A Causal Modeling Analysis of Sociodemographics and Moderate to Vigorous Physical Activity Behavior of American Adults

Abstract Confusion still exists about the relationships between sociodemographics and adult physical activity behavior. In this paper, data from a national sample of 19,110 adults is examined to determine which of the sociodemographic variables are associated with physical activity behavior, their relative importance, and why these associations may exist. Once one controls for multicollinearity among the sociodemographic variables only age, education, and income maintain any substantive correlation. However, the amount of variance explained is small (R 2 adj = 8%). Two causal models are presented to account for the education: physical activity variance. The relationship is believed to be a spurious one in that there is some other third variable(s) which is speculated to affect both schooling and physical activity behavior. Exposure to sports and physical activity in the schools, though, may enhance the skills necessary for many activities providing an argument for a direct effect of education. Understanding this spurious/direct association is important for policy decision making and further research is warranted.