Development and Validation of Self-Determination in Physical Activity Participation Inventory

Purpose: The purpose of this study was to develop and validate an instrument that assesses students' motivation for physical activity participation and adherence. Methods: A careful study of the self-determination theory and a review of research on application of this theory in a physical education setting guided the development of items on the Self-Determination in Physical Activity Participation Inventory (SDPAPI). The SDPAPI was revised repeatedly based on the experts' comments in the field. Finally, the SDPAPI was consisted of three parts: demographic information, physical activity participation, and 19-item levels of motivation rated on a five-point rating scale. Four hundreds sixty-eight fifth-grade students from two local elementary schools anonymously and voluntarily completed the inventory during their regular physical education classes. To examine the construct validity of the SDPAPI, the confirmatory factor analysis (CFA) were conducted on the initially hypothesized six-factor model and two alternative six-factor models. To analyze the internal consistency of the SDPAPI, the Cronbach alpha reliability coefficient was used. Analysis/Results: Evaluation of the model fit of the hypothesized model 1 indicated a mixed model fit results. The goodness-of-fit indices (CFI=.88 and IFI=.88) were below the criteria of .90, indicating poorly model fit, while the “badness-of-fit” indices (RMSEA=0.07 and SRMR=0.07) were lower than .08, indicating good model fit. Inspection of the factor loadings revealed the factor loadings of the two observed items lower than 0.5. Based on the modification indices, the two items were re-specified to a different factor, respectively. Next, evaluation of the modified model 2 revealed mixed fit results also. The goodness-of-fit indices remained the same as above (CFI=.88 and IFI= .88). The RMSEA and SRMR was .08, indicating good fit results, respectively. Inspection of the modification indices suggested adding an error covariance between Intrinsic Item 2 and Intrinsic Item 3, between Intrinsic Item2 and Introjected Item 5, and between Introjected Item 2 and Intrinsic Item1. Accordingly, the model 2 was modified again. Finally, the results indicated that the modified model 3 provided the best model fit to the data (CFI=.93, IFI=.93, exceeding .90 and SMSEA=.06, SRMR=.06, indicating excellent data fitting). In addition, the factor loadings of the 19 observed items were all above 0.50 at significant level (p<.05), indicating these observed items adequately defined the six factors. The Cronbach alpha reliability coefficient was .78 for the total scale of the inventory, demonstrating a good internal consistency.

Conclusions: It was concluded that the SDPAPI was a theoretically sound and psychometrically supported measure.