For analyses:

the dependent variable physical activity was dichotomized into meeting recommendations or not meeting recommendations for physical activity

U.S census categories of inside urban, outside urban rural farm and rural non-farm were collapsed in to urban, suburban and rural

Dichotomous income grouping similar to those used in other national studies were used to allow comparability ( <20K= lower, >20K higher)

Logistic regressions of meeting recommendations for six categories (3 urbanicity X 2 income) were adjusted for potential confounders: :age, gender, education and race OR’s and 95% CI reported.