Exploratory factor analyses and initial inspection of coaches' responses to a survey of adult-oriented coaching practices

Abstract

Qualitative research has shown that adults athletes have particular preferences for how they want to be coached (Callary, Rathwell & Young, 2015; Ferrari, Bloom, Gilbert & Caron, 2016), which are often discrepant from dominant approaches with younger cohorts (Callary, Rathwell & Young, 2017). A prior presentation outlined the process by which information about adults' preferences was articulated as items and vetted for content validity. This study examined the initial factor structure of these survey items using 185 coaches' (91 m, 94 f; M yrs coaching adults = 12.9, derived from USA, Canada, UK and Australia) responses relating to adult-oriented sport coaching practices. Coaches responded to 51 items representing 13 initial factors (6 factors for 'accommodating adults', 7 factors for 'coaching behaviours and attributes'). An initial exploratory factor analysis (MLE with oblique rotation, forced 13-factor model) fell short of criteria for good fit, chi square (690) = 1072.3, p = .00, CFI = .85, TLI = .73, RMSEA = .06 [90% CI = .060 - .075], SRMR = .034. Factor loadings indicated 13 flagrant items and modification indices suggested many issues with cross-loading. Consequently, exploratory structural equation modeling was employed to refine the model, resulting in seven factors, with reduced multicollinearity and divergent validity. Discussion focuses on the nature of the resulting survey instrument, including: (a) how it captures pertinent categories for Masters coaching self-assessment; and (b) its utility for examining congruency associations between coach-report and athletes' perceptions of adult-appropriate coaching behaviours, and whether congruency is associated with quality sport experiences.