USE OF PRINCIPAL COMPONENTS ANALYSIS TO IDENTIFY COMPOSITE INDICES OF PHYSICAL ACTIVITY

71 When multiple measures of physical activity (PA) are analyzed individually, the significance of treatment effects are often found to be inconsistent. A composite score composed of several measures should provide a more reliable indicator of physical activity than any single measure. This study demonstrates the use of principal components analysis (PCA) to derive conceptually meaningful composite scores of PA. PCA uses weighted linear combinations of all items to create orthogonal composites. The study sample consisted of 547 university seniors who completed 5 self-report measures of PA: 7-day recall, YRBS college survey, stages of change, and a list of 25 PAs from NHIS. From the measures, a total of 14 individual indices were derived. The sample was 56% female, 60% Caucasian, 18% Latino and 12% Asian/Pacific Islander with a mean age of 24 years. PCA resulted in 2 interpretable factors. Variables with the strongest weights on component 1 (Cl) included vigorous PA, stretching, strengthening, toning, flexibility, and exercise stages of change(eigenvalue = 3.96). Strongest weights on component 2 (C2) included moderate, walking, and housework PA items (eigenvalue = 1.80). Providing evidence of validity, significant correlations were observed for Cl with diastolic blood pressure (r=-0.24, p<0.001) and step test recovery heart rate (r=-0.382, p<0.001) and for C2 with systolic blood pressure (r=-0.16, p<0.001). The findings demonstrate the appropriateness and utility of PCA for integrating multiple measures of physical activity.