ENVIRONMENTAL, SOCIAL, AND PSYCHOLOGICAL CORRELATES OF PHYSICAL ACTIVITY IN ANGLO- AND MEXICAN-AMERICAN ADOLESCENTS

The purpose of this study was to assess the relationship of psychological, social, and environmental variables with physical activity (PA) in a Mexican-American (MA; 56 boys; 64 girls) and Anglo-American (AA; 49 boys; 45 girls) sample of adolescents (mean age = 12.1 years) from low-to-moderate SES families. PA was assessed by both the 7-day Physical Activity Recall (PAR) and Caltrac accelerometer. The PAR was administered in the child's home by trained interviewers on two occasions, approximately 6 months apart. For this study, a vigorous PA category was used. The Caltrac was worn for up to 8 days. Total activity counts for all days divided by hours worn produced the score used in analyses. An interviewer-paced survey was administered in the adolescent's homes to assess 5 psychosocial, 10 social, and 14 environmental variables previously identified to be associated with youth PA. Test-retest reliabilities (ICC) on variables over 2 weeks ranged from 0.48 to 0.88. Internal consistencies (Cronbach's Alpha) for 6 scales ranged from .48 to .68. Hierarchical multiple regression models were conducted for 4 subgroups (males, females, AA, MA). Regressions explained more variance in PA assessed by the PAR. (3–24%) than by the Caltrac (7–16%). For PAR, demographic variables were significant for AA and MA, explaining nearly 8% of the variance. Skinfolds for AA and sex (girls) for Mexican-Americans were significant contributors, with partial r's of −.29 and −.25, respectively. The psychological and environmental variables were significant only for AA (7% of the variance each) and social block significant only for girls (3% of the variance). Parent transportation was a significant contributor for girls. Regarding the Caltrac, demographic variables were not significant contributors to any model (< 5% of variance). Psychological variables were significant in 2 of 4 models (Boys, MA), explaining between 5% and 12% of the variance. Overcoming barriers for boys and physical self-perception for Mexican-Americans contributed significantly to their models with partial r's between .24 and .28. Social variables were not significant for girls, MA or AA. While numerous factors were found to be associated with adolescent PA, the regression models explained little of the variance in activity levels and they were different for the self-report and objective measures. Further study of factors associated with PA are warranted, and multiple methods for assessing PA should be used. Supported by NIH HL52449.