Distinguished Professor of Statistics
Iowa State University
“Physical Activity Measurements and Compliance with CDC Guidelines”
PAMS (Physical Activity Measurement Study) collected information about physical activity on over 1,000 randomly selected persons in three Iowa counties between 2010 and 2013. Individuals were asked to wear an arm monitor and also self-report their PA in two forms: via (replicated) 24-hour recalls administered by trained interviewers or via a self-administered frequency instrument. We do not discuss the self-administered frequency instrument today.
We estimate the population probability of compliance with CDC guidelines using a design-based or a model-based approach and starting from the minutes of physical activity measured by the arm band. The sampling model is a negative binomial hurdle model (to accommodate the large proportion of zeros in the data) and we adopt a Bayesian approach to estimation.