Measurement bias in adolescent sleep duration: Comparison of self-reported and actigraphy-assessed sleep duration among adolescents in the Future of Families and Child-Wellbeing Study

Objectives: To assess differences and potential measurement bias in adolescent sleep duration, we compared self-report and actigraphy-assessed sleep duration in the Future of Families and Child Wellbeing Study (age 15 wave). Methods: Participants (N = 634; mean age 15.4 years; 51% Black, 30% Hispanic, and 18% White) self-reported typical weekday and weekend bedtimes and waketimes before wearing a wrist actigraph for 1 week. Linear regression models estimated concordance between self-reported and actigraphy-assessed sleep duration overall and stratified by race, ethnicity, and socioeconomic status (primary caregiver education and household income-to-poverty threshold ratio). Agreement of sleep duration categories [short (<8 hours), recommended (8-10 hours), and long (>10 hours)] was estimated using Kappa statistics. Results: Self-reported sleep duration overestimated actigraphy-assessed duration by 27.87 minutes (95% CI: 34.96, 20.78) after adjustment. Agreement between duration categories was 73% (ƙw = 0.12, 95% CI: 0.07, 0.18). The mean difference was larger on weekends (74.3 ± 144.7 minutes) than weekdays (27.5 ± 92.7 minutes). Within-group analyses showed self-reported duration overestimated actigraphy-assessed duration for Black adolescents and those with lower socioeconomic status (i.e., primary caregivers with a high school education or less). Conclusions: Reliance on self-report may misclassify sleep duration and underestimate insufficient sleep, particularly in marginalized groups, underscoring the need to understand measurement bias when objective measures are unavailable.