Skip to main content
ARS Home » Plains Area » Houston, Texas » Children's Nutrition Research Center » Research » Publications at this Location » Publication #229724
Title: How Many Days Was That? We're Still Not Sure, But We're Asking the Question Better!

Author
item Baranowski, Thomas
item MASSE, LOUISE - U OF BRITISH COLUMBIA
item RAGAN, BRIAN - U OF NORTHERN IOWA
item WELK, GREG - IOWA STATE UNIV

Submitted to: Medical Science Sports Exercise
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 7/1/2008
Publication Date: 7/1/2008
Citation: Baranowski, T., Masse, L.C., Ragan, B., Welk, G. 2008. How many days was that? We're still not sure, but we're asking the question better! Medical Science Sports Exercise. 40(7 Suppl):S544-S549.

Interpretive Summary: A common problem in measuring physical activity using an acclerometer or pedometer is the unreliability or error introduced from day to day differences in physical activity. Typically this source of error is dealt with by taking an average over multiple days. What isn't clear from the literature is the optimal number of days of activity assessment that gets high reliability, but not so many days to constitute a burden to participants, and make them not want to participate. Two kinds of studies that could provide useful data on this point include: tests of variance reduction methods, and the generalizability model. More research needs to be done to identify the optimal number of days for assessment of physical activity.

Technical Abstract: Unreliable measures limit the ability to detect relationships with other variables. Day-to-day variability in measurement is a source of unreliability. Studies vary substantially in numbers of days needed to reliably assess physical activity. The required numbers of days has probably been underestimated due to violations of the assumption of compound symmetry in using the intraclass correlation. Collecting many days of data become unfeasible in real-world situations. The current dilemma could be solved by adopting distribution correction techniques from nutrition or gaining more information on the measurement model with generalizability studies. This would partition the variance into sources of error that could be minimized. More precise estimates of numbers of days to reliably assess physical activity will likely vary by purpose of the study, type of instrument, and characteristics of the sample. This work remains to be done.