Searching for Sleep: What Digital Trace Data Reveals About Infant Sleep Difficulty

Abstract

When do babies sleep worst? Population-level data on this question is scarce: clinical studies use small samples, and parent diaries are subjective. We use Google Trends data for age- specific sleep search terms (“K month old sleep,” K = 1–24) at weekly resolution across the US (2024–2026) as a revealed-preference measure of parental sleep concern. Search volume follows exponential decay with age, consistent with declining concern as infants mature. Relative to this decay baseline, leave-one-out analysis identifies a single age with significantly elevated search activity: 18 months (+422%, z = 4.14). All other ages—including those commonly associated with “sleep regressions” (4, 8, 12 months)—fall within normal variation of the decay curve. We validate the search signal using spring-forward Daylight Saving Time as a natural experiment: DST-observing states show +9.7% more “baby sleep” searches in the week after the clock change, while Arizona (which does not observe DST) shows−6.7%—a 16 percentage-point gap.

Comments