|
nif:isString
|
-
Participants were 57 female volleyball and 39 male handball players from the highest two divisions in the Belgian league. The female volleyball players were recruited from nine teams, seven of the eight Flemish first division teams and two teams in the second division. The male handball players were recruited from seven teams, all five Flemish first division teams and two second division teams. Spread over the 16 teams, 96 of the 204 athletes completed the questionnaires (i.e., 47.1%). However, the response rate for the volleyball players was significantly higher (i.e., 57.6%) than that of the handball players (i.e., 37.1%). This can be attributed to two reasons. First, relatively more foreign athletes played in the male handball teams. Because the questionnaire was drawn up in Dutch, the non-Dutch-speaking players did not participate in the study, this could have led to a lower response rate. Second, the coach of the national female volleyball team was one of the researchers, which might have stimulated them to participate. We would emphasize that we have followed sound scientific procedures to assure that our research was performed according to the strictest possible ethical standards. First, at the start of the study complete anonymity was assured. Players chose a codename to fill in the questionnaire and none of the researchers could link names to the data. Second, none of the players who filled in the questionnaires was part of the national team or trained under supervision of the national team coach. Finally, players only filled in questions about their own coach, none of the questions handled about the coach of the national team. Participants were clearly informed about these procedures to avoid biased results. The players were on average 23.15 (SD = 4.47) years old. They had been playing volleyball or handball on average for 13.58 years (SD = 4.91), worked with the same coach for 2.48 years (SD = 3.16) and trained 2.92 times (SD = 0.91) per week. Those demographical variables did not differ between the female volleyball and the male handball players.
Data were collected with two different questionnaires. The general questionnaire was used the first week of the study and gathered information about the players’ demographics and the general coaching style. The game-specific questionnaire had to be completed after each game during six consecutive weeks and gathered information about athletes’ perceived justice of the coach, decision justifications, players’ status, and the game result. Items were answered on a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree).
General questionnaire: Overall coaching style: Need support of the coach (9 items): In line with Deci and Ryan [12] we considered need support of the coach as a concept composed of three subscales: (a) autonomy-support, (b) competence-support and (c) relatedness-support. To assess the concept of need support we selected nine items from two different questionnaires which were already translated into Dutch and used in the sport setting by De Backer et al. [4,6]. An example of such an autonomy support item was: “My coach let me partially decide on my own training program”. An example of a competence support item is: “I have the impression that my coach believes in my abilities”. Finally, and example of a relatedness support item is: “My coach shows that he/she cares about me as a person”.
Control of the coach (3 items): To assess athletes’ experience of coaches’ controlling coaching style we included items that tapped coaches’ coercive behavior. Based on the study of Pelletier and colleagues [40] three items were back-translated into Dutch and adapted to fit the elite team sport context. An example item is: “My coach makes me feel guilty when I do not meet his/her expectations”.
The 4-item scale was based on research about the impact of decision justifications within business settings [13]. Greenberg’s items questioned the explanation, communication, and the clearness of the reasons of pay cuts. We transformed them to fit the team sport setting and queried the clarity and the communication of the tactical decisions and the player’s selection (e.g., “My coach explained his/her tactical decisions”).
This section was based on research of Colquitt [41] and adapted to the sport context by De Backer et al. [4]. Justice was considered as a concept composed by four subcategories of three items: personal distributive justice, group distributive justice, personal procedural justice, and group procedural justice. The personal distributive justice category questioned the perceived justice of the individual playing time during the game (e.g., “My coach rewarded me with a fair amount of playing time taking into account my contribution to the team”). The group distributive justice category informed about the outcomes for the group as a whole (e.g., “My coach based his selection of the starting team on the talent and competence of the players”). The personal procedural justice category referred to the perceived justice of the procedures used by the coach for the players’ individual treatment (e.g., “The evaluation of my performance during the game and/or training was supported by reliable information”). Finally, the group procedural justice category questioned the procedures the coach used towards the group as a whole (e.g., My coach consequently substituted players when they were underperforming). Despite the fact that people can distinguish between these sources of justice, Greenberg [42] (p. 211) suggested that individuals form justice perceptions based on a ‘‘holistic judgment in which they respond to whatever information is both available and salient”. Moreover, researchers have shown that individual’s justice perceptions may not be accurately evaluated when there is a focus on the various dimensions of justice [43]. Finally, De Backer et al. [4] indicated that the different justice subcomponents form a latent variable of overall justice, which was linked to identification and cohesion in a team sport context. Therefore, we combined the four subcomponents and shifted towards an overall justice judgment.
Status of the athlete (1 item): With a single item, we questioned whether the player was a member of the starting team (i.e., “During last game I was a starting player / a substitute”). Result of the game (1 item): With a single item, we questioned the result of the game (i.e., “Last game our team won/lost.”). The head coaches of the teams were contacted and informed about the aim of the research. Only when the coach gave his/her permission, a research assistant informed and invited the players to participate by means of a brief verbal presentation during or after a training session. The players who agreed to participate, were sent an e-mail in which they were informed about the timing and the objectives of the web-based questionnaires. Furthermore, they were guaranteed full anonymity. We collected data during six consecutive mid-season weeks. The first assessment took place after the first game of these six weeks with the aim of assessing general measures (i.e., demographical information and the coaching style). The other assessments took place after each game during six consecutive weeks and collected game-specific measures (i.e., within-athletes measures including perceived justice, decision justifications, the game specific status of the player, and the game result). An email, with the link to the questionnaire, was sent on a weekly base on Sunday. Moreover, the players who had not completed the questionnaire on Wednesday, received a reminder email. Finally, when the questionnaire was not filled out by Friday, the players were reminded a last time by phone. As part of a bigger PhD project [44], the current study was approved by The Leuven International Doctoral School Biomedical Sciences. More specific, by the Doctoral Committee of Kinesiology, Rehabilitation Sciences & Physiotherapy. The research conducted was in line with the ethical principles of the American Psychological Association (APA). No rewards were given for participation, informed consent was obtained from all participants, and confidentiality was assured.
First, we performed Confirmatory Factor Analysis (CFA) to test whether the back-translated and slightly adapted measurement instruments fitted the already validated structure of the original measurement models. The indices we used to evaluate overall model fit were: comparative fit index (CFI), Tucker–Lewis Index (TLI), and root mean square error of approximation (RMSEA A CFI and TLI value higher than .90 [45], and a value lower than .08 for the RMSEA [46] indicate an acceptable fit of the model. Consequently, internal consistencies of the different scales were calculated. Finally, Hierarchical Linear Modeling (HLM) with R was conducted to examine the fluctuations of perceived justice and the effect of the above-mentioned determinants on athletes’ perceived justice of the coach in a longitudinal perspective. Our game-to-game measures (i.e., level 1 or within-athlete variance) were nested into individuals (i.e., level 2 or between-athlete variance). Therefore, it was crucial that our statistical model explicitly recognized this hierarchical structure and that variation at within and between athlete level was allowed for in the analyses. Multilevel modelling (using R) provides an efficient way of doing this. Furthermore, these multilevel models allowed us to model cross-level interactions (Hypothesis 2a and 2b). An overview of the models can be found in Table 1. First, we compared a model with only level 1 variance to a model with both level 1 and level 2 variance and checked whether the addition of the second level significantly improved our model. We computed the Intraclass Correlation Coefficient (ICC) to examine the degree of variance underlying the within and the between-athlete level.
Table data removed from full text. Table identifier and caption: 10.1371/journal.pone.0205559.t001 The multilevel models aiming to address the different hypotheses of the study. In order to check Hypothesis 1a and 1b, we tested whether the coaching style (i.e., level 2) predicted athletes’ game-to-game perceived justice of the coach. We entered both the need supportive and the controlling coaching style (grand mean centered) as predictors at level 2 (i.e., between-athletes). Regarding Hypothesis 2a and 2b, the weekly decision justifications of the coach (group mean centered), were added to our previous model as a predictor at level 1 (within-athletes). Furthermore, the model was completed by adding the two cross-level interaction effects between both coaching styles one the hand and decision justifications on the other hand. To examine Hypothesis 3a and 3b we extended the model of the previous step. Two game-specific circumstances were entered as predictors at level 1, namely the status of the player and the result of the game.
|
![[This page as RDF]](http://data.gesis.org/somesci/static/rdf-icon.gif){kind=icon}