Too Much Sleep is Bad for Your Health

Abstract

This study determines whether long-sleep (≥ 9 hours per night) is causally related to lower self-rated health. Lewbel instrumental variable estimation is used with data on adults from the 2010 National Health Interview Survey (n = 26,607).

Lewbel instrumental variable estimation enables the derivation of valid and reliable instrumental variables from the heteroscedasticity present in the data. This allows for causal estimates to be made of the relation between self-rated health and long sleep. Controls are included for age, sex, race, education, and marital status. Long sleep is found to be causally associated with an approximate 6-10% decrease in self-rated health status. Long sleep is a significant public health issue detrimentally affecting 22.9 million adults in the US. The Lewbel instrumental variables estimator should be exploited as one approach to determine the specific causes of long-sleep patterns.

Keywords: long sleep duration, self-rated health, Lewbel instrumental variables