Abstract
Objectives: To examine the associations between demographic, socioeconomic, and health-related factors and self-reported sleep duration in a nationally representative sample of U.S. adults, with a novel focus on the conditional variability in sleep, defined as variation in sleep duration among individuals with similar backgrounds.
Methods: Data were drawn from eight waves (2011-2020) of the Behavioral Risk Factor Surveillance System, including 1,128,505 adults. A two-step regression framework was used: first, to estimate average sleep duration; second, to model inequality using the conditional variance of sleep. Shapley decomposition quantified the contribution of each factor. Analyses were stratified by sex, race-ethnicity, and separate models were estimated for short (<7 h) and long (≥10 h) sleep duration.
Results: Average sleep duration was 6.99 h; 35% of respondents were short sleepers and 5.2% long sleepers. Mental and general health were the strongest predictors of both lower sleep and higher conditional variability. The conditional variability in sleep was greater among those with poor health, lower education, unemployment, and among Black and Asian individuals. Sex differences were modest, but mental health had a stronger association with sleep among males. Asian respondents experienced declining sleep with age, unlike other groups. Short sleep was more common among younger adults, racial minorities, smokers, and the employed; long sleep was linked to poor health and unemployment.
Conclusions: Both average sleep and its conditional variability reveal important public health patterns. Disadvantaged groups not only sleep less but also experience more irregular sleep. Policies should address sleep health disparities through integrated mental healthcare and targeted, culturally responsive interventions.