0277 Improving the Validity of Self-Reported Sleep Duration: Experimental Comparisons of Question Format with Actigraphy

Abstract

Abstract Introduction Accurate measurement of sleep is a challenge for researchers and practitioners. Objective measures (i.e., polysomnography, actigraphy) are expensive and time-consuming. Thus, the practical approach is self-report – with the cost of reduced validity. The purpose of this actigraphy ­experiment was to compare the validity of differently ­formatted self-reported sleep duration measures. It was hypothesized that question formats which encourage duration reporting to the nearest 15 minutes would lead to improved validity. Methods Adults (n=56, 73.2% female, 80.4% white, mean age = 21.3 years, SD = 4.18) were recruited through university email, a psychology participant pool, and word of mouth. Sleep was recorded via actigraphy for 7 days. Participants were randomly assigned to respond via computer to one of three versions of the Pittsburgh Sleep Quality Index (Buysse et al., 1989). Specifically, they reported hours of sleep per night in one of the following formats: 1) Open-ended response with standard instructions 2) Open-ended with modified instructions to “round to the nearest 15-minutes”, or 3) Standard instructions with drop-down menus to select hours and minutes (15-minute increments). Results Self-reported and actigraphy sleep duration were strongly correlated, r = .63, p < .001. Average self-reported duration (7.05 hours) was significantly longer than actigraphy average (M = 6.65 hours), t(55) = 3.27, p < .01, d = 0.44. There were no significant differences across the three self-reported sleep duration formats, F(2, 53) = 0.02, p = .98, η2 = .00. As expected, the open-ended standard format had the weakest correlation with actigraphy (r = .47), whereas the open-ended with modified instructions (r = .74) and drop-down menu (r = .68) versions were better predictors of actigraphy duration. The effect sizes for the improvement (q = .44, .32) were medium in magnitude (Cohen, 1969). Conclusion The results suggest that researchers and practitioners should consider the format of self-report sleep duration questions. We found formats that encourage participants the reporting of more precise estimates to more strongly predict actigraphy. Additional data is needed to replicate these findings, and for a better understanding of the factors that influence the magnitude of agreement between objective and subjectively reported sleep duration. Support (if any)