Theory Of Coaching : Journal Review; Scrutnizing The Skipper a Leadership
ryamar1
scruntizing_the_skipper__leadership.pdf
Journal of Applied Psychology 1979, Vol. 64, No. 4, 391-400
Scrutinizing the Skipper: A Study of Leadership Behaviors in the Dugout
Bill Curtis General Electric Company
Arlington, Virginia
Ronald E. Smith and Frank L. Smoll University of Washington
The Coaching Behavior Assessment System was used by trained observers to classify the behaviors of coaches toward their players into 12 categories. Sam- ples of 51 and 31 coaches in a boys' baseball program were observed during the 1976 and 1977 seasons, respectively. Players were interviewed at the end of the season concerning perceptions of their coach's behaviors and attitudes toward their experience. Significant relationships were found at the team level among coaching behaviors as recorded by observers and perceived by players. The degree of punitiveness was an important factor in these relationships. Be- havioral data provided by observers and players also related significantly to the won-lost record and team attitudes toward both the coach and team. Per- ceptions of their own behavior by coaches were unrelated to the data provided by observers and players.
A controversial issue surrounding organized sports programs for children concerns how a coach's behavior affects the experience of players. For example, coaches are often ac- cused of overemphasizing the importance of winning to the detriment of their players' en- joyment of sports. Although the public media focuses attention on this topic, little scientific
The research culminating in this report was sup- ported in part by Grant MH-24248 from the Na- tional Institute of Mental Health to Ronald E. Smith, principal investigator. Portions of the data reported here were presented at the annual meet- ings of the American Psychological Association in San Francisco (August 1977) and Toronto (August 1978). Work from which this report was drawn was performed while Bill Curtis was a research assis- tant professor at the University of Washington.
We gratefully acknowledge the assistance and support of H. E. Pohlman (District Administrator, Washington District 8, Little League Baseball, Inc.). We also recognize the contributions of our col- league Earl Hunt in the design of this research, of our research assistants David Coppel, Samuel Clarke, and Nolan Zane in data collection and analysis, and of Allan Jones in commenting on a previous draft.
Requests for reprints should be sent to Bill Curtis, Software Management Research, General Electric-ISP, Suite 200, 1755 Jefferson Davis High- way, Arlington, Virginia 2 2 2 0 2 .
evidence is available concerning the impact of coaching behaviors on children (Smith & Smoll, 1978).
Studying interactions between coaches and players holds implications not only for the psychology of sports, but also for the field of leader-group relations. In fact, sports offer an excellent setting for studying the behaviors of leaders. Ilgen and Fujii (1976) stressed the importance of field studies relating group performance and morale to behavioral data on leaders. They suggested that 'both the greater importance of a leader's behavior to subordi- nates and the increased time for interactions may produce stronger relationships in a field setting than have been obtained in the lab- oratory.
To study coach-player interactions, Smith, Smoll, and Hunt (1977b) developed the Coaching Behavior Assessment System (CBAS). Using this system, trained observers can record the ongoing behaviors of coaches into one of 12 categories. These categories are divided into two classes, reactive behaviors (responses to a player's behavior) and spon- taneous behaviors (self-initiated responses, not requiring prior activity by a player). Coaching behaviors are distinguished partly
Copyright 1979 by the American Psychological Association, Inc. 0021-9010/79/6404-0391$00.75
391
392 B. CURTIS, R. SMITH, AND F. SMOLL
in terms of events immediately preceding the behavior because the situational context sur- rounding interactions between coaches and players presumably sets the stage for the psy- chological effect of the behavior.
In a mediational model of leadership in athletic situations, Smoll, Smith, Curtis, and Hunt (1978) proposed that the effect of coaching behaviors on team attitudes is de- termined by how players perceive these be- haviors. That is, the effects of coaching behav- iors are mediated by the meaning players attribute to them. For instance, players may respond not only to the informational content of technical instruction, but also to some message a coach unwarily communicates through the manner or timing of such instruc- tion. Thus, players' attitudes should be more closely related to their perceptions of a coach's behavior than to his actual behavior. This mediational approach is more than a recursive causal model, however, since it al- lows for reciprocal interactions among relevant variables (Heise, 1975). Significant relation- ships may indicate not only that coaching behaviors affect team performance or morale but also that the subsequent behaviors of the coach may be influenced by these outcomes.
In this study, behavioral data on Little League Baseball coaches were obtained from trained observers, players, and the coaches themselves. In addition, data were collected on won-lost records and players' attitudes toward their experience. Data were collected over two seasons, allowing a replication of the first year's results. These data also al- lowed a field replication of the results ob- tained by Ilgen and Fujii (1976) in a lab- oratory simulation concerning different sources of information on leader behavior.
Method
Sample
Coaches of Little League Baseball teams in the Seattle area were studied during the 1976 and 1977 seasons. The 1976 sample consisted of 51 male coaches who had given their permission to be ob- served during league games, and 542 of their play- ers (all boys) who were available for interviews at the close of the season. Of the 64 coaches originally contacted, 13 did not give permission to
be observed. The coaches studied had from 0 to 17 years of coaching experience (M = 6.8, SZ) = 4.3). The S42 boys (8-15 years of age) constituted 83% of the players on the 51 teams. Most of the un- interviewed boys were on vacation or had moved after the season. Fewer than \% were not given permission by their parents to be interviewed. The 51 teams were drawn from three levels of compe- tition separated by age: minors (8-9 years), ma- jors (10-12 years), and seniors (13-15 years).
The 1977 sample was limited to teams in the majors and seniors divisions. All 31 coaches from the leagues selected were observed, 19 of whom had been included in 1976 sample. These coaches had from 0 to 22 years of coaching experience (M = 8.3, SD-6A). From these 31 teams, 325 boys were interviewed at the close of the season (82% of all players).
Coaching Behavior Assessment System
The CBAS, designed by Smith et al. ( 1 9 7 7 b ) , is a data collection technique that permits the classification of coaching behaviors into 12 cate- gories. These categories are divided between reac- tive behaviors (responses to good performance, mistakes, or misbehaviors) and spontaneous be- haviors (self-initiated responses that may be rele- vant or irrelevant to the game). The organization of the 12 CBAS behaviors is presented in Table 1.
Undergraduate research assistants (31 in 1976 and 18 in 1977) completed an audiovisual training sequence (Smith, Smoll, Hunt, & Clarke, 1976) that taught them how to collect CBAS data as observers. A subsequent reliability study of the CBAS, using 19 of the 1976 observers, yielded an average inter- observer reliability of .88 (Smith, Smoll, & Hunt, 1977a). In using the CBAS, observers positioned themselves behind the team bench, where the coach could be heard and seen clearly. From the opening pitch to the final out, the observer recorded every behavior the coach directed toward players into one of the 12 CBAS categories.
Each coach was observed for from three to six games, with approximately 280 behaviors being coded during each game. A behavioral profile was formulated for each coach as follows: (a) for each game, the frequency of each behavioral category was divided by the total number of behaviors coded for that game, yielding a proportion for each category, and (b) the proportions for each category were then averaged across games to create the average proportion of behaviors belonging to each category. These proportions are reported as per- centages.
Player Perceptions and Attitudes
At the conclusion of each season, trained under- graduate research assistants who had not partici- pated as observers went to the home of each player
COACHING BEHAVIORS 393
Table 1 The Coaching Behavior Assessment System
Behavior Definition
Responses to desirable performance Reinforcement
Nonreinforcement Responses to mistakes
Encouragement after a mistake Technical instruction after a mistake Punishment Punitive technical instruction
Ignoring a mistake Response to misbehavior
Keeping control
Reactive behaviors
A positive, rewarding reaction (verbal or nonverbal) to a good play or good effort
Failure to respond to a good performance
A positive, consoling reaction to a player following a mistake Instructing a player how to correct a mistake he has just made A negative reaction, verbal or nonverbal, following a mistake Technical instruction, following a mistake, that is given in a
harsh or sarcastic manner Failure to respond to a player mistake
Reactions intended to restore or maintain order among team members
Game related General technical instruction
General encouragement Organization
Game irrelevant General communication
Spontaneous behaviors
Spontaneous instruction in the techniques and strategies of the sport (not following a mistake)
Spontaneous encouragement that does not follow a mistake Administrative behavior that sets the stage for play by as-
signing duties, responsibilities, positions, etc.
Interactions with players unrelated to the game
and conducted a structured interview. The inter- view began by giving the player a description and examples of each of the 12 CBAS behavioral cate- gories. The player was then asked to indicate how often his coach engaged in each of the 12 behav- iors on a scale from 1 (never) to 7 (almost al- ways). Later in the interview, the player answered 10 questions about his attitudes toward his coach, team, and baseball, again measured on a 7-point scale (least favorable to most favorable).
Scores from the 10 attitudinal items measured during 1976 were entered into a factor analysis (Smith, Smoll, & Curtis, 1978), and two factors emerged reflecting players' feelings toward the coach and toward the team. Based on this analysis, attitudinal criteria were developed by summing intercorrelated items into two composites: (a) attitude toward the coach—How much did you like playing for your coach? How much would you like having the same coach again next year? How much do you like your coach? How much do your par- ents like your coach? How much does your coach like you? How much does your coach know about baseball?, (b) attitude toward the team—How well did the players on your team get along? How well did you like the other players on your team? Lower bound estimates of internal consistency (Cronbach's alpha) for these two composites were .94 and .86, respectively.
Based on the 1976 results, a reduced set of items was chosen for use in 1977. Attitudes toward the coach were measured by Items 1, 2, 6, and a new item concerning how good the coach was at teach- ing baseball. Attitudes toward the team were mea- sured by Item 7.
Team perceptions of the 12 CBAS behaviors and team attraction toward each coach and team were computed as average scores on the appropriate measures across members of each team. In aggre- gating data to the team level, we are not trying to minimize the importance of studying the effect of coaching behavior on individual players. How- ever, limitations inherent in this field study re- stricted our ability to collect data at the individual player level. That is, the speed with which coaching behaviors had to be recorded limited the observer's ability to identify and record the player toward which the behavior was directed, and many be- haviors were directed toward groups of players.
Coach Selj'-Perceptions
As part of a larger questionnaire on coaching philosophy, coaches were asked to indicate how frequently they engaged in each of the 12 CBAS behaviors. These responses were made on the same 7-point scale used by the players (never to almost always).
394 B. CURTIS, R. SMITH, AND F. SMOLL
Analysis
The initial analyses investigated the extent to which coaching behaviors were stable over time, and whether observers, players, and coaches agreed on their perceptions of these behaviors. The next phase involved correlating observed behaviors and ratings of the 12 CBAS categories by players and coaches with team attitudes toward the coach and team and the won-lost record. In addition, multiple regressions were performed to determine the extent to which these criteria were related to behavioral data provided by observers, players, and coaches. The multiple correlations reported represent step- wise results in which only variables contributing significantly to the overall prediction (significant b weights) were entered into the analysis. Squared multiple correlations, adjusted to account for the number of cases and predictors (Theil, 1971), are also reported.
Results
Analyses of Behavioral Data
Means. Data presented in Table 2 indicate that the behaviors most frequently observed across all coaches during both years were gen-
eral technical instruction, general encourage- ment, and reinforcement. The least frequently observed behaviors were keeping control, pun- ishment, and punitive technical instruction, each accounting for no more than 2% of total game behaviors. The mean profiles of coach- ing behaviors provided by players and coaches were similar to each other across both years. Both coaches and players perceived coaches as engaging in punishment, punitive technical instruction, nonreinforcement, and ignoring mistakes less often than other behaviors. Means for these four behaviors typically fell below 3.6 on a 7-point 'scale compared to means above 4.1 for other behaviors.
Consistency. Table 2 also presents corre- lations between 1976 and 1977 observed, player-perceived, and self-rated behavioral data for the 19 coaches who were studied during both years. Significant correlations were obtained on 7 of the 12 observed cate- gories and were largest for punitive behaviors and general encouragement. Fewer significant correlations occurred between 1976 and 1977
Table 2 Means for Observed, Player-Perceived, and Self-Rated Behaviors and Correlations Between 1976 and 1977 Results for Coaches Appearing in Both Samples
Observed
M
Behavior
Reinforcement Nonreinforcement Encouragement after
mistakes Technical instruction
after mistakes Punishment Punitive technical
instruction Ignoring mistakes Keeping control General technical
instruction General encouragement Organization General communication
1976 (51)
17.1 4.2
3.1
4.2 1.8
1.0 3.7 1.7
27.4 21.3 8.4 6.1
1977 (31)
23.7 3.1
3.9
3.3 1.6
.8 1.7 1.3
22.7 30.8 4.8 2.8
r (19)
.36
.50*
.40*
.50*
.64**
.68**
.09
.37
.44*
.73**
.32
.11
Player perceived
M
1976 (51)
5.31 2.71
5.24
5.32 2.51
2.71 2.89 5.37
5.33 5.67 5.63 4.68
1977 (31)
5.53 2.65
5.56
4.91 2.29
2.41 2.80 5.11
4.84 6.04 5.71 4.32
r (19)
.26 -.19
.44*
.35
.76**
.65**
.23 -.02
.11
.38
.14
.35
Self-rated
M
1976 (50)
6.56 1'.96
5.82
5.90 3.00
4.08 3.28 5.78
6.02 6.46 5.90 4.74
1977 (30)
6.07 1.93
5.10
6.23 3.53
3.60 2.77 5.50
5.50 4.93 5.70 4.17
r (19)
.14
.52*
.32
.42*
.46*
.58** -.16
.27
-.14 .37 .33 .09
Note. Means for observed behaviors represent percentages, whereas means for player-perceived and self- rated behaviors represent responses on a 7-point scale. Numbers in parentheses are ns. *p < .05.
**p< .01.
COACHING BEHAVIORS 395
Table 3 Correlations Among Observers', Coaches', and Players' Behavioral Data
Observers
Behavior
Reinforcement Nonreinforcement Encouragement after
mistakes Technical instruction
after mistakes Punishment Punitive technical
instruction Ignoring mistakes Keeping control General technical
instruction General encouragement Organization General communication
and
1976 (SI)
.38**
.14
.37**
.31*
.54**
.45**
.06 -.08
-.03 .05
-.11 .26*
players
1977 (31)
.29
.12
-.07
.03
.59**
.46**
.28 -.06
-.30* .21
-.06 .26
Observers and
1976 (50)
.19
.21
.01
.19
.45**
.13 -.02 -.02
.05
.03 -.07
.10
coaches
1977 (30)
.13
.02
.35*
.01
.17
.27 -.08
.05
.21
.12 -.07 -.03
Players and
1976 (50)
-.04 .20
.23
.06
.26*
.06
.23
.20
.01
.09
.22
.14
coaches
1977 (30)
.00
.13
.13
.18
.53**
.25
.23
.28
-.05 .09 .13 .31*
Note. Numbers in parentheses are ns. * p < .05.
**p < .01.
Table 4 Correlations of Behavioral Categories With Attitude Toward the Coach
Behavior
Reinforcement Nonreinforcement Encouragement after
mistakes Technical instruction
after mistakes Punishment Punitive technical
instruction Ignoring mistakes Keeping control General technical
instruction General encouragement Organization General communication
R Adjusted K>
Observed
1976 (51)
.20
.09
.08
.08 -.02
-.10 .01
-.31*
.16 -.11
.12 -.47'**
.47**
.22
behavior
1977 (31)
.05
.05
-.13
.10 -.26
-.21 .18
-.33"*
-.27 .22
-.08 .32'* .46* .16
Team perceptions
1976 (51)
.17 -.07
.33*
.31* -.27
-.40"** .16 .21
.42"**
.39**
.28* -.03
.54**
.30
1977 (31)
.43** -.48**
.68**
.41** -.64'**
-.61** -.25
.16
.52**
.64"**
.70"**
.47**
.89**
.76
Coach self-rating
1976 (SO)
.00
.01
.13
-.06 -.02
-.20 -.19 -.02
-.02 -.14
.05 ,29»* .29* .08
1977 (30)
.13
.19
.20
.03 -.14
.05
.05 -.17
.30
.02
.02
.19
Note. Numbers in parentheses are ns. • This category was entered into the multiple regression analysis. * p < .05.
**p < .01.
396 B. CURTIS, R. SMITH, AND F. SMOLL
scores on player-perceived and self-rated coaching behaviors, and these correlations oc- curred primarily for punitive behaviors. Thus, although there was some consistency in the frequency of observed behaviors across years, players and coaches were most aware of this consistency in punitive behaviors.
Intersource agreement. Correlations repre- senting intersource agreement on each of the 12 behaviors among observers, players, and coaches are presented in Table 3. During 1976 there was agreement on 6 of the 12 behavioral categories in the data provided by observers and players. These categories involved pri- marily reinforcement and reactions to mis- takes. During 1977, however, this agreement was limited primarily to punitive behaviors. With the exception of punishment, there was little agreement between perceptions of their own behavior by coaches and the data pro- vided by observers and players.
Relationships With Criteria
Attitude toward the coach. Correlations of the team's attitude toward the coach with
observed behaviors, team perceptions, and self- ratings by the coach on the 12 CBAS cate- gories are presented for both samples in Table 4. Adjusted R2 values indicated that the largest percentage of variance in team attitude toward the coach was related to team percep- tions of behavior, especially during 1977 ( 7 6 % ) . During 1977, correlations for team perceptions of all but two CBAS categories achieved significance. Among observer data, keeping control correlated negatively with at- titudes toward the coach during both years. Although significant relationships were also obtained for observed general communication during both years, the direction of the rela- tionship differed from one year to the next. Inspection of scatterplots indicated that the 1976 result was accounted for entirely by two coaches in the youngest league with extreme scores on general communication. Little rela- tionship was obtained for coach self-ratings during either year.
Attitude toward the team. Table 5 pre- sents correlations of intrateam attraction with the behavioral data on coaches for both sam-
Table 5 Correlations of Behavioral Categories With Attitudes Toward the Team
Observed behavior Team perceptions
Note. Numbers in parentheses are ns. » This category was entered into the multiple regression analysis. *p < .05. **p < .01.
Coach self-rating
Behavior
Reinforcement Nonreinforcement Encouragement after
mistakes Technical instruction
after mistakes Punishment Punitive technical
instruction Ignoring mistakes Keeping control General technical
instruction General encouragement Organization General communication
R Adjusted R1
1976 (51)
.30*
.07
.07
.07
.04
-.14 .03
-.23*
.01 -.09
.15 -.43"**
.43**
.18
1977 (31)
.08 -.05
-.29
-.27 -.34*
_.44a**
.07 -.59"**
-.09 .26"
-.31* .44"** .85** .68
1976 (51)
.17 -.10
.04
.12 -.23
-.33* -.08
.15
.15
.43"**
.07 -.12
.43**
.18
1977 (31)
.44** -.47**
.69"**
.21 -.54**
-.45** .01* .12«
.37*
.55"**
.46**
.58"**
.90**
.77
1976 (50)
-.02 -.02
.01
.07 -.08
-.26 -.02 -.04
-.02 -.11
.08
.05
1977 (30)
.14 -.08
.15
-.07 -.12
-.10 .00
-.29
.37'* -.06 -.02
.15
.37*
.14
COACHING BEHAVIORS 397
Table 6 Correlations of Behavioral Categories With the Won-Lost Record
Observed behavior Team perceptions
Note. Numbers in parentheses are ns. » This category was entered into the multiple regression analysis. *p < .05. ** p < .01.
Coach self-rating
Behavior
Reinforcement Nonreinforcennent Encouragement after
mistakes Technical instruction
after mistakes Punishment Punitive technical
instruction Ignoring mistakes Keeping control General technical
instruction General encouragement Organization General communication
R Adjusted R2
1976 (51)
.08
.27"*
-.09
-.14 -.35'**
-.20 -.08" -.01
.06 -.11
.02
.06
.51**
.26
1977 (31)
-.01 .07
-.15
-.31"* -.37*
-.32* -.32* -.42"**
-.19 .38*
-.24 .16 .52* .22
1976 (51)
.44"** -.34**
.40**
.29* -.42"**
-.31* -.08
.08
.18
.36**
.20
.09
.52**
.27
1977 (31)
.18 -.17
.34**
.14 -.53s**
-.45** -.12
.09
.20
.35*
.37*
.51"**
.63**
.35
1976 (SO)
.15 -.02
.33"**
.17 -.32*
.19 -.04
.07
.11
.02
.02
.07
.33**
.11
1977 {30)
.11
.11
.20
.02 -.30
-.30 .04
-.04
.16 -.11 -.01
.14
pies. These results were similar to those pre- sented for team attitude toward the coach. That is, significant correlations were obtained for player ratings of most CBAS categories during 1977. The 1976 results, however, were not as strong. Among observed behaviors, the correlations for keeping control and general communication with intrateam attraction were almost identical to those obtained with team attitude toward the coach. During 1977, how- ever, additional negative relationships were obtained for punitive behaviors and organiza- tion. Adjusted R- values indicated that during 1977, substantial variance in intrateam attrac- tion was related to both observed behaviors and team perceptions (68% and 77%, respec- tively). Results for coach self-ratings were generally weak.
Team performance. Correlations of the team's won-lost record with behavioral data for both samples are presented in Table 6. Coaches of losing teams in the 1977 sample were observed engaging in proportionately more reactions to player mistakes and mis- behaviors than were winning coaches. As seen
by their players, coaches of losing teams were more punitive toward mistakes, whereas coaches of winning teams were more reinforc- ing and engaged in more spontaneous behav- iors. Adjusted R2 values indicated that slightly more variance in won-lost records was related to team perceptions than to ob- served behavior. Again, correlations for coach self-ratings were weak.
Moderate similarities in the correlations of behavioral data both with won-lost records and team attitudes suggested a possible rela- tionship among these criteria. As is apparent
Table 7 Correlations of the Won-Lost Record With Attitudes Toward the Coach and Team
Group Coach Team
1976 sample8
Minors'1
Majors and seniors" 1977 sample11
.26*
.15
.40*
.50**
.15
.07
.28
.59**
•» = 51. b n = 23. "r *p < .05. **p< .01.
28. d w = 31.
398 B. CURTIS, R. SMITH, AND F. SMOLL
in Table 7, there was a modest relationship between the won-lost record and team atti- tudes toward the coach in the 1976 sample, but stronger relationships were evident in the 1977 results. When the 1976 sample was separated into divisions, the won-lost record was correlated with team attitude toward the coach in the older major and senior divisions, but not in the minors. Although relationships among criteria differed across divisions in the 1976 sample, few differences occurred across divisions in relationships between behavioral data and the criteria.
Discussion
The data reported here demonstrated and replicated relationships between data on coaching behaviors and team performance and morale. Some of these relationships replicated the laboratory findings of Ilgen and Fujii (1976) relating behavioral data on leaders to subordinate perceptions, attitudes, and per- formance. Within these data it was also pos- sible to identify which aspects of a coach's behavior were most salient to the criteria studied.
The strongest agreement on assessments of coaching behavior among observers, players, and coaches concerned punitive behaviors. Although punishment and punitive technical instruction accounted for only 3% of the coaching behavior observed, players seemed especially sensitive to punitiveness in their perceptions of a coach. Agreement on other behaviors among the three sources of behav- ioral data was weak and inconsistent. In par- ticular, coaches were found to be no better at judging their own behavior than were leaders in other settings (Stogdill, 1974). Research by Graen and his associates (Dansereau, Graen, & Haga, 1975; Graen, Dansereau, & Minami, 1972; Graen & Schiemann, 1978) suggests that greater agreement among sources might have been achieved had it been possible to determine the behavior directed toward in- dividual players, since coaches may have be- haved differently toward each player. These relationships between coaching behaviors and team perceptions represent the first link in determining how a coach affects team morale.
The second link in the mediational model of Smoll et al. (1978) predicted that team attitudes toward the coach would be more highly related to perceptions of his behavior than to his actual behavior. Assuming that the data provided by observers is our best ap- proximation of actual coaching behavior, this prediction was supported and replicated. In both samples, team perceptions of most be- havioral categories were related to attitudes toward the coach, whereas observed behavior in only two CBAS categories exhibited such relationships.
On the other hand, the prediction of the mediational model in accounting for players' attitudes toward their team was not strongly supported. The direct relationships of atti- tudes toward the team with observed behav- iors were about as strong as those with team perceptions of behavior. It was curious that observed coaching behaviors were more strongly related to attitudes toward the team than to those toward the coach during 1977. However, the observed behavior that related most strongly to intrateam attraction was keeping control, and its relative frequency may provide one index of team harmony.
Data on coaching behaviors from observers and players were also related to team per- formance. Although observers may have been aware of the score during a game, there was little evidence to suggest that this knowledge affected their observations. When compared to winning coaches, proportionately more of the observed behaviors of coaches on losing teams were reactions to player mistakes and misbehaviors. This result is not surprising, since players on losing teams usually make more mistakes. Furthermore, players perceived coaches of losing teams as more punitive and less supportive than winning coaches. Team performance was related to team attitudes pri- marily in the older age groups.
The generally stronger correlations observed in the 1977 replication may have resulted from several factors. First, the 1976 results may have been conservative, since 13 coaches chose not to participate, resulting in a possible restriction of range on some coaching behav- iors. Anecdotal evidence suggested that at least some of these 13 coaches were more
COACHING BEHAVIORS 399
punitive than those who gave permission to be observed. Second, experience gained during the 1976 study led to improved selection and training of observers and interviewers, result- ing in more reliable data during 1977. Finally, it is possible that significant relationships among these measures were more easily ob- tained with the older players studied in 1977.
At a descriptive level, -it is reassuring to note that the majority of boys interviewed had very positive attitudes toward their coach and team. It is also refreshing to find that contrary to some stereotypes, observers and players perceived punitiveness relatively in- frequently in the overall stream of coaching behavior. One caution in interpreting these data, however, is that they represent only re- sults based on boys 8-15 years old playing for male coaches in an urban baseball program. Additional data need to be collected from other sports, at other levels of competition, in other sociocultural settings, and from girls' programs to determine the generalizability of these results. Further research also needs to address how these relationships differ across players as a function of their individual char- acteristics.
Nevertheless, the findings reported here are valuable in applying the principles and tech- niques of behavioral science to improving organized athletic programs, especially those for children. These data suggest that some improvement in team performance or morale might result from training coaches how to interact more effectively with players, in ad- dition to instruction on conditioning and athletic techniques (Smith, Smoll, & Curtis, 1979). Focusing on the training of coaches, however, implies that their behavior is a causal factor in team performance and morale.
The question of causality is of paramount importance both in coaching and in leadership theory more generally. Beyond the issue of training, the cyclic firing of professional coaches stems from an assumption that they are the primary causal factor in team per- formance. Yet research by Barrow (1976), Farris and Lim (1969), Haythorn, Couch, Haefner, Langham, and Carter (1956), and Lowin and Craig (1968) indicated that the characteristics and performance of subordi-
nates can, in turn, exert a causal influence on the behavior of a leader. For instance, al- though punitiveness and the handling of mis- behaviors may affect players' attitudes (Keller & Szilagyi, 1978), poor performance by play- ers or team disharmony may give coaches more opportunities to display such behaviors.
Unfortunately, causality cannot be tested directly in these data. An attempt to establish causal relationships between actual coaching behaviors and team performance or morale might best proceed by analyzing the reciprocal relationships (James & Singh, 1978) among selected variables. As an outgrowth of this research, Smoll et al. (1978) have presented a model that identifies interactions and feed- back cycles between variables studied here and such additional factors as coaching phi- losophy and players' individual characteristics. The validation of such dynamic, interactional models will be an important step both in sports psychology and in leadership research.
References
Barrow, J. C. Worker performance and task com- plexity as causal determinants of leader behav- ior style and flexibility. Journal of Applied Psy- chology, 1976, 61, 433-440.
Dansereau, F., Jr., Graen, G., & Haga, W. J. A vertical dyad linkage approach to leadership within formal organizations: A longitudinal in- vestigation of the role-making process. Organi- zational Behavior and Human Performance, 1975, 13, 46-78.
Farris, G. F., & Lim, F. G. Effects of performance on leadership, cohesiveness, influence, satisfaction, and subsequent performance. Journal of Applied Psychology, 1969, 53, 490-497.
Graen, G., Dansereau, F., Jr., & Minami, T. Dys- functional leadership styles. Organizational Be- havior and Human Performance, 1972, 7, 216-236.
Graen, G., Schiemann, W. Leader-member agree- ment: A vertical dyad linkage approach. Journal of Applied Psychology, 1978, 63, 206-212.
Haythorn, W. W., Couch, A., Haefner, D., Lang- ham, P., & Carter, L. F. The effects of varying combinations of authoritarian and equalitarian leaders and followers. Journal of Abnormal and Social Psychology, 19S6, S3, 210-219.
Heise, D. R. Causal analysis. New York: Wiley, 19 75.
Ilgen, D. R., & Fujii, D. S. An investigation of the validity of leader behavior descriptions ob- tained from subordinates. Journal of Applied Psychology, 1976, 61, 642-651.
400 B. CURTIS, R. SMITH, AND F. SMOLL
James, L. R., & Singh, B. K. An introduction to the logic, assumptions, and basic analytic proce- dures of two-stage least squares. Psychological Bulletin, 1978, 85, 1104-1122.
Keller, R. T., & Szilagyi, A. D. A longitudinal study of leader reward behavior, subordinate expec- tancies, and satisfaction. Personnel Psychology, 1978, 31, 119-129.
Lowin, A., & Craig, J. The influence of level of per- formance on managerial style: An experimental object-lesson in the ambiguity of correlational data. Organizational Behavior and Human Perfor- mance, 1968, 3, 440-458.
Smith, R. E., & Smoll, F. L. Sports and the child: Conceptual and research perspectives. In F. L. Smoll & R. E. Smith (Eds.), Psychological per- spectives in youth sports. New York: Hemi- sphere, 1978.
Smith, R. E., Smoll, F. L., & Curtis, B. Coaching behaviors in little league baseball. In F. L. Smoll & R. E. Smith (Eds.), Psychological perspectives in youth sports. New York: Hemisphere, 1978.
Smith, R. E., Smoll, F. L., & Curtis, B. Coach
effectiveness training: A cognitive-behavioral ap- proach to enhancing relationship skills in youth sport coaches. Journal of Sports Psychology, 1979, 1, 59-75.
Smith, R. E., Smoll, F. L., & Hunt, E. A system for the behavioral assessment of coaches. Re- search Quarterly, 1977, 48, 401-407. (a)
Smith, R. E., Smoll, F. L., & Hunt, E. B. Training manual for the coaching behavior assessment sys- tem. JSAS Catalog of Selected Documents in Psychology, 1977, 7, 2. (Ms. No. 1406) (b)
Smith, R. E., Smoll, F. L., Hunt, E. B., & Clarke, S. J. CBAS audio-visual training module. Seattle: University of Washington, 1976. (Film)
Smoll, F. L., Smith, R. E., Curtis, B., & Hunt, E. Toward a mediational model of coach-player re- lationships. Research Quarterly, 1978, 49, 528-541.
Stogdill, R. M. Handbook of leadership. New York: Free Press, 1974.
Theil, H. Principles of econometrics. New York: Wiley, 1971.
Received December 21, 1978 •
