the relationship between perceived coaching behaviors and team cohesion ... · pdf filethe...

15
The Sport Psychologist, 1996,10,367-381 O 1996 Human Kinetics Publishers, Inc. The Relationship Between Perceived Coaching Behaviors and Team Cohesion Among Baseball and Softball Players Douglas E. Gardner Boston University David L. Light Shields John F. Kennedy University Brenda Jo Light Bredemeier University of California at Berkeley Alan Bostrom University of California-San Francisco The relationship between perceived leadership behaviors and team cohesion in high school and junior college baseball and softball teams was researched. Study participants, 307 athletes representing 23 teams, responded to the per- ceived version of the Leadership Scale for Sports (LSS) and the Group Envi- ronment Questionnaire (GEQ). Correlational and multivariate analyses indi- cated significant relationships between perceived leader behaviors and team cohesion. Specifically, coaches who were perceived as high in training and instruction, democratic behavior, social support, and positive feedback, and low in autocratic behavior, had teams that were more cohesive. A MANOVA indicated there were significant differences between genders and athletes at the two school levels in their perceptions of coaching behaviors and team cohesion, though these demographic variables did not significantly moderate the leadership-cohesion relationship. Douglas E. Gardner is with the School of Education at Boston University, Boston, MA 02215. David L. Light Shields is with the Graduate School of Professional Psychology at John F. Kennedy University, Orinda, CA 94563. Brenda Jo Light Bredemeier is with the Department of Human Biodynamics at the University of California at Berkeley, CA 94720. Alan Bostrom is with the Center for AIDS Prevention Studies at the University of Califor- nia-San Francisco, CA 94143.

Upload: phamtruc

Post on 18-Mar-2018

233 views

Category:

Documents


6 download

TRANSCRIPT

The Sport Psychologist, 1996,10,367-381 O 1996 Human Kinetics Publishers, Inc.

The Relationship Between Perceived Coaching Behaviors and Team Cohesion

Among Baseball and Softball Players

Douglas E. Gardner Boston University

David L. Light Shields John F. Kennedy University

Brenda Jo Light Bredemeier University of California at Berkeley

Alan Bostrom University of California-San Francisco

The relationship between perceived leadership behaviors and team cohesion in high school and junior college baseball and softball teams was researched. Study participants, 307 athletes representing 23 teams, responded to the per- ceived version of the Leadership Scale for Sports (LSS) and the Group Envi- ronment Questionnaire (GEQ). Correlational and multivariate analyses indi- cated significant relationships between perceived leader behaviors and team cohesion. Specifically, coaches who were perceived as high in training and instruction, democratic behavior, social support, and positive feedback, and low in autocratic behavior, had teams that were more cohesive. A MANOVA indicated there were significant differences between genders and athletes at the two school levels in their perceptions of coaching behaviors and team cohesion, though these demographic variables did not significantly moderate the leadership-cohesion relationship.

Douglas E. Gardner is with the School of Education at Boston University, Boston, MA 02215. David L. Light Shields is with the Graduate School of Professional Psychology at John F. Kennedy University, Orinda, CA 94563. Brenda Jo Light Bredemeier is with the Department of Human Biodynamics at the University of California at Berkeley, CA 94720. Alan Bostrom is with the Center for AIDS Prevention Studies at the University of Califor- nia-San Francisco, CA 94143.

368 Gardner, Shields, Bredemeier, and Bostrom

It is generally accepted by sports scientists that the personality and leader- ship style of a coach has an important influence on both the team and its individual athletes (Terry, 1984). Despite this prevailing belief in the effects coaches have on their athletes, very little empirical research has put hypothesized relationships to the test. In this study, we focused on the relationship between coaches' leadership behaviors and one important potential correlate-team cohesion.

Chelladurai and Carron's (1978) Multidiwensional Model of Leadership (MML) provides a sport-specific framework for studying the possible influences of leadership behaviors on such variables as athletic performance, personal satis- faction, and team cohesion. The MML's major proposition is that the performance and satisfaction of the athlete are a function of the congruence between three inter- acting leadership aspects: (a) actual leader behavior, (b) athletes' preferred leader behavior, and (c) required leader behavior (as determined by situational demands). It has been argued that this line of thinking also should hold true for the effect of coaching behaviors on team cohesion (Westre & Weiss, 1991).

Evolving from the Multidimensional Model of Leadership was a sport-spe- cific questionnaire called the Leadership Scale for Sports (LSS) (Chelladurai & Saleh, 1980). The five LSS scales consist of one instructional behavior (training and instruction), two decision-making styles (autocratic and democratic behav- ior), and two motivational tendencies (social support and positive feedback). The LSS can be administered in three different versions: (a) a coaches' version in which the coach self-describes his or her behavior, (b) a perceived version in which ath- letes describe the behaviors of their coach, and (c) a preferred version in which athletes describe the types of coaching behaviors they desire.

Previous sport-leadership research has suggested that athletes' perceptions of leader behavior have differed by player ability (Garland & Barry, 1988) and by overall team success (Gordon, 1986). With the preferred version of the LSS, com- petitive level differences also have been found (Carron & Chelladurai, 1983; Serpa, 1990). Oddly, neither school level nor gender have been studied in relation to the perceived version of the LSS, despite theoretical and empirical reasons to believe they may have an influence. For example, the professionalization of values litera- ture indicates that females and athletes at lower competitive levels are less on- ented toward a win orientation than males or athletes at more elite levels (Blair, 1985; Card, 1981; Kidd & Woodman, 1975; Gill, 1986, 1988; Knoppers, 1985; Maloney & Petrie, 1972; Nixon, 1980; Sage, 1980). If this is the case, female and younger athletes may perceive coaches, who exhibit behaviors premised on a pro- fessional orientation, differently than their male and more elite counterparts.

In terms of potential leadership correlates or outcome variables, the LSS has been used to examine relationships among leader behaviors and athlete satisfac- tion andlor performance (Chelladurai, 1984; Chelladurai, Imamura, Yamaguchi, Oinuma, & Miyauchi,1988; Home & Carron, 1985; Schliesman, 1987; Serpa, Pataco, & Santos, 1991). Recently, it also has been used to study the relationship between leadership behaviors and team cohesion (Pease & Kozub, 1993; Westre & Weiss, 1991).

Carron (1982) proposed an influential definition of group cohesion that im- plied its multidimensional character, and he offered a model of the antecedents and consequences of group cohesion. Four categories of antecedents were identi- fied: (a) leadership, (b) situational, (c) personal, and (d) team factors. The two identified categories of consequences were individual and group outcomes.

Coaching Behaviors and Team Cohesion 369

The definition offered by Carron (1982) was elaborated into a conceptual model and given operational specificity in terms of an instrument to measure team cohesion-the Group Environment Questionnaire (GEQ) by Carron, Widmeyer, and Brawley (1985). In the Carron et al. (1985) approach, team cohesion is con- ceived as multidimensional in nature with distinctions drawn between individual attraction to the group and group integration, and a further distinction within each of these two categories between the task and social aspects of group affiliation.

There is relatively little information available about possible group differ- ences in GEQ responding. Granito and Rainey (1988) reported differences be- tween high school and collegiate level football players, while Westre and Weiss (1991) reported differences in perceptions of cohesion due to starting status. No studies have directly compared male and female respondents, despite evidence that males and females may have different orientations with regard to connection/ separation issues (Chodorow, 1978; Gilligan, 1982). As with leadership research, there is a lack of empirical data pertaining to potential group differences in percep- tions of group cohesion.

Research using the GEQ has been concerned with such potential group co- hesion correlates as group resistance (Brawley, Canon, & Widmeyer, 1988); group size (Widmeyer, Brawley, & Carron, 1992); adherence to physical activity (Carron, Widmeyer, & Brawley, 1988); and group performance (Williams & Widmeyer, 1991). As previously noted, there also has been a growing interest in the relation- ship between leadership behaviors and team cohesion.

The importance of studying the relationship between leadership behaviors and group cohesion rests on two premises. First, group cohesion is an important influence on group performance. Though the relationship between cohesion and performance is complex, there is quite ample empirical research to demonstrate a link (see Widmeyer, Carron, & Brawley, 1993, for a review). Second, of the ante- cedents to group cohesion, leadership is perhaps the most important because, as Williams (1993) noted, "the coach is probably in the best position to influence change [in team cohesion]" (p. 120). Nonetheless, the relationship between lead- ership factors and team cohesion has received little empirical attention.

Westre and Weiss (1991) were the first to study the leadership-cohesion re- lationship using the LSS and GEQ. Six high school varsity football teams (N = 182) completed the perceived version of the LSS, the GEQ, and additional mea- sures to determine perceived team and individual success, playing status, and po- sition played. It was predicted that "higher levels of task cohesion would be re- lated to the leader-behavior dimensions of training and instruction and autocratic style" (p. 51), and, secondly, that "higher levels of social cohesion would be re- lated to the leader dimensions of democratic style, social-support behaviors, and rewarding behaviors" (p. 51). As predicted, coaches perceived by their athletes as exhibiting more training and instruction also perceived higher team-task cohesion. The hypothesized relationship between cohesion and autocratic behavior could not be tested due to an unacceptably low reliability for the autocratic scale, a prob- lem that plagues the LSS literature (cf. Chelladurai, 1993). Though it was not predicted, a positive relationship was found between task cohesiveness and the LSS dimensions of positive feedback, social support, and democratic style.

Pease and Kozub (1994) replicated the Westre and Weiss (1991) study with 10 high school female basketball teams (N = 94). Findings from this study re- vealed an overall significant relationship between coaching behaviors and the two

370 Gardner, Shields, Bredemeier, and Bostrom

task dimensions of team cohesion (Pease & Kozub, 1994). Specifically, perceived training and instruction and democratic behavior were significantly related to Group Integration-Task (GI-T), and the leader behavior of training and instruction was related to Attraction to the Group-Task (ATG-T). Coaching behavior was not re- lated to either of the GEQ's social dimensions. Like Westre and Weiss (1991), the Pease and Kozub (1994) study had limited generalizability due to an exclusive focus on one gender and one competitive level.

Based on previous research and theory (Pease & Kozub, 1994; Westre & Weiss, 1991), we formulated two hypotheses: (a) higher levels of task cohesion would be positively related to the leader dimensions of training and instruction, democratic behavior, social support, and positive feedback, and negatively related to autocratic behavior; and, (b) higher levels of social cohesion would be posi- tively related to the leader dimensions of democratic behavior and social support, while being negatively related to autocratic behavior. In addition to these hypoth- eses, we were interested in exploring whether there were gender or school level differences in responding to the LSS or GEQ and, if so, whether these variables might mediate the leadership-cohesion relationship.

Method

Study Participants Baseball players (n = 189) and softball players (n = 118) representing six junior colleges and six high schools from large urban areas of California were used for this study. All baseball players were male, all softball players were fe- male. Altogether, players from 23 teams participated (n per team = 12.41, SD = 6.07).

High school athletes (n = 111) were drawn from five high school varsity baseball teams (n = 55) and six varsity softball teams (n = 56). The high school athletes ranged in age from 13 to 19 (M = 16.3; SD = 1.07). Eighty high school athletes described themselves as starters, and 29 as nonstarters, with two leaving the question blank. All high school teams were from inner-city public schools and study participants reflected the following raciallethnic groups: African American (n = 68), HispanicILatino (n = 20), Caucasian (n = 1 I), Asian (n = 6), Filipino (n = 5) , and Native American (n = 1).

The six baseball (n = 134) and six softball (n = 62) junior college teams were all from the same conference in northern California. The junior college athletes ranged in age from 18 to 33 (M = 19.7; SD = 1.89). Of the junior college athletes, 129 described themselves as starters, and 64 as nonstarters, with three athletes not responding. Junior college athletes (n = 196) reflected the following racial-ethnic composition: Caucasian (n = 113), Latino (n = 39), African American (n = 26), Asian (n = 8), Filipino (n = 2), Native American (n = 2), and Hawaiian (n = 2); four respondents declined to answer.

Measures Three instruments were used in the present study: The perceived version of the Leadership Scale for Sports (LSS), the Group Environment Questionnaire (GEQ), and a demographic questionnaire.

Coaching Behaviors and Team Cohesion 371

The LSS contains 40 items that ask athletes to indicate the frequency with which their coach engages in specific types of coaching behavior. Item responses are based on a 5-point Likert scale, ranging from "never to always," and scores for each scale are produced by summing the item responses and dividing by the num- ber of items in that category. The LSS assesses 5 dimensions of leader behavior: (a) training and instruction, the task-oriented behaviors of the coach designed to improve and refine the performance level of athletes; (b) democratic behavior, which reflects the extent to which a coach encourages athletetteam participation in team decisions; (c) autocratic behavior, which indicates the extent to which a coach stays aloof or distant from the athletestteam and stresses his or her authority when dealing with the team; (d) social support, which refers to the degree to which a coach is involved in attempting to satisfy interpersonal needs of athletes; and, (e) positive feedback, which is performance-based rewarding behavior on the part of the coach toward the athlete.

The GEQ consists of 18 items designed to measure individual group mem- bers' perceptions of team cohesiveness. Responses for all 4 categories are made on a 9-point Likert scale that ranges from "strongly disagree" to "strongly agree." Scores for each scale were computed by summing the relevant item values and dividing by the number of items in that category. The GEQ measures 4 aspects of team cohesiveness. Specifically, Individual Attraction to the Group-Task (ATG-T) is a measure of individual team members' feelings about their personal involve- ment with the group task, productivity, goals, and objectives. IndividualAttraction to the Group-Social (ATG-S) is a measure of individual team members' feelings about personal involvement, desire to be accepted, and social interaction with the group. These first two subscales have the individual assess his or her personal involvement with the group's task and the group's social aspects. In contrast, the next two subscales have the individual assess the group as a whole in terms of its coherence around task and social activities. Specifically, Group Integration-Task (GI- T) is a measure of the individual team members' feelings about the similarity, close- ness, and bonding within the team as a whole around the group's task; and Group Integration-Social (GI-S) measures the individual team members' feelings about the similarity, closeness, and bonding within the team, but for the group as a social unit.

The demographic questionnaire asked about participants' age, gender, race/ ethnicity, school level, starting status, and playing history. For each team, the win- loss record also was obtained.

Procedures At the high school level, permission to conduct the study was obtained from dis- trict commissioners, athletic directors, and team coaches. All 12 high school coaches gave consent for administering the questionnaires, but one baseball team could not be tested due to conflicting schedules which left the researchers with an odd num- ber of teams (N = 11). For the junior college teams, permission was obtained through each school's athletic director and head coach. Questionnaires were administered during the last third of the regular season to maximize opportunity for the develop- ment of team cohesion, without running into the complications and possible medi- ating influence of postseason play. Players first completed the demographic ques- tionnaire, then the LSS, and finally the GEQ. The measures were administered by one of the authors, and the coach was not present.

372 Gardner, Shields, Bredemeier, and Bostrom

Data Analyses The relationship between leadership behaviors and team cohesion was determined through use of correlation analyses and multivariate multiple regression and ca- nonical correlation analyses. The canonical correlation analyses were used to de- termine the strength of association between the leadership scales and the cohesion scales, and to understand which scales contributed the most to the multivariate relationship. Next, a multivariate analysis of variance (MANOVA) was conducted to determine whether perceptions of leader behaviors and group cohesion differed as a function of gender and competitive level. Finally, finding gender and school level differences in LSS and GEQ responding, we conducted another canonical correlation analysis to determine whether either or both of these variables acted as a moderator of the perceived leader behaviors and team cohesion relationship.

Results

Scale Reliabilities Internal reliability scores using Cronbach's (195 1) alpha coefficient for four of the LSS scales were reasonably strong: .88 (training & instruction), .83 (democratic behavior), .8 1 (social support), .85 (positive feedback). The alpha reliability score for the autocratic behavior scale was rather low (.65), as it often has been in previ- ous research (Chelladurai & Saleh, 1980; Chelladurai & Carron, 1981; Westre & Weiss, 1991). It was higher, however, than in either the Westre and Weiss (1991) or Pease and Kozub (1994) study, so we decided to retain it despite its marginal reliability to offer a preliminary examination of hypotheses relating autocratic be- havior to team cohesion.

Internal reliability for the four GEQ subscales was somewhat low: .60 (ATG- T), .61 (ATG-S), .68 (GI-T), and .60 (GI-S). Since our hypotheses pertained only to aggregate task and social cohesion, we combined the two task subscales into a single composite task cohesion scale, and the two social subscales into a single composite social cohesion scale. When the two task and social dimensions were combined, the reliability scores improved: .72 (task) and .71 (social). Only the composite scales were used in further analyses.

Relationship Between Leader Behaviors and Team Cohesion

To examine the relationship between LSS scales and group cohesion, we initially conducted a correlation analysis between the scales of the two measures. As can be seen in Table 1, all LSS scales were significantly correlated with task cohesion; autocratic behavior was negatively correlated while the other four scales posi- tively correlated. The correlation pattern was parallel for the total group, for both genders, and for both school levels. For social cohesion, only training and instruc- tion and social support were positively correlated for both genders. For males, autocratic behavior achieved a significant negative correlation, and for females, democratic behavior achieved a significant positive correlation. For junior college athletes, all five LSS scales were significantly correlated with social cohesion, following the same pattern as was evident with task cohesion. Only social support was positively correlated with social cohesion for high school athletes.

Coaching Behaviors and Team Cohesion 373

Table 1 Individual-Level Correlations: LSS Behaviors with Task and Social Cohesion

LSS Scale Task cohesion Social cohesion

Training and instruction (overall) Malelfemale H.S.1J.C. Democratic behavior (overall) Malelfemale H.S.1J.C. Autocratic behavior (overall) Malelfemale H.S.1J.C. Social support (overall) Malelfemale H.S.1J.C. Positive feedback (overall) Malelfemale H.S.1J.C.

Note. a = p <.05, b = p <.01, c = p <.001.

The overall test of a multivariate relationship between leader behaviors and group cohesion was significant, Wilks' lambda = .699, F(10,600) = 11.79, p < .0001. The canonical correlation between the two sets of variables was rc = .53. Despite these strong findings, the leadership variables accounted for only 6.53% of the variance in the group cohesion scores, according to the redundancy index (Pedhazur, 1982).

Labeling variables with loadings greater than .30 as significant (Pedhazur, 1982), canonical loadings were used to determine which specific variables con- tributed to the multivariate relationship. All five of the LSS scales were found to be significant contributors, with training and instruction (.77) most important, fol- lowed by social support (.68), democratic behavior (.66), autocratic behavior (-.63), and positive feedback (.59). For the GEQ, both the task and social scales were significant contributors to the relationship, though task cohesion (.998) clearly dominated over social cohesion (.54).

These multivariate results indicate there was a significant relationship be- tween perceived coaching behaviors and team cohesion among baseball and soft- ball players at two school levels. Specifically, the analyses suggest coaches who were perceived by their players to engage in more training and instruction, social support, democratic behavior, and positive feedback, while avoiding autocratic behavior, had athletes who described higher levels of task and social cohesion within their teams.

Gender and School Level Diflerences In order to determine whether leader behaviors and group cohesion were perceived differently by gender and school level, a multivariate analysis of variance

374 Gardnw, Shields, Bredemeier, and Bostrom

(MANOVA) was conducted. The independent variables considered were gender and school level; the dependent variables were the five LSS scales and the two combined GEQ scales (task and social).

As indicated by Table 2, both main effects were significant, as was the inter- action between gender and school level. Female athletes perceived significantly more training and instruction, democratic behavior, positive feedback, and task cohesion than did their male counterparts. Males perceived their coaches as being significantly higher in autocratic behavior than females. Junior college athletes showed significantly higher perceptions of social support and social cohesion than high school athletes.

For the interaction effect, significant differences occurred in perceptions of training and instruction, autocratic behavior, social support, and social cohesion. To clarify which differences in these interaction patterns were significant, these variables were reanalyzed using a one-factor analysis of variance with four groups (four combinations of gender and competitive level). A Tukey multiple compari- son procedure was used to compare all pairs of groups. The comparisons between all four groups which were significant ( p < .05) by this test are presented in Table 3.

Table 2 Multivariate Relationship Between Perceived Leadership Behaviors and Team Cohesion

Source Lambda F(7,297) P

Gender .86 School level .93 Gender X school level .88

Table 3 One-Factor Analyses of Variance Using a 'hkey Multiple Comparison to Compare the Groups

Training and instruction JCMs perceived more than HSFs, JCFs, and HSMs HSFs perceived more than JCFs and HSMs

Democratic behavior HSMs perceived more than HSFs, JCMs, and JCFs

Autocratic behavior JCMs perceived more than HSMs and JCFs

Social support JCMs perceived more than HSMs, HSFs, and JCFs HSMs perceived more than HSFs and JCFs HSFs perceived more than JCFs

Positive feedback JCMs perceived more than HSMs, HSFs, and JCFs

Task cohesion HSFs perceived more than HSMs and JCMs JCFs perceived more than HSMs and JCMs

Social cohesion JCMs perceived more than HSMs

Note. HSM = high school males; HSF = high school females; JCM =junior college males; JCF =junior college females.

Coaching Behaviors and Team Cohesion 375

Moderating Eflects of Gender and School Level, Since gender and school level differences were found in both perceptions of lead- ership, as measured by the LSS, and perceptions of team cohesion, as measured by the GEQ, we posited that gender and/or school level might mediate the relation- ship between the leadership and cohesion variables. To test for this possibility, another canonical correlation was conducted, adding both demographic variables into the equation to see if either contributed to the relationship. It was found, how- ever, neither demographic variable contributed significantly to the leadership-co- hesion relationship (rc = .55).

Discussion

The primary purpose of this study was to further examine the relationship between leader behaviors and team cohesion. Both correlational and multivariate analyses supported the contention that there is a relationship between perceived coaching be- haviors and team cohesion, especially team cohesion around task-related functions.

Perceived Coaching Behaviors and Team Cohesion We hypothesized that higher levels of task cohesion would be positively related to the leader dimensions of training and instruction, democratic behavior, social sup- port, and positive feedback, and negatively related to autocratic behavior. The re- sults supported this hypothesis for the overall sample, both genders, and athletes at both school levels. Canonical loadings of all five scales of leader behaviors were significant and relatively close, suggesting that no single dimension of coaching behavior can be viewed in isolation.

No cause-effect relationship can be inferred from these findings. Still, the group-cohesion theory, on which the GEQ is based, states that leadership behav- iors are one antecedent of team cohesion. Thus, it is reasonable to hypothesize, based on both theory and these results, that coaches can promote greater task cohe- sion among their team members through augmenting their training and instruc- tion, democratic behavior, social support, and positive feedback, while avoiding behaviors likely to be perceived as autocratic.

The underlying reasons for these results are not entirely clear. Most likely, training and instruction augments task cohesion by helping team members focus on a common, shared goal. Since positive feedback is contingent on successful implementation of the coach's training and instruction, the relationship between higher levels of task cohesion and more perceived positive feedback is logically coherent with the task cohesion and training and instruction relationship.

Westre and Weiss (1991) plausibly explained the positive relationship be- tween democratic behavior and task cohesion as a reflection of group participation in setting task-related goals. Additionally, a coach's democratic behavior may en- courage task-related independent thought, heightening a sense of both personal growth and team ownership which, in turn, may facilitate task cohesion. A coach who encourages his or her athletes to make situation-specific decisions-such as the catcher calling pitches, infielders deciding when to attempt pickoff plays, or a batter determining whether to swing at a 3-0 pitch-may heighten commitment to team tasks because everyone will experience outcomes as related not only to the coach's decisions, but their own.

376 Gardner, Shields, Bredemeier, and Bostrom

The coach's social-support behavior may encourage task cohesion in a couple of ways. First, the social-support behaviors of the coach may assist the athlete in dealing with nonsport distractions that may have been hindering a focus on team goals. In addition, the coach who exhibits a high level of interest in his or her athletes-not as athletic performers, but as people-may foster a reciprocal sense of commitment by the athletes to the coach, which could readily translate into commitment to team goals. The negative relationship between autocratic behavior and task cohesion is parsimoniously explained in terms of the athlete's decreased sense of ownership of team goals.

It also is possible the causal relationships are the reverse of what was just implied. In a recent meta-analysis, Mullen and Copper (1995) demonstrated that group performance affects cohesion more strongly than group cohesion affects performance. Maybe there is a similar relationship between leader behavior and group cohesion. As teams become more cohesive, they elicit from the coach more positive coaching behaviors. In a similar vein, it has been empirically demonstrated that group members' perceptions of their leaders are often biased by the outcomes those leaders have produced (Lord, 1985; Rush & Russell, 1988). It may be that athletes in more cohesive teams, because of their positive experience, exhibit a more bi- ased interpretation of their coach's behavior than athletes on less cohesive teams.

Our second hypothesis stated that higher levels of social cohesion would be positively related to the leader dimensions of democratic behavior and social sup- port, and negatively related to autocratic behavior. This hypothesis also was gen- erally supported, though not consistently in the subgroup analyses. It is not sur- prising, however, that the link between coaching behavior and social cohesion was less robust than between coaching behavior and task cohesion since the coaches'

u

primary function is task related. The one dimension of leadership behavior probably most directly tethered

to social cohesion is the coach's social support. Since social support is provided outside of the athletic context (Danielson, Zelhart, & Drake, 1975), and communi- cates a high level of caring and commitment, it may heighten the general sense of mutual care and concern on the team. Such a team atmosphere certainly would facilitate bonding around the social dimension. Consequently, the finding that per- ceived social support was correlated with social cohesion for the entire sample and all subgroups was not a surprise.

After social support, the leadership scale most closely linked to social cohe- sion was neither democratic nor autocratic behavior, but training and instruction. One possible explanation for this unanticipated finding is that coaches, who were perceived as providing more training and instruction, were likely to have teams who experienced themselves as more interdependent. In a team sport such as base- ball or softball, effective team play requires highly coordinated actions. In turn, perceived interdependence generally fosters interpersonal attraction (Deutsch, 1962; Johnson & Johnson, 1989).

One other important trend in the social cohesion data is worth noting. All perceived leader behaviors were correlated with social cohesion for the junior col- lege athletes, but for the high school athletes only perceived social support corre- lated. One likely explanation for this pattern has to do with the time athletes at the junior college level spend interacting with their coach throughout the year. Re- searchers have found that coaching behaviors may vary depending on the time of season (Lacy & Darst, 1985) and in game versus practice situations (Horn, 1985).

Coaching Behaviors and Team Cohesion 377

Demographic Variables and Perceived Leader Behaviors The MANOVA demonstrated there were significant main effects for both gender and school level in perceptions of leadership behaviors; in addition, the interaction effect was significant. That males perceived more autocratic behavior than females was not surprising, but we were surprised the female athletes perceived more training and instruction, democratic behavior, and positive feedback. This finding may in- dicate that coaches of female teams generally employ different leadership behav- iors than coaches of male teams, responding, perhaps, to distinct gender prefer- ences in leadership style (Chelladurai & Saleh, 1978; Terry, 1984; cf. Terry & Howe, 1984). Alternately, male and female athletes may perceive or report a given level of leadership behavior differently, due perhaps to differences in their previ- ous sport histories. Clearly, additional research is needed to clarify the operative dynamics involved.

For school level, it was found that junior college athletes perceived more social support than their high school counterparts. Since previous research has indicated that preference for more social support increases across this age span (Chelladurai & Carron, 1983), this finding may suggest coaches are generally re- sponsive to their athletes' preferences.

An examination of the Tukey test results on the significant interactions re- veals an interesting pattern. In each case, with one minor exception, junior college males were significantly different from the other three groups. They perceived their coaches as providing more training and instruction, social support, and auto- cratic behavior than the other groups. One likely explanation for this finding is that of the groups studied, the male junior college teams were the ones most likely to imitate the professional model. The gap between the high school and the profes- sional ranks is quite large, and the lack of postcollege opportunities for softball players is likely to dampen the influence of the pros on female teams. In profes- sional baseball, coaches are likely to exhibit the kinds of behaviors that junior college baseball players attribute to theirs.

Demographic Variables and Team Cohesion In terms of perceptions of team cohesion, there were again significant differences by gender and school level. Softball teams at both school levels were significantly higher in task cohesion than baseball teams, while college students were signifi- cantly higher than high school students in social cohesion. The first finding may suggest task cohesion is achieved more easily with female teams, perhaps because female socialization emphasizes connection, relationship, and interdependence (Gilligan, 1982). The latter finding may reflect increased opportunities junior col- lege athletes have to form social bonds with teammates due to an extended focus on one sport, multiple day road trips, and interaction throughout an entire year instead of a season. Another possible interpretation has to do with the GEQ itself. The GEQ was originally written for university-level athletes. Questions like "I did not enjoy being a part of the social activities of the team," and "I enjoyed other parties more than the team parties" are much more appropriate for athletes at the collegiate level and may not accurately assess the social reasons for affiliation at the high school level. The failure in previous research with high school athletes to achieve acceptable reliability ratings for the two social cohesion subscales of the GEQ (Pease & Kozub, 1993; Westre & Weiss, 1991) may reflect this problem.

378 * Gardner, Shields, Bredemeier, and Bostrom

The one significant interaction effect relevant to team cohesion demonstrated that junior college males perceived more social cohesion than high school males. This may again reflect the possibly greater professional tenor of these teams. Par- alleling what was said before, this also may reflect the fact that the junior college baseball players spend more time together than members of the other teams.

Conclusion The results from this study add support for the notion that a significant relation- ship exists between leadership behaviors and team cohesion. Coaches who are perceived as providing more training and instruction, democratic behavior, social support, and positive feedback, while limiting autocratic behavior, tend to have more cohesive teams. Of course, the present study does not directly provide evi- dence of causality or causal direction. Though the previous discussion has empha- sized the potential impact of coaching behaviors on team cohesion, it also is likely that team cohesion will influence coaching behaviors. In addition to implementing research designs that might allow causal connections to be demonstrated, future work assessing the relationship between leadership and cohesion might also profit from a qualitative approach to the study of the interaction between these two con- structs.

Several important gender and school level differences were found in both the perceptions of leader behaviors and team cohesion. Future research using the LSS or GEQ would benefit from a closer examination of how demographic factors influence responses to these instruments. In particular, the suitability of the GEQ for assessing social cohesion among high school athletes needs to be addressed. In terms of the present study, however, neither gender nor school level appeared to moderate the leadership-cohesion relationship.

References

Bird, A.M. (1977). Development of a model for predicting team performance. Research Quarterly, 48,24-32.

Blair, S. (1985). Professionalization of attitude toward play in children and adults. Research Quarterly in Exercise and Sport, 56, 82-83.

Brawley, L.R., Carron, A.V., & Widmeyer, W.N. (1988). Exploring the relationship be- tween cohesion and group resistance to disruption. Journal of Sport & Exercise Psy- chology, 10, 199-213.

Brawley, L.R., Carron, A.V., & Widmeyer, W.N. (1987). Assessing the cohesion of teams: Validity of the Group Environment Questionnaire. Journal of Sport Psychology, 9, 275-294.

Card, A. (198 1, April). Orientation toward winning as afinction of athletic participation, grade level, and gender. Paper presented at the annual meeting of the AAHPERD, Detroit, MI.

Carron, A.V. (1982). Cohesiveness in sport groups: Interpretations and considerations. Jour- nal of Sport Psychology, 4,123-128.

Carron, A.V. (1988). Group dynamics in sport. London, ON: Spodym Publishers. Carron, A.V., Widmeyer, N., & Brawley, L. (1985). The development of an instrument to

assess cohesion in sport teams: The Group Environment Questionnaire. Journal of Sport & Exercise Psychology, 10,244-266.

Coaching Behaviors and Team Cohesion 379

Carron, A.V., Widmeyer, W.N., & Brawley, L.R. (1988). Group cohesion and individual adherence to physical activity. Journal of Sport & Exercise Psychology, 10,119- 126.

Chelladurai, P. (1984). Discrepancy scores between preferences and perceptions of leader- ship behavior and satisfaction of athletes in varying sports. Journal of Sport Psychol- ogy, 6,27-41.

Chelladurai, P. (1993). Leadership. In R. Singer, M. Murphey, & L. Tennant (Eds.), Hand- book of research on sportpsychology. New York: Macmillan.

Chelladurai, P., & Carron, A.V. (1978). Leadership. [Monograph]. Canadian Association for Health, Physical Education, and Recreation Sociology of Sport, Series A. Calgary, AB: University of Calgary.

Chelladurai, P., & Carron, A.V. (1981). Applicability to youth sports of the Leadership Scale for Sports. Perceptual and Motor Skills, 53, 361-362.

Chelladurai, P., & Carron, A.V. (1983). Athletic maturity and preferred leadership. Journal of Sport Psychology, 5,371-380.

Chelladurai, P., Imamura, H., Yamaguchi, Y., Oinuma, Y., & Miyauchi, T. (1988). Sport leadership in a cross-national setting: The case of Japanese and Canadian university athletes. Journal of Sport & Exercise Psychology, 10, 374-389.

Chelladurai, P., & Saleh, S. (1978). Preferred leadership in sports. Canadian Journal of Applied Sport Sciences, 3, 85-92.

Chelladurai, P., & Saleh, S. (1980). Dimensions of leader behavior in sports: Development of a leadership scale. Journal of Sport Psychology, 2,34-45.

Chodorow, N . (1978). The reproduction of mothering. Berkeley, CA: University of Califor- nia Press.

Cronbach, L.J. (195 1). Coefficient alpha and internal structure of tests. Psychometricka, 2, 135-138.

Danielson, R.R., Zelhart, P.F., & Drake, C.J. (1975). Multidimensional scaling and factor analysis of coaching behavior as perceived by high school hockey players. Research Quarterly, 46,323-334.

Deutsch, M. (1962). Cooperation and trust: Some theoretical notes. In M.R. Jones (Ed.), Nebraska Symposium on Motivation (pp. 275-319). Lincoln, NE: University of Ne- braska Press.

Dwyer, J.J.M., & Fischer, D.G. (1988). Psychometric properties of the coach's version of the Leadership Scale for Sports. Perceptual and Motor Skills, 67,795-798.

Garland, D.J., & Bany, J.R. (1988). The effects of personality and perceived leader behav- ior on performance in collegiate football. The Psychological Record, 38,237-247.

Gill, D.L. (1986). Competitiveness among females and males in physical activity classes. Sex Roles, 15,233-247.

Gill, D.L. (1988). Gender differences in competitive orientation and sport participation. International Journal of Sport Psychology, 19, 145-159.

Gilligan, C. (1982). In a different voice: Psychological theory and women's development. Cambridge, MA: Harvard University.

Gordon, A.M.D. (1986). Behavioral correlates of coaching effectiveness. Unpublished doc- toral dissertation. University of Alberta, Edmonton, AB.

Granito, V., & Rainey, D. (1988). Differences in cohesion between high school and college football teams and starters and nonstarters. Perceptual and Motor Skills, 66, 471- 477.

Hastie, P. (1993). Coaching preferences of high school girl volleyball players. Perceptual and Motor Skills, 77, 1309-13 10.

Horn, T. (1985). Coaches' feedback and changes in children's perceptions of their physical competence. Journal of Educational Psychology, 77, 174-186.

380 Gardner, Shields, Bredemeier, and Bosfrom

Home, T., & Carron, A.V. (1985). Compatibility in coach-athlete relationships. Journal of Sport Psychology, 7, 137-149.

Johnson, D.W., &Johnson, R.T. (1989). Cooperation and competition: Theory and research. Edina, MN: Interaction Books.

Kidd, T., & Woodman, W. (1975). Sex and orientation toward winning in sport. Research Quarterly, 46,476-483.

Knoppers, A. (1985). Professionalization of attitudes: Areview and critique. Quest, 37,92- 102.

Lacy, A., & Darst, P. (1985). Systematic observations of behaviors of winning high school head football coaches. Journal of Teaching in Physical Education, 4,256-270.

Landers, D.M., & Luschen, G. (1974). Team performance outcome and the cohesiveness of competitive coacting groups. International Review of Sport Sociology, 9,57-69.

Lenk, H. (1969). Top performance despite internal conflict. An antithesis to a functionalis- tic proposition. In J. Loy & G. Kenyon (Eds.), Sport, culture, and society (pp. 393- 397). London: Macmillan.

Maloney, T.L., & Petrie, B. (1972). Professionalization of attitude toward play among Ca- nadian school pupils as a function of sex, grade, and athletic participation. Journal of Leisure Research, 4, 184- 195.

Martens, R., & Peterson, J. (1971). Group cohesiveness as a determinant of success and member satisfaction in team performance. International Review of Sport Sociology, 4,49-59.

Mullen, B., & Copper, C. (1994). The relation between group cohesiveness and perfor- mance: An integration. Psychological Bulletin, 115, 210-227.

Nixon, H. (1980). Orientation toward sports participation among college students. Journal of Sport Behavior, 3,29-45.

Pease, D.G., & Kozub, S. (1994). Perceived coaching behaviors and team cohesion in high school girls basketball teams. Journal of Sport & Exercise Psychology: NASPSPA abstracts, 16, S93.

Pedhazur, E.J. (1982). Multiple regression and behavioral research. New York: Holt, Rinehart, & Winston.

Robinson, T.T., & Carron, A.V. (1982). Personal and situational factors associated with dropping out versus maintaining participation in competitive sport. Journal of Sport Psychology, 4,364-378.

Sage, G. (1980). Orientation toward sport of male and female intercollegiate athletes. Jour- nal of Sport Psychology, 2,355-362.

Schliesman, E. (1987) Relationship between the congruence of preferred and actual leader behavior and subordinate satisfaction with leadership. Journal of Sport Behavior, 10, 157-166.

Serpa, S. (1990). Research work on sport leadership in Portugal. Unpublished manuscript, Lisbon Technical University.

Serpa, S., & Antunes, I. (1989). Leadership styles of elite Portuguese women's volleyball coaches. Paper presented at the 6th International Congress on Sport Psychology, Lahti, Finland.

Serpa, S., Pataco, V., & Santos, F. (1991). Leadership patterns in handball international competition. International Journal of Sport Psychology, 22,78-89.

Terry, P. (1984). The coaching preferences of elite athletes competing at Universiade '83. Canadian Journal of Applied Sport Sciences, 9, 201-208.

Teny, P.C., & Howe, B.L. (1984). The coaching preferences of athletes. The Canadian Journal ofApplied Sport Sciences, 9, 188-193.

Coaching Behaviors and Team Cohesion 381

Weiss, M., & Friedrichs, W. (1986). The influence of leader behaviors, coach attributions, and institutional variables on performance and satisfaction of collegiate basketball teams. Journal of Sport Psychology, 8,332-346.

Westre, K., & Weiss, M. (1991). The relationship between perceived coaching behaviors and group cohesion in high school football teams. The Sport Psychologist, 5,41-54.

Widmeyer, W.N., Brawley, L.R., & Carron, A.V. (1985). The measurement of cohesion in sport teams: The Group Environment Questionnaire. London, ON: Sports Dynamics.

Widmeyer, W.N., Carron, A.V., & Brawley, L.R. (1993). Group cohesion in sport and exer- cise. In R. Singer, M. Murphey, & L. Tennant (Eds.), Handbook of research on sport psychology. New York: Macmillan.

Williams, J. (1993). Applied sportpsychology: Personal growth topeakperformance (2nd ed.) Mountain View, CA: Mayfield.

Williams, J., & Widmeyer, W.N. (1991). The cohesion-performance outcome relationship in a coacting sport. Journal of Sport & Exercise Psychology, 13,364-371.

Yukelson, D., Weinberg, R., & Jackson, A. (1984). A multidimensional group cohesion instrument for intercollegiate basketball teams. Journal of Sport Psychology, 6,103- 117.

Zanna, M.P., & Fazio, R.H. (1982). The attitude-behavior relation: Moving toward a third generation of research. In M.P. Zanna, E.T. Higgins, & C.P. Herman (Eds.), Consis- tency in social behavior: The Ontario Symposium: Vol. 2 (pp. 283-301). Hillsdale, NJ: Erlbaum.

Manuscript submitted: June 1995 Revision received: March 1996