James A. Bovaird
Date of this Version
Miramonti, A.A. (2018). Evaluation of the Feasibility of a Two-Method Measurement Design for the Assessment of Healthy Physical Activity Behavior in Youth (Master’s thesis).
Purpose: Assess the reliability and validity of self- and parent-report survey responses regarding physical activity (PA), sedentary behaviors, and PA self-efficacy and determine if these data can be combined with objective physical activity monitor data to model the latent construct healthy physical activity behavior (HPAB).
Methods: 126 underserved 4th-5th grade students participated in a 12-week after-school nutrition, cooking, and physical activity program (WeCook: Fun with Food and Fitness). Participants and parents (n=103) completed surveys pre- and post-program and participants wore PA monitors for one week at PRE and POST. Unidimensionality and internal consistency reliability were assessed for survey measures and objective PA measures (ST=step counts, FL=floors climbed) and predictive validity of survey measures was assessed through correlation with ST and FL.
HPAB was modeled using z-scores (standardized across time, averaged) for youth self-reported physical activity (YPA), youth PA self-efficacy (YSE), parent-reported youth PA and sedentary behavior (AS), ST, and FL. Metric and intercept invariance were established across time and between groups selected for assessing construct validity (gender, grade, socioeconomic status, weight status, school, minority, season, grant year).
Results: ST, FL, YSE, and AS were unidimensional. For ST coefficient α was 0.735 (PRE) and 0.805 (POST), for FL α was 0.686 and 0.684 (PRE and POST), for YSE, α was 0.829 (PRE) and 0.897 (POST), and for AS α was 0.545 (PRE) and 0.729 (POST). YPA was most predictive of the objective PA measures and was correlated with ST at PRE and POST and FL at PRE (pp>0.05). HPAB exhibited measurement invariance across time and between groups of interest and some evidence for latent construct validity based on nomothetic span and construct representation was established.
Conclusions: This study establishes some evidence supporting the feasibility of modeling HPAB using survey and objective measures of PA in youth.
Advisor: James A. Bovaird