study were tested on a sample of 25 tank crews which parti- 
 cipated in Project STALK (Cf. 10). 
 Procedure 
 
 Background. The Ballistics Research Laboratory of the 
 Ordnance Corps, and the Office of the Chief of Army Field Forces 
 jointly conducted "Project STALK" in the Fall and Winter of 1953 to 
 determine the relative effectiveness of five models of tanks. 
 The present study supplements this Army experiment by attempt- 
 ing to determine the psychological factors which influence the rela- 
 tive effectiveness of tank crews. The specific results and proced- 
 ures of Project STALK are classified insofar as they pertain to the 
 operation and performance of equipment. For this reason they have 
 
 * 
 
 This study was made with permission of the Director, Ballistic 
 Research Laboratories, Aberdeen, Md., and Mr. Floyd I. Hill, Tech- 
 nical Director of STALK. We are especially indebted to Messrs. Hill, 
 Andrew J. Eckles, III, and Stanford C. Ericksen, whose active collab- 
 oration enabled us to conduct this study. 
 
15 
 been deleted from this report. 
 
 Subjects. Twenty-five tank crews from a fairly typical 
 battalion served as subjects. Their scores on the Army General 
 Classification Test ranged from 58 to 139 with a mean of 90.6. Of 
 the 142 men who participated in this study, about 30% had completed 
 no more than eight years of school. 
 
 Crew Structure. Each crew consisted of the following 5 men: 
 
 Tank Commander (TC), in charge of the crew 
 
 Gunner (G) 
 
 Driver (D) 
 
 Loader (L), who also assists the Gunner 
 Bowgunner (BG), who also assists the Driver. Under some condi- 
 tions the bowgunner did not participate as a member of the crew. 
 
 Several alternate crew members were also tested. 
 
 Design of STALK. The experiment was designed as a 
 Graeco-Latin Square. Each platoon, consisting of five crews, worked 
 with a different one of five tank models in each of the five phases of 
 the experiment. Criterion scores for our purposes could thus be 
 obtained by comparing crews within a platoon during one particular 
 phase, or by comparing all platoons over all phases. 
 
 Test Instrument . Similarity scores were obtained as in the 
 bomber crew study. The AS test contained 60 items selected from 
 among the 80 items of the test used in the B-29 study; it was admini- 
 stered one week before the experiment. 
 
 The split-half reliability of ASo scores, derived from this 60 
 item instrument, was .91 (N = 132). 
 
16 
 
 Sociometric Questionnaire. Sociometric questionnaires were 
 administered to the tank crews on three occasions. Ss responded to 
 the first questionnaire just prior to the beginning of the experiment, 
 to the second at the beginning of the fifth phase (three months later) , 
 and to the third questionnaire in the week after completion of the 
 project, almost four months after the first administration. 
 
 The first and second questionnaires asked each crew member 
 to rank in order of preference seven men in his platoon for three 
 situations: (a) men S would want to have as fellow crew members 
 under combat conditions (b) men he would be likely to nominate for 
 battlefield commissions, and finally (c) men with whom S was on 
 friendliest terms. The third questionnaire used the same stimulus 
 questions but differed from the first two by asking for the five most 
 preferred, and five least preferred men in the platoon. 
 
 Inspection of the questionnaires indicated high intercorrela- 
 tion of responses to the three sociometric stimulus questions and 
 we therefore averaged the responses to the three questions. Re- 
 sponses on the three questionnaires could then similarly be averaged 
 and "combined" into one sociometric preference rating. - This "com- 
 bined" rating was used in connection with analyses which considered 
 the average criterion performance rather than subcriteria obtained 
 within a "Phase". 
 
 Criteria of Crew Effectiveness 
 
 Each crew competed with other crews of the same platoon 
 during each of the five phases. Thus, each crew within the platoon 
 can be ranked with respect to other crews within the same platoon. 
 It is not feasible, however, to rank crews of different platoons on 
 the same tank models since the crew which worked with Tank Model 
 

 
 
 
17 
 A during a late phase would have had more experience in handling 
 tanks and in working together than crews which worked with Tank 
 Model A_ during an early phase of the experiment. 
 
 Three criteria which are described below were used for 
 our study. These were obtained by personnel of the Office of the 
 Chief of Army Field Forces and the Ballistics Research Labora- 
 tory and are based on test results from 25 different targets. 
 
 1. Time per Hit (T/H) is the average time in seconds 
 which a crew required to hit 5 assigned targets in each phase. 
 This score depends primarily on the gunner. The Time per Hit 
 can be obtained for each of the five phases (T/H., T/H,,, etc.) 
 and also as the sum over all phases (ST/H). (This treatment also 
 applies to Travel Time.) 
 
 2. Travel Time(T/T) is the average time of travel from 
 target to target on the course. It is a score which largely reflects 
 the driver's skill and ability in handling the tank. 
 
 3 # Probability of a crew winning an engagement ( F ). 
 This is a composite criterion score which indicates the probability 
 that a particular crew will win a duel with a similar tank. 
 
 The score is based on the formula 
 
 ,« 00 
 
 P A t P n, W h <t 34 )dt * P m (A) re(t z4)dt 
 
 * 4 (A) ^t 4 (A) 
 
 Where P = Probability of crew A winning the engagement 
 
 P (A) = Fraction of all crews having travel time to target 
 greater than travel time of crew A 
 
 g(t,.) = Frequency distribution of time of all crews 
 from target-in-View till hit 
 
 t^ = Time target comes in view 
 
 t~ = Time target is recognized 
 
 t, = Time target is hit. 
 

 
18 
 
 The formula weights the gunnery performance more highly than 
 
 other criterion performances and we have therefore considered 
 
 the Gunner (G) as the keyman for this criterion performance. 
 
 As indicated by Table 5, with exception of Time per Hit 
 
 and P , the average criterion scores (summed over all phases) 
 
 are uncorrelated. 
 
 Table 5 
 
 Inter cor relations of combined criteria of 
 Project STALK (N = 25 crews) 
 
 Z T/H 2 T/T P a 
 
 X T/H - -.07 .69 
 
 S T/T - .10 
 
 P. 
 
 ** 
 
 P< .01 
 
 2 T/H = Average Time per Hit 
 
 S T/T = Average Travel Time 
 
 P. = Probability of Tank A winning 
 an engagement 
 
 Operational Hypotheses 
 
 As indicated above we tested the hypothesis that ASo of 
 the formal leader will correlate negatively with the criteria in 
 crews having certain sociometric structure. 
 
 Stated as operational hypotheses, these read: 
 1. a. ASo of the Tank Commander (ASo ) will correlate 
 negatively with Time per Hit and P in those crews in which the 
 accepted Tank Commander endorses the Gunner. 
 
19 
 
 1. b. ASo Tr will correlate negatively with Time per Hit 
 
 in crews in which the Tank Commander and Gunner sociometrically 
 endorse each other. 
 
 a. ASo_ c will correlate negatively with Travel Time where 
 the accepted Tank Commander endorses the Driver. 
 
 2. b. ASo T - will correlate negatively with Travel Time in 
 crews in which the Tank Commander and Driver sociometrically 
 endorse each other. 
 
 Definition of Sociometric Endorsement. 
 
 The sociometrics required nomination of five to seven men 
 in order of preference. The average choice of another person placing 
 him among the five most preferred was used as a cutting score. 
 
 Since sociometric preferences tend to change over time, our 
 hypothesis is exactly applicable only to two phases, namely, the 
 first and the fifth phase (T/H,, T/H v ; TT,, TT v ). However, we 
 have also combined data over all phases for purposes of this 
 analysis. Each crew was ranked as best, second best, etc., to 
 poorest (5th) in its platoon for each of the five phases and the cri- 
 terion ranks were then summed over all phases to obtain combined 
 scores for both Time per Hit (2 T/H) and Travel Time (S T/T). 
 This procedure assumes that the crew is competing on equal terms 
 with other crews within the platoon. 
 
 In combining the sociometrics into a single index we defined 
 endorsement as one person's choice for another in fifth preference 
 rank or above on at least two of the three sociometrics. Similarly, 
 a Tank Commander was considered Most Preferred Crew Member 
 when he received the highest number of weighted ranks in his crew 
 in two of the three sociometrics. 
 

 
20 
 Results 
 
 The first hypothesis is related to the Time per Hit (T/H) 
 criterion and P. and was stated as two subhypotheses: (a) that 
 ASo Tr would correlate negatively with the Time per Hit criterion 
 when the accepted Tank Commander endorses the keyman, here 
 the Gunner (Table 6) and (b) where the Tank Commander and 
 Gunner choose each other. (Table 7) As can be seen from Tables 
 6 and 7, both subhypotheses la and lb are supported. 
 
 Table 6 
 
 Correlation of the Time per Hit Criterion with the Tank 
 Commander's ASo Score under a Specified Sociometric Condition 
 
 (TC = MPC ->G) 
 
 F hase subsequent 
 
 
 
 
 
 to sociometric 
 
 Criterion 
 
 Rho 
 
 N 
 
 P 
 
 I 
 
 T/H, 
 
 -.53 
 
 6 
 
 — 
 
 V 
 
 T/H y 
 
 -.51 
 
 12 
 
 .05 
 
 Combined 
 
 2T/H 
 
 -.76 
 
 1 
 
 .05 
 
 I 
 
 P A 
 
 -.37 
 
 6 
 
 - 
 
 V 
 
 P A 
 
 -.71 
 
 12 
 
 .01 
 
 Combined 
 
 P A 
 
 -.60 
 
 7 
 
 .10 
 
 Operational Hypothesis 2a stated that ASo would be 
 negatively correlated with Travel Time, when the accepted Tank 
 Commander endorses the Driver. Table 8 presents correlations 
 pertaining to the Travel Time criterion. Operational Hypothesis 
 2b could not be tested because of the insufficient number of avail- 
 able erews (N = 4). 
 
 We therefore consider the major hypothesis of the valida- 
 tion study as supported, in particular since the uncorrelated Gunnery 
 and Travel Time Criteria give consistent results. 
 

21 
 Table 7 
 
 Correlation between the Gunnery Criterion with the 
 ASo of the Tank Commander under a Specified Sociometric Condition 
 
 TC *-» G 
 
 Phase subsequent 
 to sociometric 
 
 Criterion 
 
 Rho 
 
 N 
 
 P 
 
 I 
 
 T/Hj 
 
 -.35 
 
 9 
 
 - 
 
 V 
 
 T/H y 
 
 --32 
 
 9 
 
 - 
 
 Combined 
 
 ST/H 
 
 -.61 
 
 9 
 
 .05 
 
 I 
 
 P A 
 
 -.43 
 
 9 
 
 - 
 
 V 
 
 P A 
 
 -.72 
 
 9 
 
 .02 
 
 Combined 
 
 P A 
 
 -.7? 
 
 9 
 
 .02 
 
 Table 8 
 
 Correlation of the Travel Time Criterion with Tank 
 Commanders 1 ASo Score under a Specified Sociometric Conditio* 
 
 (TC = MFC --*D) 
 
 Sociometric s 
 in Phase Criterion Rho N P 
 
 I 
 
 V 
 
 Combined 
 
 T/Tj 
 
 - 
 
 2 
 
 r/T v 
 
 -.74 
 
 6 
 
 st/t 
 
 -.33 
 
 5 
 
 Related Findings 
 
 Further analyses were confined to crews haviiag 
 accepted leaders since the leader was not accepted in only 
 eight of the tank crews which were available for validation. 
 However, groups not having accepted formal leaAert do appear 
 to differ markedly from those in which the formal leader is 
 accepted. 
 
 As we have seen above, ASo of the accepted leader who 
 endorsed his keyman correlated negati vely with group effectiveness 
 
22 
 
 criteria. In addition, Table 9 shows that ASo and effectiveness 
 correlate positively in crews in which the accepted leader does 
 not endorse his keyman. This suggests that two separate attitudes 
 of the leader toward his co-workers interact in affecting the per- 
 formance of his group" (a) the leader's generalized attitude toward 
 co-workers (ASo), and (b) his specific attitude toward his keyman 
 (measured sociometrically). 
 
 Table 9 
 
 Summary of correlations obtained from bomber and tank 
 crews in which the Leader is Most Preferred Crew Member* 
 
 
 
 
 Leade 
 
 r's 
 
 Prefsrer.ce for Keyman 
 
 Sample 
 
 Criterion 
 
 Keyman 
 
 Positive 
 
 Neutr 
 
 al 1 Negative 
 
 
 
 
 
 Rho 
 
 N 
 
 Rho 
 
 N 1 Rho N 
 
 i 
 
 
 B-29 
 
 RBS 
 
 VO/N 
 
 -.81 
 
 11 
 
 -.13 
 
 7 .42 7 
 
 B-29 
 
 %SVC 
 
 B 
 
 -.52 
 
 7 
 
 .47* 
 
 9 .27** 7 
 
 Tanks 
 
 st/h 
 
 G 
 
 -.60 
 
 6 
 
 .11 
 
 6 .60 5 
 
 Tanks 
 
 st/t 
 
 D 
 
 -.33 
 
 5 
 
 .3 9 
 
 6 .43 6 
 
 Friedman's £ test (7) indicates that this table departs signifi- 
 cantly from randomness. 
 
 ♦♦ 
 
 See Footnote, p. 13 
 
 Figure 1 presents the data of Table 9 in graphic form. 
 The correlation between the leader's generalized interpersonal 
 attitude (ASo) and his group's effectiveness is plotted along the 
 vertical axis; the leader's positive, neutral, or negative socio- 
 metric rating of his keyman (indicated by +, 0, -, respectively) 
 is plotted on the horizontal axis. The interaction between ASo 
 and sociometric preferences is readily apparent in this figure. 
 ASo and group effectiveness correlate negatively in crews in 
 which the leader endorses his keyman (points labelled "A") but 
 positively in crews where the leader does not endorse his keyman 
 (points labelled "C"). The point labelled "B" identifies 
 
' 
 
 
 
23 
 Rho's between leaders' ASo and effectiveness criteria 
 
 i 
 i- 1 
 
 i i 
 
 i i 
 
 H 
 
 
 
 = 5 
 
 ft 
 U 
 
 H 
 
 P 
 
 
 o 
 
 tj o 
 
 <V> *-i 
 
 »-*» ni 
 
 H V> 
 
 (T> rt 
 
 s ° 
 
 re & 
 
 § ^* 
 
 P »-• (T) 
 
 "o •■*» ,r 
 p 
 
 - - a 
 re 
 
 
 re 
 
 
 X 
 
 
 V 
 
 
 4 
 
 £ ~ 
 
 rt 
 
 
 en 
 
 
 u 
 
 
 re 
 
 
 0) 
 
 ro re 
 
 >C 
 
 
 
 "J C 
 
 V) 
 
 0) rt 
 
 h»» 
 
 ft 4 
 > 1— ' 
 
 ft 
 
 rt 
 < 
 
 re 
 p 
 
 re 
 
 o 
 
 Of 
 
 - o 
 re 
 
 w» 
 
 : O 
 
 re 
 _*< 
 
 - 3 
 p> 
 
 
 cr w) 
 o o 
 
 3 p 
 
 re a 
 
 H re 
 
 t" Si 
 
 5 « 
 a- o 
 
 ~ ?. 
 p> < 
 
 2L » 
 
 * t« 
 
 re w 
 S»tf 
 w Z 
 
 rt 
 
 rt 
 
 re 
 a 
 
 p 
 
 OQ 
 {0 
 
 M« 
 
 
 U> 
 rt 
 
 ►-* 
 
 re 
 P) 
 
 a. 
 re 
 4 
 
 re 
 
 re 
 •i 
 p 
 
 O 
 re 
 
 rt» 
 
 o 
 
 re 
 
 3 
 P 
 
 f ^??^ ?9?f i ?999?q 9 99 9 . 
 
 O-i 
 
 o-i 
 
 O 
 
 rt- 
 
 re 
 o 
 
 
 
 II II 
 
 
 o 
 
 
 H 
 
 
 !»•• 
 
 
 rt- 
 
 
 re 
 
 
 H 
 
 
 h»» 
 
 
 O 
 
 r 
 
 
 
 tf o 
 
 ») 
 
 ii ii 
 
 en 
 < 
 
 o 
 
 o 
 
 
 •i 
 
 
 *-•• 
 
 
 rt- 
 
 
 re 
 
 
 H 
 
 
 
 
 H 
 
 R 
 
 o 
 
 II II 
 
 n 
 
 h a 
 
 s 
 
 a 
 
 h3 
 o 
 
 O 
 
 
 H 
 
 
 H* 
 
 
 rt- 
 
 
 re 
 
 
 M 
 
 
 
 H 
 
 
 
 ttO 
 
 tl 
 
 ii n 
 
 H^g 
 
 ►4 
 
 
 CD 
 
 o 
 
 -v] On 
 O O 
 
 vrv t - Vjj ro i_j o I- 1 
 O O O O O O O 
 
 ro\jjfVn.CN-ocovo 
 OOQO O OOO 
 
 O 
 
 o 
 

^4 
 
 crews in which the leaders preference for his keyman was 
 neutral. 
 
 Discussion 
 
 The studies raise a number of problems. In addition to 
 the interaction effect presented above we will consider questions 
 concerning the leader's status in the group and concerning the 
 classification of tasks in terms of the hypotheses to which they 
 give rise* 
 
 Interaction between Sociometric Preference and ASo 
 
 The studies of bomber and tank crews have yielded re- 
 sults which indicate that effective work groups tend to have 
 accepted leaders (a) who are generally distant toward their co- 
 workers (low ASo) and have a close relationship with their key- 
 men, or (b) who have generally close relations with co-workers 
 but who are distant toward their particular keyman. 
 
 In an earlier paper dealing with informal teams we sug- 
 gested that effective leadership requires a certain distance 
 between the leader and his co-workers (6). The present studies 
 lead to the additional hypothesis that the effective leader can 
 maintain this distance in one of two ways. If he is generally 
 distant, i.e., has a tendency to perceive, and react to, persons 
 in a generally cold and reserved manner, he must compensate 
 for this feeling of distance to others by a feeling of closeness 
 
 Categorization of crews into groups in which the leader's 
 preference was positive, neutral, or negative, was based on 
 available data. The results therefore require further validation; 
 future work may indicate that more or fewer subdivisions provide 
 the most efficient utilization of data for purposes of prediction. 
 
-.. 
 
 
 
 , 
 
 
25 
 
 toward his keyman. 
 
 Multiple correlations, using the variables, leader's ASo, 
 
 sociometric preference (SP) and their product ASo x SP, yielded 
 
 R's as shown in Table 10. 
 
 Table 10 
 
 Prediction of three effectiveness criteria by the Assumed 
 Similarity and Sociometric Preference variables, singly, 
 jointly, and in combination with the Assumed Similarity x 
 Sociometric Preference Score 
 
 Criterion N ASo SP R^ 3 ^1*2 3 4* 
 (1) (2) (3) 
 
 Radar Bomb- 22 -.21 -.04 .21 .45 
 
 ing 
 
 Time per Hit 17 -.30 -.15 .32 .45 
 
 Travel Time 17 .12 .15 .20 .3 
 
 * Variable 4 is ASo x SP; these scores were obtained by multi- 
 plying the standard scores of ASo and SP 
 
 Si Table 10, neither the multiple correlations nor the incre- 
 ments of R^.2 3 4 ar e statistically significant when considered in- 
 dividually. The beta weights for the product term ASo x SP were 
 -.56, -.33, and -.24 for RBS, T/H, and T/T, respectively. The first 
 two of these weights are considerably greater than the weights for 
 ASo and SP in the RBS and T/H correlations. The third beta weight 
 for ASo x SP is equal to the weight for ASo in the T/T correlation 
 and these two weights are again considerably greater than the weight 
 for SP. This finding is related to the interpretation that we are deal- 
 ing with a single dimension of emotional distance which is a composite 
 of ASo and sociometric preference. The negative signs show that 
 an increase of the cross product is 
 
 In terms of the work by Halpin (8), Hemphill (9) and their 
 associates, these may well be the leaders who have high "considera- 
 tion" as well as initiation-of -structure scores. 
 
 The writer is indebted to Dr. C. F. Wrigley for suggesting the use 
 of the cross product term as a variable and for his assistance in 
 interpretation of the results. It may be noted that this use of the 
 product term is arithmetically identical to that of Saunders' (15 ) 
 "moderator variable" model. 
 

26 
 
 detrimental to effectiveness, thus again indicating that effective 
 group leaders tend to be persons with high ASo and low Socio- 
 metric preference for their keyman, or vice versa. 
 
 On the whole, these multiple regressions lend further 
 support to the interpretation that the effective leader's gener- 
 alized and specific attitudes toward co-workers complexly inter- 
 act, resulting in an optimum leader-keyman distance. This 
 hypothesized distance may be visualized as the sum of the dis- 
 tance engendered by a generalized attitude (ASo) and of the 
 distance resulting from a particular interpersonal attitude 
 (expressed by sociometric choice). An attempt has been made 
 to present this relationship in Figure 2. 
 
 ASo is visualized as an index of the leader's generalized 
 relations to his co-workers. The less similarity the leader 
 assumes, i.e., the lower his ASo, the less approachable will he 
 be in his relations with others. In this figure, this distance is 
 shown as the radius of the circle. The sociometric preference 
 is also visualized as a distance. The higher the leader' 3 pref- 
 erence for his keyman, the closer his psychological distance to 
 him. The less his sociometric preference, the greater the dis- 
 tance between himself and his keyman. This distance is optimal 
 in diagrams b and c. Overly distant or overly close relations 
 are detrimental to group effectiveness. The latter are diagrammed 
 in a and d, respectively. 
 
 Most large organizations implicitly or explicitly recognize 
 the importance of social and psychological distance in work rela- 
 tions, and have established elaborate rules for maintaining this 
 distance between leader and follower. Thus the armed forces 
 

27 
 
 Optimal distance between 
 leader and keyman 
 
 High ASo and high socio- 
 metric preference for 
 keyman - Poor Crew 
 (L = MPC-JK 
 
 K High ASo and low socio- 
 metric preference for 
 keyman - Good Crew 
 (L = MPC ~fT) 
 
 Low ASo and high socio 
 metric preference for 
 keyman - Good Crew 
 (L = MPC — » K) 
 
 ■>K Low ASo and low 
 
 sociometric preference for 
 keyman - Poor Crew 
 (L = MPC/->K) 
 
 Figure 2. Schematic Representation of the Interaction between 
 ASo and Sociometric Preference in Good and Poor Crews. 
 
Z8 
 separate the senior officer from the junior officer, and the 
 junior officer from his enlisted men. We may point to the 
 restricted officers' and NCO messes and washrooms, the of- 
 ficers' country aboard ship, the Admiral's deck, the periodic 
 job rotations, and many other rules which operate to restrict 
 informal contacts between leader and follower. Large industrial 
 organizations have similar, although perhaps less explicit, rules 
 in industry. We also find the executive dining room and the em- 
 ployees' cafeteria, informal sanctions against vice presidents and 
 bookkeepers playing golf together, and many others. These customs 
 and institutions, whether by design or not, operate to maintain a 
 distance between leader and followers. In light of our studies, the 
 maintenance of such a distance by means of various barriers 
 appears justified when the leader tends to have generally close 
 relations with others. These same barriers may be detrimental 
 in cases where the accepted leader naturally tends to have distant 
 relations with others: here w^ would expect poorer group perform- 
 ance since the extreme distance between the generally reserved 
 leader who has negative feelings toward his keyman may perhaps 
 make successful communication too difficult. On the other hand, 
 the approachable outgoing leader may well become emotionally 
 too involved with the keyman whom he likes, and he may therefore 
 have difficulty in making sound decisions involving his subordinate, 
 and in applying sanctions in the case of poor keyman performance. 
 
 Thus, Katz, et al. report a greater role differentiation between 
 supervisor and supervisees in high productivity than in low 
 productivity groups (13). 
 

 -'■ 
 
 
 
29 
 
 The leader's status in the group 
 
 The relations which we have found in our studies are 
 limited to groups in which the formal leader is also the group's 
 informal leader. It is probable that the informal leader status 
 provides the formal leader with considerable influence and power 
 over his crew members. Because the group supports his actions, 
 the leader* s demands can be reinforced not only by official sanc- 
 tions but also by the much more potent pressures which the group 
 itself might bring to bear on the non-conforming individual. Where 
 the leader does not have informal leader status another person may 
 perhaps take over some of his leadership functions as has been sug- 
 gested by other investigators (1, 12). It is also possible that effective 
 teams, which do not accept their leaders, utilize the keyman's skill 
 and the leadership ability of other crew members to a relatively 
 greater extent. While our data provide some leads, further research 
 is required to develop theory and methods for predicting the effect- 
 iveness of these groups. 
 
 Categorization of group tasks . 
 
 Our studies have been mainly concerned with situations 
 which seem to require positive, ie., direction-giving leadership 
 behavior. However, we found in the bomber crew study that the 
 correlation between Control Time Error and the Aircraft Com- 
 mander's ASo is in the opposite direction from that obtained in 
 other studies. While the finding may be due to chance, we are 
 inclined to believe that Control Time Error requires a receptive 
 attitude on the part of the Aircraft Commander. If this is the case, 
 
30 
 a categorization of group tasks in terms of optimum leader atti- 
 tudes would be of considerable practical and theoretical importance. 
 Such studies may well lead to new conceptions of training and place- 
 ment of potential leaders, a conclusion which is also suggested by 
 research on Naval Leadership by Shartle, Stogdill, and others 
 (16, 17). Research on this aspect of the group effectiveness problem 
 is now under way. 
 
 Summary and Conclusions 
 
 Two studies are here reported which investigate the re- 
 lation between the leader's interpersonal attitudes and the effect- 
 iveness of small military combat crews. 
 
 Two types of leader attitudes were considered; a general- 
 ized attitude toward co-workers, and the attitude toward a specific 
 crew member. The former is obtained when we ask S to predict 
 the personality test responses of the persons whom he considers 
 to be his most and least preferred co-workers. The statistical 
 comparison of these two predictions yields the interpersonal per- 
 ception score, Assumed Similarity between Opposites (ASo). 
 The second attitude measure is concerned with sociometric pref- 
 erences of the leader for a particular co-worker, namely, the 
 k«yman, or the specialist on the crew who is most directly con- 
 cerned with the criterion relevant operations. 
 
 An exploratory study using 53 B-29 bomber crews led 
 to the hypothesis that the ASo of the leader would be negatively 
 correlated with crew effectiveness criteria in groups in which 
 the accepted leader sociometrically endorsed his keyman, 
 
31 
 or in which leader and keyman mutually chose each other. 
 
 In addition, the presence of an interaction effect between the 
 leader's status, his attitude toward generalized co-workers (.aSo) 
 and his attitude toward his keyman (his sociometric rating of his 
 keyman) was found, suggesting that the psychological distance 
 between leader and keyman is related to the effectiveness of 
 the team. 
 
 a validation study was conducted on 25 Army tank crews 
 which participated in a carefully controlled weapons analysis 
 experiment. Two uncorrelated criteria were available for the 
 study. The hypothesis was supported that leaders* aSo scores 
 correlate negatively with criteria in crews of specified socio- 
 metric structure. The study also provided corroborating evidence 
 that the psychological distance between accepted leaders and key- 
 men, here defined by ASo and sociometric preference, is related 
 to effective team work. 
 
32 
 References 
 
 1. Cattell, R. B. New concepts for measuring leadership in 
 
 terms of group syntality. Hum . Relat., 1951, 4, 161-184. 
 
 2. Cronbach, L. J., and Gleser, Goldine C. Similarity between 
 
 persons and related problems of profile analysis. Cham- 
 paign-Urbana, Illinois: 1952. (Mimeographed, Technical 
 Report No. 2, Contract N6ori-07135 between the University 
 of Illinois and the Office of Naval Research.) 
 
 3. Cronbach, L. J., Hartmann, W., and Ehart, Mary E. Investi- 
 
 gation of the character and properties of assumed similarity 
 measures. Champaign-Urbana. 1953. (Mimeographed, 
 Technical Report No. 7, Contract N6ori-07135 between the 
 University of Illinois and the Office of Naval Research.) 
 
 4. Fiedler, F^ E., Hartmann, W., and Rudin, S. A. Interpersonal 
 
 perception in the prediction of effectiveness of basketball 
 teams. Champaign-Urbana, 1952. (Mineographed, Technical 
 Report No. 3, Contract N6ori-07135 between the University 
 of Illinois and the Office of Naval Research.) 
 
 5. Fiedler, F. E. Assumed similarity measures as predictors of 
 
 team effectiveness in surveying. Champaign-Urbana, 1953. 
 (Mimeographed, Technical Report No. 6, Contract N6ori- 
 07135 between the University of Illinois and the Office of 
 Naval Research.) 
 
 6. Fiedler, F. E. The psychological distance dimension in inter- 
 
 personal relations. J. Pers ., 1953, 22, 142-150. 
 
 7» Friedman, M. The use of ranks to avoid the assumption of 
 normality. £. Am . Stat. Assn ., 1937, 3_2, 675-701. 
 
 8. Halpin, A. W. The leadership behavior and combat perform- 
 ance of airplane commanders. J. abnorm . soc . Psychol ., 
 1954,49, 19-22. 
 
 9« Hemphill, J. K. and Coons, A. E. Leader Behavior Descrip- 
 tion . Columbus, Ohio, Pers. Res. Bd. Ohio State U. 1950 
 
 10. Hill, F. I., Scott, G. W., Eckles, A. J., Hardison, D. C, 
 McAuley, J. A. An assembly of Project STALK data . 
 Ballistics Research Laboratory Memorandum Report No. 
 ?45, Vol I (Classified). 
 
 11. Jennings, Helen. Leadership and isolation. ( 2nd Ed.) New 
 York: Longmans* Green, and Co, 1950. 
 
■ 
 
 
33 
 
 jl2. Kahn, R. L. and Katz, D. Leadership practices in relation 
 
 to productivity and morale, in Cartwright, D. and Zander, A., 
 Group Dynamics . Evanston, 111.: Row Peterson, 1953. 
 
 13. Katz, D., Maccoby, N., Gurin, G., Floor, Lucretia G. 
 
 Productivity supervision and morale among railroad 
 workers. Ann Arbor, Mich.: Univ. of Michigan, 1951 
 
 14. Knoell, Dorothy and Forgays, D. G. Interrelationships 
 
 of combat crew performance in the B-29 « San Antonio, 
 Tex.: Human Resources Research Center, Lackland AFB, 
 December 1952. (Research Note CCT 52-1) 
 
 15. Saunders, D. R. Moderator variables in prediction. Res. 
 
 Bulletin , RB-53-23. Educ. Testing Service, Princeton.N. J. 
 
 16. Shartle, C. L. and Stogdill, R. M. Studies in naval leadership : 
 
 methods, results, and applications. Final technical report. 
 Columbus, Ohio, Pers. Res. Bd. Ohio State U. 1953. Litho, 
 
 17. Stogdill, R. M., Wherry, R. J., and Jaynes, W. E. Patterns 
 
 of leadership behavior: A factorial study of navy officer 
 performance . Columbus, Ohio, Pers. Res. Bd. Ohio State U. 
 Mimeo* 
 
 18. Wherry, R. J. and Fryer, D. H. Buddy ratings: Popularity 
 
 contests or leadership criteria. Pers. Psychol ., 1942, 2, 
 147-149. 
 
 19. Whyte, W. F. Street Corner Society . Chicago: Univ. Press, 
 
 1943. 
 
I 
 
 ■■*.*, . ,. <«*.,*• 
 
 
/ 
 
 I 
 
 •I 
 
 « 
 
 \ 
 
 t 
 
 i 
 
 J 
 
 

 
 til 
 
 
 
 
 
 i W tj if If 
 
 
 
 1 lif tiff 1 
 
 
 iiiiidil 
 
 
 • l I ! t 
 
 
 
 
 
 
 It 
 
 111