Factor Validity and Invariance of the Original and Modified Physical Activity Enjoyment Scales

Abstract

PURPOSE: Enjoyment is cited as an important correlate of physical activity, but is often measured with a single-item or group of items with no previous evidence for reliability or validity. The Physical Activity Enjoyment Scale (PACES) and the modified PACES (PACES-M) show potential as standardized measures of physical activity enjoyment. The purpose of this study was to evaluate the factor structure of the PACES and PACES-M and test measurement invariance across gender and race. METHODS: Participants (N = 1023; mean age = 19.60 ± 1.55 years) were recruited from two universities. Data were collected by web-based survey. During pilot testing no significant differences were found between responses to a paper & pencil and web version of the survey. Confirmatory factor analysis was used to examine factor validity and invariance. RESULTS: Findings supported the hypothesized factor structure of the PACES-M (CFI = 0.977; RMSEA = 0.055), but not the PACES (CFI = 0.844; RMSEA = 0.112). Based on exploratory analyses, a two factor model was found to represent the PACES (CFI = 0.940, RMSEA = 0.078). The factor invariance between black & white and male & female participants was supported for both scales. While statistically significant, differences between genders for PACES (Male = 102.84; Female =97.14) and PACES-M (Male = 67.32; Female = 65.64) were small and may be of little practical importance (effect size = 0.16 & 0.32). Mean scores on the PACES were also significantly higher for white (99.93) compared to black (96.10) participants, but these differences were small (effect size = 0.21). Correlations among the PACES-M and PACES were moderate to large (0.59-0.76). CONCLUSION: The results support the hypothesized single factor model of the PACES-M and a two-factor model for the PACES. This two-factor model should be explored in greater detail to determine how it might impact understanding of the enjoyment construct. This study provides evidence of factor invariance across race and gender in young adults and indicates substantial but not complete construct convergence between scales. Continued validity evidence from diverse populations involved in various physical activities will improve confidence in using PACES and PACES-M scores to examine longitudinal change, intervention effects, and mediation of physical activity behavior.