Test-retest reliability of the twin cities walking survey

Background: The Twin Cities Walking Study measured the associations of built environment versus socioeconomic and psychosocial variables with total physical activity and walking for 716 adults. Methods: This article reports on the test-retest reliability of the survey portion of the study. To test the reliability of the study measures, 158 respondents completed measures twice within 1 to 4 weeks. Agreement between participants' responses was measured using Pearson r and Spearman rho, and kappa statistics. Results: Demographic questions are highly reliable (R > .8). Questions about environmental and social features are typically less reliable (rho range = 0.42-0.91). Reliability of the International Physical Activity Questionnaire (last 7 days version) was low (rho = 0.15 for total activity). Conclusions: Much of the survey has acceptable-to-good reliability. The low test-retest reliability points to potential limitations of using a single administration of the IPAQ to characterize habitual physical activity. Implications for sound inference are accordingly complicated.