Factorial Validity Of The Coach-athlete Relationship Questionnaire With a Sample Of Team Sports Performers

Introduction
The aim of the present study was to provide psychometric evidence regarding the factorial construct validity of the two
versions of the Coach-Athlete Relationship Questionnaire (CART-Q self-perceptions) [1] CART-Q meta-perceptions
[2]). These measurement scales were developed to measure the quality of the coach-athlete relationship through the
constructs of Closeness, Commitment and complementarity (3Cs).

Methods
A total of 591 British athletes (414 male, 177 female) performing in team sports (hockey, football, rugby, basketball,
volleyball) participated in the study. Age of the participants ranged from 16 to 38 years. Different levels of participation
were included: premiership (90), national (102), county (125), regional (240) and recreational (34). Also participants
had a varied length of relationship with coaches ranging from as little as a week to more than 3 years. Both versions of
the CART-Q were administered before a training session. Confirmatory Factor Analysis using EQS, was employed to
examine the validity of the instrumentations’ factorial structure. A hierarchical second-order factor model in which the
three constructs of closeness, commitment and complementarity were subsumed was tested. The same fit indexes were
employed as in Jowett and Ntoumanis’ (2004) study for comparison reasons.

Results
Table 1 presents the fit indexes of the hierarchical model tested. The S-Bχ2 for both models were statistically significant
whereas the CFI (Robust) and NNFI were approached and reached 0.95 [3]. Factor loadings of the three first-order
factors in both models were high and ranging from 0.93 to 0.99 and were statistically significant. Results also illustrated
that all items loaded as expected on their designated constructs providing evidence for convergent validity. The items of
both questionnaires recorded squared multiple correlation (R2) values beyond 0.50 except for two items (both of which
loaded on the Commitment scale).

Discussion / Conclusions
The main purpose of the study was to provide further evidence of the factorial structure of the self- and meta-perception
versions of the CART-Q utilising a sample of team sport performers. The information presented shows that the CARTQ
(both versions) has adequate psychometric properties. The low R2 values recorded for the two items of Commitment
may have resulted due to the composition of this particular sample which consisted of over 50% of sport performers
with less than 6 month relationship with coach.

References
[1]. Jowett, S., & Ntoumanis, N. (in press). Scand J of Med and Sc in Sp.
[2]. Jowett, S. (2002). The Coach - Athlete Relationship Questionnaire and DyadMaps Manual (Monograph) Staffs ,
Uni, Sch of Health.
[3]. Hu, L., & Bentler, P.M. (1999). Str Eq Model, 6, 1-55.

NOTA: O texto com a iconografia está no anexo.

Arquivo