Repository logo

Teenage sleep and technology engagement across the week.

Accepted version

No Thumbnail Available



Change log


Orben, Amy 
Przybylski, Andrew K  ORCID logo


BACKGROUND: Throughout the developed world, adolescents are growing up with increased access to and engagement with a range of screen-based technologies, allowing them to encounter ideas and people on a global scale from the intimacy of their bedroom. The concerns about digital technologies negatively influencing sleep are therefore especially noteworthy, as sleep has been proven to greatly affect both cognitive and emotional well-being. The associations between digital engagement and adolescent sleep should therefore be carefully investigated in research adhering to the highest methodological standards. This understood, studies published to date have not often done so and have instead focused mainly on data derived from general retrospective self-report questionnaires. The value of this work has been called into question by recent research showing that retrospective questionnaires might fail to accurately measure these variables of interest. Novel and diverse approaches to measurement are therefore necessary for academic study to progress. METHODS: This study analyses data from 11,884 adolescents included in the UK Millennium Cohort Study to examine the association between digital engagement and adolescent sleep, comparing the relative effects of retrospective self-report vs. time-use diary measures of technology use. By doing so, it provides an empirical lens to understand the effects of digital engagement both throughout the day and before bedtime and adds nuance to a research area primarily relying on retrospective self-report. RESULTS: The study finds that there is a small negative association relating digital engagement to adolescent sleep both on weekdays and weekend days (median standardized association βweekday = -0.06 and βweekend = -0.03). There is a more negative association between digital engagement and total sleep time on weekdays compared to weekend days (median standardized βweekday = -0.08, median standardized βweekend = -0.02), while there is no such difference when examining adolescents' bedtime. Surprisingly, and contrary to our expectations, digital technology use before bedtime is not substantively associated with the amount of sleep and the tardiness of bedtime in adolescents. CONCLUSIONS: Results derived from the use of transparent Specification Curve Analysis methods show that the negative associations in evidence are mainly driven by retrospective technology use measures and measures of total time spent on digital devices during the day. The effects are overall very small: for example, an additional hour of digital screen time per day was only related to a 9 min decrease in total time spent sleeping on weekdays and a 3 min decrease on weekends. Using digital screens 30 min before bed led to a 1 min decrease in total time spent sleeping on weekdays and weekends. The study shows that more work should be done examining how to measure digital screen time before interventions are designed.



Adolescents, Digital technology use, Large-scale social data, Measurement, Screen time, Sleep, Specification curve analysis, Time-use diary

Journal Title


Conference Name

Journal ISSN


Volume Title





All rights reserved
Medical Research Council (MC_UU_00005/9)
Emmanuel College, Barnardo's UK, Huo Foundation