The impact of leadership behavior on satisfaction of college tennis players: a test of the leadership behavior congruency hypothesis of the multidimensional model of leadership.

Andrew, Damon P.S.

Most research on coaching effectiveness has assumed that coaches greatly influence athletes' performance and behavior, as well as their general psychological and emotional well-being (Chelladurai, 1990; 1993). The study of the role of the leader/coach in athletics has been aided greatly over the past two decades or so by the existence of the Multidimensional Model of Leadership (MML--See Figure 1), developed by Chelladurai (1978) as an attempt to both bring parsimony to the numerous approaches to studying leadership in the mainstream, and to bring a sport-specific focus to the study of leadership. Further, the MML addresses the concerns put forth by Chelladurai and Carron (1978), who asserted that the direct application of mainstream leadership theory to sport situations may not adequately account for the distinctness of the sport context.

[FIGURE 1 OMITTED]

The MML proposes that three aspects of leader behavior need to be in congruence to achieve effective group performance and member satisfaction. The aspects of leader behavior include required (behavior that is prescribed for a particular situation), preferred (behavior preferred of the coach by the athletes), and actual, hereafter referred to as perceived (the coach's behavior as perceived by the athletes). Required leader behavior is influenced by situational characteristics such as organizational goals, formal structure, group task, social norms, government regulations, technology, and the nature of the group (Chelladurai, 2006). In 1990, Chelladurai revised the antecedents of required leader behavior to also include member characteristics. For example, in situations where members lack the intelligence, ability, experience, and/or personality dispositions to make judgments about situational requirements, the leader must make an appropriate decision for the members. Therefore, required leader behavior is determined by situational and member characteristics. Preferred leader behavior stems from both the aforementioned situational characteristics and member characteristics such as task-relevant ability (House, 1971; House & Dressier, 1974), personality traits, attitude toward authority (Lorsch & Morse, 1974; Morse, 1976), cognitive complexity (Wynne & Hunsaker, 1975), authoritarianism and the need for independence (Vroom, 1959). Perceived leader behaviors are partially determined by the characteristics and behaviors of the leader (i.e., personality, ability, experience, and style), but are also determined to some extent by required and preferred leader behavior. Therefore, the leader may alter his or her behavior to the requirements of the situation and the preferences of the members to some degree.

The main proposition of the MML is that, to a large degree, group performance and member satisfaction are dependent upon the congruency of required, preferred, and perceived leader behaviors. In other words, group performance and member satisfaction can be enhanced when the leadership behavior required by the situation, the leadership behavior preferred by the followers, and the leadership behavior perceived by the followers are similar. In contrast, when the leadership behavior required by the situation, the leadership behavior preferred by the followers, and the leadership behavior perceived by the followers are not similar, group performance and member satisfaction are compromised. Although more study is warranted, initial research has supported this proposition and the individual tenets of the MML. Strong support has been shown, for example, for the link between member characteristics and coaching behaviors (e.g., Chelladurai & Carron, 1983; Chelladurai, Imamura, Yamaguchi, Oinmuma, & Miyauchi, 1988; Chelladurai, Malloy, Imamura, & Yamaguchi, 1987). Research has also clearly identified a link between leadership behavior congruency and athlete satisfaction (Chelladurai, 1978; 1984; Chelladurai et al., 1988; Dwyer & Fischer, 1988; Home & Carron, 1985; McMillin, 1990; Riemer & Chelladurai, 1995; Schliesman, 1987; Summers, 1983; Weiss & Friedrichs, 1986) as well as group performance (Gordon, 1986; Serpa, Pataco, & Santos, 1991; Weiss & Friedrichs, 1986). Collectively, these studies provided initial support for the leadership behavior congruency hypothesis within the MML.

However, in 1998, Chelladurai and Riemer noted that the majority of research concerning the leadership behavior congruency hypothesis of the MML had incorporated difference scores in their analysis. That is, researchers commonly created a hybrid independent variable by subtracting one form of leadership behavior from another (e.g., subtracting perceived leadership behavior scores from preferred leadership behavior scores) and then attempted to associate that hybrid independent variable with an outcome variable (e.g., member satisfaction, group performance). Unfortunately, as noted by Chelladurai and Riemer (1998), Peter, Churchill, and Brown (1993) had outlined numerous negative consequences of incorporating difference scores in data analysis techniques including problems with reliability (Johns, 1981; Lord, 1958; Mosier, 1951; Prakash & Lounsbury, 1983), discriminant validity (Churchill, 1979; Peter, 1981), spurious correlations (Wall & Payne, 1973), and variance restriction (Wall & Payne, 1973). After reviewing the negative consequences of difference scores, Chelladurai and Riemer (1998, p. 245) concluded, "An implication of the [use of discrepancy scores] is that any results of research using discrepancy (i.e., difference) scores are not tenable." This conclusion essentially invalidated all studies examining the leadership behavior congruency hypothesis of the MML up to that time.

In order to avoid these potential problems associated with the use of difference scores researchers have suggested a hierarchical regression technique (Berger-Gross, 1982; BergerGross & Kraut, 1984; Cronbach, 1958; Johns, 1981; Rice, McFarlin, & Bennett, 1989; Riemer & Chelladurai, 1995). Riemer and Chelladurai (1995) pointed out that Cronbach (1958) has demonstrated how the interaction of two component parts (e.g., preferences and perceptions) is equivalent to the differences of those components. Therefore, researchers can avoid difference scores by entering each component separately into a regression followed by the interaction of those two terms. This approach has been advocated by others (Riemer & Chelladurai, 1995; Riemer & Toon, 2001) to provide information regarding the unique variance accounted for by each leadership variable as well as its relative dominance. Cronbach (1958, p. 356) noted that an interaction hypothesis is justified "only if it improves significantly" upon the simpler (non-interactive) prediction. Therefore, the leadership behavior congruency hypothesis is accepted if the interaction significantly increases the amount of variance explained.

In order to test the efficacy of the leadership behavior congruency hypothesis in the MML with this new procedure, Riemer and Toon (2001) collected data from a sample of 148 tennis players who were competing in the NCAA Division I and II Championships. Their study found no support for the leadership behavior congruency hypothesis when examining leadership satisfaction as an outcome of the congruency of preferred and perceived leadership behavior. These results contrasted with the findings of previous research that had incorporated difference scores in their analyses (Chelladurai, 1984; Home & Carron, 1985; Schliesman, 1987). Given that prior research examining the leadership behavior congruency hypothesis of the MML has either found no support (Riemer & Toon, 2001) or incorporated difference scores in their analysis (Chelladurai, 1984; Home & Carton, 1985; Schliesman, 1987), there remains a need to explore the efficacy of the leadership behavior congruency hypothesis to determine whether the MML is in need of revision. Further, since the only study to examine the leadership behavior congruency hypothesis without the use of difference scores incorporated a sample of collegiate tennis players who were successful enough to advance to the NCAA Division I and II Championships, more research is needed to confirm whether the conclusions drawn from that relatively elite sample are applicable to other skill levels and settings. Further, that study measured leadership behavior via the Leadership Scale for Sports (Chelladurai & Saleh, 1980) rather than the Revised Leadership Scale for Sports, which Zhang, Jensen, and Mann (1997) argue is more contextually appropriate for collegiate athletes in the United States than the original Leadership Scale for Sports. Therefore, the purpose of this study was to explore whether the congruency of preferred and perceived leadership behaviors significantly predicted leadership satisfaction in a sample of NCAA Division I, II, and III tennis players. It was hoped that the more representative sample along with a contextually appropriate measurement tool for leadership in the present study would provide an interesting contrast with the findings of Riemer and Toon (2001), who examined elite collegiate athletes in a similar setting.

Method

Due to the inherent structured and seasonal competition of NCAA athletic teams, intercollegiate tennis players were targeted as a sample. Upon receiving approval from the Institutional Review Board to conduct the study, a total of 1107 collegiate tennis coaches at the NCAA Division I, II, and III playing levels were informed of the study via a pre-notification email (see Kent & Turner, 2002) during the last two weeks of their regular season competition. This time period was deemed long enough for new and existing players to establish a coach-player relationship, but early enough to avoid excessive time conflicts with the post-season championships and final exams. E-mail addresses for each coach were obtained from the College Tennis Connect website (http://www.collegetennisconnect.com). However, a total of 96 email messages sent to coaches were returned to sender due to complications such as incorrect e-mail addresses, terminated e-mail accounts, exceeded storage quota limits, and temporary absence of head coach due to administrative or maternity leave. A total of 14 of the returned messages were successfully corrected to reflect the present e-mail address of the coaches, so only 82 messages were undeliverable, leaving a total coaching tally of 1025.

Following the recommendations of Dillman (2000), an e-mail message was sent to each coach one week after the remittance of the pre-notification message asking him or her to encourage and facilitate athlete participation. The letter included a summary of the risks and benefits of participation along with directions to complete the survey at a secure website. Since individual e-mail addresses of the athletes were not available to the researchers, the coaches were asked to forward the electronic message to each of their respective athletes and carbon copy ("CC") the message to the primary investigator's e-mail address. Receipt of the carbon copied message allowed the primary investigator to determine the number of athletes who received invitations to participate in the survey. Follow-up reminders were sent to the coaches and athletes each week for a total of four weeks. The surveys were conducted in an online format in an attempt to maximize player convenience, secure response confidentiality, and minimize necessary paper. The survey was administered through a third-party company entitled FormSite (http://www.formsite.com).

Participants

Coaches from 81 institutions forwarded the survey to their tennis players. A total of 514 intercollegiate tennis players received invitations from their respective coaches to participate in the survey. A total of 245 athletes responded to the request to participate, for a response rate of 47.7%. Demographically, the respondents were 31.8% (n = 78) male and 68.2% (n = 167) female, with ages ranging from 18 to 24 years and a mean age of 20.01 (SD = 1.38). The majority (78.8%; n = 193) of the sample cited the United States as their nationality, with the remainder of the sample (22.2%; n = 52) citing one of 31 different foreign countries. The majority of the respondents competed in NCAA Divisions I (42.0%; n = 103) and III (48.6%; n = 119), with only 9.4% (n = 23) of the athletes attending Division II programs. A total of 76 different colleges were represented in the sample by at least one student athlete. Years of playing experience in the sport of tennis ranged from one to 18 years with a mean of 11.05 (SD = 3.53). Years of collegiate playing experience ranged from one to five years with a mean of 2.05 (SD = 1.10). The singles playing position for each team member ranged from one to 20 with a mean of 4.47 (SD = 2.84). It is important to note that the competition format in collegiate tennis in the United States allows for six singles players to compete in a dual match versus another team. Therefore, the singles players ranked seven through 20 in the present study may not participate in singles competition as often as those ranked in the top six singles players on their respective teams. Only 15.9% (n = 39) of respondents received full athletic scholarships, while 20.8% (n = 51) received partial athletic scholarships and 63.3% (n = 155) received no athletic scholarship funding, which corresponds with the fact that 48.6% (n = 193) of the respondents competed in NCAA Division III programs, which do not offer athletic scholarships. Of those eligible to receive athletic scholarships (i.e., Divisions I and II; n = 126), a total of 71.4% (n = 90) of athletes received at least some form of athletic scholarship funding. Only 4.5% (n = 11) of respondents received full academic scholarships, while 36.7% (n = 90) received partial academic scholarships and 58.8% (n = 144) received no academic scholarship funding.

Instrumentation

In addition to general demographic information (age, gender, nationality, years of overall playing experience, years of collegiate playing experience, name of current institution, singles playing position on team, NCAA playing division, level of academic scholarship funding, and level of athletic scholarship funding), perceived leadership behavior, preferred leadership behavior, and satisfaction were assessed. Instruments were carefully selected based on their demonstrated reliability and validity in past studies.

Zhang, Jensen, and Mann's (1997) Revised Leadership Scale for Sports (RLSS) was utilized to assess perceived and preferred leadership behavior. The RLSS is a 60-item questionnaire containing the following subscales: Training and Instruction, Democratic Behavior, Autocratic Behavior, Social Support, Positive Feedback, and Situation Consideration. Training and instruction reflects the coach's ability to improve the performance level of the athlete. The extent to which the coach permits participation by the athletes in decision-making is termed democratic behavior. Autocratic behavior indicates the extent to which a coach keeps apart from the athletes and stresses his or her authority in dealing with them. The social support factor refers to the extent to which the coach is involved in satisfying the interpersonal needs of the athletes. The positive feedback factor represents the coach's expressions of appreciation and willingness to compliment the athletes for their performance and contribution. Finally, situational consideration behaviors includes proper coaching behaviors aimed at considering the situational factors (i.e., time, individual, environment, team, and game), setting up individual goals and clarifying ways to reach the goals, differentiating coaching methods at different stages, and assigning an athlete to the right game position. Cronbach alpha reliability tests confirmed factor ratings equal to or greater than .81 on the democratic, positive feedback, situation consideration, teaching and instruction, and social support dimensions; however, the autocratic behavior subscale received ratings of .59, .48, and .35, for the versions of athlete preference, athlete perception, and coach self-evaluation respectively. Issues with the internal consistency of the autocratic behavior subscale have also been noted for the original Leadership Scale for Sports (Chelladurai & Riemer, 1998). Respondents usually complete the RLSS by using a five point Likert scale, which signifies "always", "often", "occasionally", "seldom", and "never", but a seven point Likert scale was utilized in the present study to allow for the use of a standard seven point response for all survey instruments.

The Athlete Satisfaction Questionnaire (ASQ; Riemer & Chelladurai, 1998) was developed to measure the facets of satisfaction identified previously by Chelladurai and Riemer (1997). The ASQ is a 56-item questionnaire that contains 15 dimensions of athlete satisfaction. Respondents complete the ASQ by using a seven point Likert scale ranging from "not at all satisfied" to "extremely satisfied." Using a sample of 614 Canadian university athletes (basketball, hockey, and volleyball), the researchers reported Cronbach alpha coefficients ranging from .78 to .95 (M= .88), and 12 of the 15 subscales were higher than .85. The following four subscales were included: training and instruction satisfaction (three items), personal treatment satisfaction (five items), team performance satisfaction (three items), and individual performance satisfaction (three items). Of the 15 subscales in the ASQ, only these four subscales are conceptually related to leadership behavior. Specifically, the first two subscales focus on satisfaction with the processes of coaching behavior, while the latter two subscales evaluate satisfaction with outcomes associated with the process of leadership (Riemer & Chelladurai, 1998). Other studies utilizing the ASQ to measure athlete satisfaction as an outcome of leadership behavior have reduced the number of questionnaire items for the ASQ in this same manner (Al-Tahayneh, 2003; Riemer & Toon, 2001) and for the same reason. Therefore, only the 14 items in these four subscales of the ASQ were presented to the respondents.

A pilot study surveying the collegiate tennis players at the host institution indicated that the average response time to complete the survey ranged from 20 to 25 minutes. The results from the pilot study were not included in the data analysis for the present study since it was completed at an earlier time during the season.

Data Analysis

Descriptive statistics were calculated for each of the demographic variables. Cronbach's alpha coefficients were calculated for the components of each measurement scale to verify internal consistency. Confirmatory factor analyses were performed on each scale to assess an appropriate fit of the data to the identified model factors. Hierarchical regression analyses were performed whereby each component was entered separately into a regression followed by the interaction of those two terms, with the four facets of satisfaction serving as the criterion variables in each equation. Specifically, the base scores (leadership preferences and perceptions) were entered fast followed by their interactional term (preferred x perceived). Two sets of multiple regression equations were then calculated. In the first set, preference scores were entered first followed by the perceptions and the interaction term, and the second set followed a similar format but reversed the order of the preference and perception terms. This approach has been advocated by others (Riemer & Chelladurai, 1995; Riemer & Toon, 2001) to provide information regarding the unique variance accounted for by each leadership variable as well as its relative dominance. The leadership behavior congruency hypothesis was accepted if the interaction significantly increased the amount of variance explained.

Results

Instrument Assessment

Internal consistency coefficients of all subscales exceeded the value of .70, accepted by most as an adequate benchmark (Nunnally & Bernstein, 1994) (see Table 1 for means, standard deviations, and internal consistency estimates of all subscales). Scale fit was assessed via a confirmatory factor analysis. The calculated fit indices for all scales can be found in Table 2. The Root Mean Square Error of Approximation (RMSEA) for all scales was at or below 0.1, the maximum value threshold suggested by Steiger (1990) and Kelloway (1998) to indicate reasonable fit of the data to the model. Comparative fit was assessed via the Normed Fit Index (NFI), Non-Normed Fit Index (NNFI), and Comparative Fit Index (CFI). The fit indices for all scales met or exceeded the minimum threshold value of 0.90 suggested by Kelloway (1998) with the exception of the Normed Fit Index of the perceptions version of the RLSS (0.87). However, the 'perceptions' version of the RLSS was retained for further analysis (i.e., hierarchical regression) on the basis of its favorable performances on other comparative (NNFI and CFI) and absolute (RMSEA) fit indices as well as its acceptable overall reliability scores.

Leadership Behavior Congruency

The regression results detailing the impact of leadership behavior congruency on the dependent variables of individual performance, personal treatment, team performance, and training and instruction satisfaction are found in Table 3. As noted by Riemer and Toon (2001, p. 244), "The congruence hypothesis is accepted if the interaction significantly increases the amount of variance explained; relative size of the interaction to other equation terms is not important." In general, the results indicated the congruency of preferred and perceived training and instruction behaviors resulted in a significant [F(1, 241) = 66.48; p = .028] increase in [R.sup.2] (.011) for the dependent variable of personal treatment satisfaction. Furthermore, the congruency of preferred and perceived autocratic behavior significantly predicted individual performance satisfaction [F(1,241) = 13.28; [DELTA][R.sup.2] = .030; p = .004], personal treatment satisfaction [F(1, 241) = 22.50; [DELTA][R.sup.2] = .019;p = .015], team performance satisfaction IF(1, 241) = 7.46; [DELTA][R.sup.2] = .032; p =. 004], and training and instruction satisfaction [F(1,241) = 17.75; [DELTA][R.sup.2] = .040; p = .001 ]. In sum, these results indicate that some, but not all, facets of leadership behavior have an impact on satisfaction when the coach's leadership behavior as perceived by the athlete is similar to the behavior preferred by the athlete.

Discussion

The primary purpose of this study was to determine whether athlete satisfaction would be dependent on the congruence between preferred and perceived leadership behaviors. The results of this study are in line with propositions put forth in the MML that, in general, the congruency of leadership behaviors of coaches impact facets of athlete satisfaction (Chelladurai, 1984; Home & Carron, 1985; Riemer & Chelladurai, 1995; Schliesman, 1987). The congruency of two particular leadership behaviors, training and instruction and autocratic behavior, played a role in the determination of athlete satisfaction in the present study.

The congruency of training and instruction behavior, or behavior directed at improving the performance level of the athlete, resulted in a significant increase in R2 for the dependent variable of personal treatment satisfaction. Therefore, in the present sample, when the tennis coach provided the desired level of training and instruction behavior, he or she had the capability to influence the athlete's satisfaction with those coaching behaviors that directly affect the individual yet indirectly affect team development (i.e., social support and positive feedback).

The congruency of autocratic behavior appears to have played a dominant role in the determination of athlete satisfaction in the present study. The congruency of preferred and perceived autocratic leader behaviors resulted in a significant increase in [R.sup.2] for individual performance satisfaction, a significant increase in [R.sup.2] for personal treatment satisfaction, a significant increase in [R.sup.2] for team performance satisfaction, and a significant increase in [R.sup.2] for the dependent variable of training and instruction satisfaction. Taken together, these findings indicate that, if the coach provides the desired level of autocratic behavior, he or she has the capability to influence the athlete's satisfaction with: (a) his or her own task performance (i.e., absolute performance, improvements in performance, and goal achievement), (b) those coaching behaviors that directly affect the individual yet indirectly affect team development (i.e., social support and positive feedback), (c) his or her team's level of performance (i.e., absolute performance, goal achievement, and implies performance improvements), and (d) the training and instruction provided by the coach. However, it is important to note that the amount of preferred autocratic behavior in this study was relatively low (3.51 +/- 0.99), and that the amount of autocratic behavior perceived was even lower (2.89 +/- 1.04).

Comparing the findings of the present study to past research is challenging given that many past studies incorporated difference scores in their analysis of the leadership behavior congruency hypothesis in the MML (Chelladurai, 1984; Home & Carron, 1985; Schliesman, 1987). However, Riemer and Toon (2001) utilized an appropriate analysis in their evaluation of the leadership behavior congruency hypothesis of the MML. Their study explored five facets of leadership behavior (training and instruction, democratic behavior, autocratic behavior, social support, and positive feedback) and found no support for the leadership behavior congruency hypothesis among a sample of 148 tennis players who were competing in the NCAA Division I and II Championships. These results diverge with those of the present study, which examined six facets of leadership behavior (training and instruction, democratic behavior, autocratic behavior, social support, positive feedback, and situational consideration) and found preferred and perceived autocratic behavior to impact satisfaction with individual performance, personal treatment, team performance, and training and instruction. In addition, the congruence of preferred and perceived training and instruction behavior predicted satisfaction with personal treatment.

The contrasting results between the study of Riemer and Toon (2001) and the current study could be the result of several differences between the studies. First, Riemer and Toon (2001) utilized the Leadership Scale for Sports (Chelladurai & Saleh, 1980) rather than the Revised Leadership Scale for Sports (Zhang et al., 1997). Zhang et al. (1997) claimed to have improved the original LSS in several ways: (a) the items were generated through interviewing the coaches, hence, they are sports specific; (b) the study was conducted in the United States and the regulations of the National Collegiate Athletic Association were considered, thus, the scale is more culturally specific to the United States; (c) involvement of large samples of subjects in a variety of sports improves the generalizability and the application of the scale; (d) the measurement properties of the coaching self-evaluation version were tested and improved; and (e) overall factor structures in determining the constructs of the scale were notably improved. Therefore, Riemer and Toon (2001) could have obtained different results since they sampled a United States population with a scale that was not as culturally specific to the United States as the scale used in the present study.

Second, Riemer and Toon (2001) noted low internal consistency estimates for preferred autocratic behavior (0.67) and perceived autocratic behavior (0.59), but chose to retain those dimensions based on the practice of prior research (Chelladurai, 1993; Chelladurai & Riemer, 1998). The current study was not affected by low internal consistency estimates for measures of leader behavior. Therefore, the enhanced scale reliability of the present study may have resulted in a more accurate measurement of leadership behavior, and therefore, a purer analysis of the relationship between leadership behavior congruency and athlete satisfaction.

Third, the present study incorporated an online survey while Riemer and Toon (2001) collected their data via a more traditional paper-and-pencil survey. Dillman (2000) noted several limitations of web-based surveys. First, not everyone is connected to the internet, so this survey method will not be optimal for all populations. This particular concern was somewhat alleviated given the sample of the present study, which was comprised of college students who are typically reliant upon the internet to facilitate their studies. However, even if connected to the internet, not all potential respondents are equally computer literate. Screen configurations may appear significantly different from one respondent to another depending on settings of individual computers. Finally, since e-mail addresses are not standardized, sampling of e-mail addresses is difficult (i.e., sometimes there is more than one e-mail address per respondent). In consideration of the above limitations, it is possible that data generated from online sources may differ from that collected by more traditional means.

Finally, Riemer and Toon (2001) only sampled participants in the NCAA Divisions I and II Championships. Since participation in the NCAA Championships is based on individual and team performance, Riemer and Toon's (2001) sample is likely composed of the most highly skilled athletes in each division. Given that group member ability, a member characteristic, is an antecedent of required and preferred leadership behavior in the MML, it is an important variable to consider when attempting to generalize the results of a study. The present sample incorporates tennis players from all three NCAA divisions, regardless of individual or team performance, therefore, it should be more generalizable to the average NCAA tennis player than the results of Riemer and Toon (2001).

The results of the present study open up numerous additional avenues of research activity. First, future research should explore a full test (i.e., required, preferred and perceived leadership behaviors) of the leadership behavior congruency hypothesis in the MML with the outcome of satisfaction. Second, future studies should analyze a variety of different sport settings. Chelladurai (1984) classified the moderating influence of type of sport as independent or interdependent and open or closed. Athletes who play sports classified as independent (e.g., golf, bowling, and track and field) rely very little on their teammates during their individual performance. On the contrary, sports classified as interdependent include basketball, football, volleyball, and soccer. Open sports (e.g., football, basketball, and baseball) are characterized by an unstable, changing environment, as opposed to closed sports (e.g., golf, swimming, and track and field), which feature relatively stable environments. Given this typology, the sport of tennis is characterized overall as an independent, open sport, and further research is needed to determine if the relationships noted in the current study exist when the sports vary according to task dependence and type. Third, Chelladurai (in press) recently asserted that the athlete's pursuit of pleasure versus his/her pursuit of excellence must be accounted for when examining leadership in sports. For athletes pursuing excellence, Chelladurai recommends that leaders provide demanding and directive leadership. Assuming collegiate tennis players are driven by the pursuit of excellence, one could argue that elevated displays of democratic behavior could perhaps be perceived by the athletes as a form of laissez-faire leadership. Such a "hands off" approach would not be appropriate for athletes motivated by the pursuit of excellence according to Chelladurai (in press), however, further research is needed to investigate this relationship.

In conclusion, the results of the present study highlight the importance of congruency of training and instruction behavior and autocratic behavior in the determination of athlete satisfaction. It is unknown why the congruency of the other four leadership behaviors that were studied (democratic behavior, social support, positive feedback, and situational consideration) did not have an impact on athlete satisfaction. Should future studies replicate the findings of the present study, the congruency hypothesis of the MML may need modification to reflect a higher degree of specificity regarding the relationship between specific leadership behavior and individual outcomes. In consideration of the psychological outcomes impacted by leadership behavior that were identified in the present study, tennis coaches should moderate their behavior in order to enhance satisfaction of their athletes. Furthermore, athletic administrators should be cognizant of the impact a coach's leadership behavior has on athletes when hiring new coaches. Finally, those involved in the training of new coaches should emphasize leadership behaviors that promote positive outcomes.

Acknowledgements

This study was funded in part by a Research Grant from the United States Tennis Association.

References

Al-Tahayneh, Z. (2003). The effects of coaches' behaviors and burnout on the satisfaction and burnout of athletes. Unpublished doctoral dissertation. Florida State University.

Berger-Gross, V. (1982). Difference score measures of social perceptions revisited: A comparison of alternatives. Organizational Behavior and Human Performance, 29, 279-285.

Berger-Gross, V., & Kraut, A. I. (1984). Great expectations: A no-conflict explanation of role conflict. Journal of Applied Psychology, 69, 261-271.

Chelladurai, P. (1978). A contingency model of leadership in athletics. Unpublished doctoral dissertation, Department of Management Sciences, University of Waterloo, Canada.

Chelladurai, P. (1984). Discrepancy between preferences and perceptions of leadership behavior and satisfaction of athletes in varying sports. Journal of Sport Psychology, 6, 27-41.

Chelladurai, P. (1990). Leadership in sports: A review. International Journal of Sport Psychology, 21, 328-354.

Chelladurai, P. (1993). Leadership. In R. N. Singer, M. Murphy, & L. K. Tennant (Eds.), Handbook on Research on Sport Psychology (pp. 647-671). New York: McMillan.

Chelladurai, P. (2006). Human Resource Management in Sport and Recreation (2nd ed.). Champaign, IL: Human Kinetics.

Chelladurai, P. (in press). Leadership in sports. In G. Tenenbaum & R. C. Eklund (Eds.). Handbook of Sport Psychology (3rd ed.). New York: John Wiley and Sons, Inc.

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

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 and Exercise Psychology, 10, 374-389.

Chelladurai, P., Malloy, D., Imamura, H., & Yamaguchi, Y. (1987). A cross-cultural study of preferred leadership in sports. Canadian Journal of Applied Sport Sciences, 12, 106-110.

Chelladurai, P., & Riemer, H. A. (1997). A classification of the facets of athlete satisfaction. Journal of Sport Management, 11, 133-159.

Chelladurai, P., & Riemer, H. A. (1998). Measurement of leadership in sport. In J.L. Duda (Ed.), Advances in Sport and Exercise Psychology Measurement (pp. 227-253). Morgantown, WV: Fitness Information Technology, Inc.

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

Churchill, G. A., Jr. (1979). A paradigm for developing better measures of marketing constructs. Journal of Marketing Research, 16, 64-73.

Cronbach, L. (1958). Proposals leading to analytic treatment of social perception scores. In R. Tagiuri & L. Petrullo (Eds.), Person perception and interpersonal behavior (pp. 353-359). Stanford: Stanford University Press.

Dillman, D. A. (2000). Mail and internet surveys (2nd ed.). New York: John Wiley & Sons. Inc.

Dwyer, J. M., & Fischer, D. G. (1988). Leadership style of wrestling coaches. Perceptual and Motor Skills, 67, 706.

Gordon, A. M. D. (1986). Behavioral correlates of coaching effectiveness. Unpublished doctoral dissertation. University of Alberta, Canada.

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

House, R. J. (1971). A path-goal theory of leader effectiveness. Administrative Science Quarterly, 16, 321-338.

House, R. J. & Dressier, G. (1974). A path-goal theory of leadership. In J. G. Hunt & L. L. Larson (Eds.), Contingency Approaches to Leadership (pp. 29-55). Carbondale, IL: Southern Illinois University Press.

Johns, G. (1981). Difference score measures of organizational behavior variables: A critique. Organizational Behavior and Human Performance, 27, 443-463.

Kelloway, E. K. (1998). Using LISREL for structural equation modeling: A researcher's guide. Sage Publications: Thousand Oaks, CA.

Kent, A., & Turner, B. (2002). Increasing response rates among coaches: The role prenotification methods. Journal of Sport Management, 16, 230-238.

Lord, F. M. (1958). The utilization of unreliable difference scores. Journal of Educational Psychology, 49(3), 150-152.

Lorsch, J. W., & Morse, J. J. (1974). Organizations and their members: A contingency approach. New York: Harper and Row.

McMillin, C. J. (1990). The relationship of athlete self-perceptions and athlete perceptions of leader behaviors to athlete satisfaction. Unpublished doctoral dissertation, University of Virginia.

Morse, J. J. (1976). Person-job congruence and individual adjustment and development. Human Relations, 28, 841- 861.

Mosier, C. I. (1951). Batteries and profiles. In E. F. Lindquist (Ed.), Educational Measurement. Washington, DC: American Council on Education.

Nunnally, J. C., & Bernstein, I. H. (1994). Psychometric theory (3rd ed.). New York: McGrawl-Hill.

Peter, J. P. (1981). Construct validity: A review of basic issues and marketing practices. Journal of Marketing Research, 19, 133-145.

Peter, J. P., Churchill, G. A., Brown, T. J. (1993). Caution in the use of difference scores in consumer research. Journal of Consumer Research, 19, 655-662.

Prakash, V., & Lounsbury, J. W. (1983). A reliability problem in the measurement of

disconfirmation of expectations. In R. P. Bagozzi & A. M. Tybout (Eds.), Advances in Consumer Research. Ann Arbor, MI: Association for Consumer Research.

Rice, R. W., McFarlin, D. E., & Bennett, D. E. (1989). Standards of comparison and job satisfaction. Journal of Applied Psychology, 74, 591-598.

Riemer, H. A., & Chelladurai, P. (1995). Leadership and satisfaction in athletics. Journal of Sport and Exercise Psychology, 17, 276-293.

Riemer, H. A., & Chelladurai, P. (1998). Development of the Athlete Satisfaction Questionnaire. Journal of Sport & Exercise Psychology, 20(2), 127-156.

Riemer, H. A., & Toon, K. (2001). Leadership and satisfaction in tennis: Examination of congruence, gender, and ability. Research Quarterly for Exercise and Sport, 72(3), 243-256.

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

Steiger, J. H. (1990). Structural model evaluation and modification: An interval estimation approach. Multivariate Behavioral Research, 25, 173-180.

Summers, R. J. (1983). A study of leadership in a sport setting. Unpublished master's thesis. University of Waterloo, Canada.

Vroom, V. H. (1959). Some personality determinants of the effects of participation. Journal of Abnormal and Social Psychology, 59, 322-327.

Wall, T. D., & Payne, R. (1973). Are deficiency scores deficient? Journal of Applied Psychology, 58(3), 322-326.

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

Wynne, B. E., & Hunsaker, P. L. (1975). A human information-processing approach to the study of leadership. In J. G. Hunt and L. L. Larson (Eds.), Leadership frontiers. Kent, OH: Kent State University.

Zhang, J., Jensen, B. E., & Mann, B. L. (1997). Modification and revision of the Leadership Scale for Sport. Journal of Sport Behavior, 20(1), 105-122.

Damon P.S. Andrew

The University of Tennessee

Address Correspondence to: Dr. Damon P. S. Andrew, Dept. of Exercise, Sport, and Leisure Studies, 1914 Andy Holt Avenue, HPER 349, University of Tennessee, Knoxville, TN 37996-2700, Office: (865) 974-8891, Fax: (865) 974-8981, E-mail: dandrew@utk.edu

Table 1. Means, Standard Deviations, and Internal Consistency Estimates
of Subscales

 [alpha]-
Scale and Factor Mean Std. Dev. level

RLSS preferred Democratic Behavior 4.40 1.07 0.88
RLSS perceived Democratic Behavior 5.13 0.86 0.86
RLSS preferred Positive Feedback 5.17 1.23 0.93
RLSS perceived Positive Feedback 5.94 0.78 0.85
RLSS preferred Training and Instruction 4.84 1.30 0.93
RLSS perceived Training and Instruction 5.84 0.81 0.86
RLSS preferred Situation Consideration 5.21 1.23 0.91
RLSS perceived Situation Consideration 6.16 0.73 0.84
RLSS preferred Social Support 4.98 1.32 0.91
RLSS perceived Social Support 5.40 0.87 0.77
RLSS preferred Autocratic Behavior 3.51 0.99 0.71
RLSS perceived Autocratic Behavior 2.89 1.04 0.82
ASQ Individual Performance Satisfaction 4.80 1.52 0.88
ASQ Personal Treatment Satisfaction 5.40 1.51 0.95
ASQ Team Performance Satisfaction 4.79 1.53 0.91
ASQ Training and Instruction Satisfaction 4.70 1.73 0.95

Note: RLSS = Revised Leadership Scale for Sport; ASQ = Athlete
Satisfaction Questionnaire

Table 2. Fit Indices of Subscales

Fit Index RLSS preferred RLSS perceived ASQ

RMSEA 0.072 0.068 0.095
NFI 0.94 0.87 0.97
NNFI 0.97 0.93 0.97
CFI 0.97 0.93 0.98

Note: RLSS = Revised Leadership Scale for Sport; ASQ = Athlete
Satisfaction Questionnaire

Table 3. Summary of Sign cant Leadership Behavior Congruency Outcomes

 [R.sup.2]
Leadership Behavior Dependent Variable increase P-value

Training and Personal Treatment .011 .028
 Instruction Satisfaction
Autocratic Individual Performance .030 .004
 Satisfaction
Autocratic Personal Treatment .019 .015
 Satisfaction
Autocratic Team Performance .032 .004
 Satisfaction
Autocratic Training, and .040 .001
 Instruction
 Satisfaction