Marilyn Johnson-Kozlow1, Georg E Matt. 1. Joint Doctoral Program, Public Health, Epidemiology, San Diego State University/University of California at San Diego, San Diego, CA 92093-0607, USA. mfjohnson@projects.sdsu.edu
Abstract
BACKGROUND: Research on the cognitive processing of survey questions suggests that the distribution of walking bouts (e.g., mean, skew) may influence retrospective self-report responses. METHODS: Undergraduate women (N = 209) kept a weekly log of their walking behavior, then responded to four different self-report measures of walking [National Health Interview Survey (NHIS), Context, Fuzzy-Typical, and Fuzzy-Maximum]. The logs were evaluated to determine which walking characteristics best summarized their walking behavior. Principal components analysis yielded three components of walking behaviors: Frequency, Duration, and Maximum. Multiple regression was used to determine if the walking components differentially predicted the four self-report measures of walking. RESULTS: The Maximum walking component was a significant predictor of all self-report measures of walking. The Context score was also predicted by the Frequency Component. The Frequency and Duration walking components did not predict the NHIS or Fuzzy self-report measures independently of Maximum walking. CONCLUSIONS: Frequency-by-duration self-report measures of walking appear to be more sensitive to the maximum rather than the frequency or duration of walking behavior. Measurement of the frequency of bouts may be best accomplished by providing a framework of recall cues around which walks of different types occur.
BACKGROUND: Research on the cognitive processing of survey questions suggests that the distribution of walking bouts (e.g., mean, skew) may influence retrospective self-report responses. METHODS: Undergraduate women (N = 209) kept a weekly log of their walking behavior, then responded to four different self-report measures of walking [National Health Interview Survey (NHIS), Context, Fuzzy-Typical, and Fuzzy-Maximum]. The logs were evaluated to determine which walking characteristics best summarized their walking behavior. Principal components analysis yielded three components of walking behaviors: Frequency, Duration, and Maximum. Multiple regression was used to determine if the walking components differentially predicted the four self-report measures of walking. RESULTS: The Maximum walking component was a significant predictor of all self-report measures of walking. The Context score was also predicted by the Frequency Component. The Frequency and Duration walking components did not predict the NHIS or Fuzzy self-report measures independently of Maximum walking. CONCLUSIONS: Frequency-by-duration self-report measures of walking appear to be more sensitive to the maximum rather than the frequency or duration of walking behavior. Measurement of the frequency of bouts may be best accomplished by providing a framework of recall cues around which walks of different types occur.