Passively sensing smartphone use in teens with rates of use by sex and across operating systems

Jordan D. Alexander, Janosch Linkersdörfer, Katherine Toda-Thorne, Ryan M. Sullivan, Kevin Cummins, Rachel L. Tomko, Nicholas B. Allen, Kara Bagot, Fiona C. Baker, Bernard F. Fuemmeler, Elizabeth A. Hoffman, Orsolya Kiss, Michael J. Mason, Tam T. Nguyen‐Louie, Susan F. Tapert, Calen J. Smith, Lindsay M. Squeglia, Natasha E. WadeScientific 2024 · DOI

TL;DR

This study introduces and validates a passive sensing method for tracking adolescent smartphone use, revealing that teens significantly underestimate their screen time (by about 2 hours daily). The findings highlight differences in usage patterns by sex and operating system, with social media and texting dominating usage.

What Problem Does This Paper Address?

Adolescent screen media activity (SMA) is a growing public health concern, but research has been hampered by reliance on self-reported data. Self-reports are prone to recall bias and effortful for participants, leading to infrequent and potentially inaccurate data. This is particularly problematic for understanding smartphone use, which is rapidly evolving and deeply integrated into adolescent life. Objective, continuous measurement is needed to accurately capture these behaviors and their potential impact on development.

Methods

The researchers utilized the Effortless Assessment Research System (EARS) application, a specialized tool within the Adolescent Brain Cognitive Development (ABCD) Study. EARS passively collected data on smartphone usage, including:

  • Keyboard Use: Recording when the keyboard was activated within an app, serving as a proxy for app engagement.
  • App Use (Android only): Directly tracking which applications were in the foreground on Android devices.
  • Keystrokes: Quantifying the number of keystrokes made.

A key methodological contribution is the development of a harmonization method to make keyboard and app usage data comparable across different operating systems (iOS and Android). This was crucial because iOS devices historically offered less access to sensor data for third-party apps. The study also involved comparing these passively collected metrics with self-reported smartphone usage data.

Datasets

The primary dataset used is from the Adolescent Brain Cognitive Development (ABCD) Study, a large, longitudinal, U.S.-based cohort. Specifically, the study analyzed cross-sectional data from the Year 4 follow-up, including 1,463 participants (aged 14 on average) who had usable ABCD-EARS data.

Key Insights and Ideas

  • Significant Underreporting: Adolescents consistently underestimate their smartphone usage. The passive sensing data revealed an average of 5 hours of daily smartphone use, which is approximately 2 hours more than what participants self-reported. This suggests a potential lack of self-awareness regarding actual screen time.
  • Gender Differences Emerge: Females reported and were passively observed to use their smartphones significantly more than males across all measures (app use, keystrokes, and self-report).
  • Operating System Nuances: While the study aimed for cross-OS comparability, there were differences. Android users showed more daily keystrokes than iOS users on average. Missing data was also more prevalent for iOS keystroke data, potentially due to the use of a third-party keyboard.
  • Dominance of Social and Communication Apps: Social media and texting applications accounted for nearly half of all smartphone use, highlighting their central role in adolescent communication and social life.
  • Late-Night Usage is Common: A substantial portion of participants remained active on their smartphones late into the night, with 80% of Android users recording app use between midnight and 5 AM on weeknights. This raises questions about the impact on sleep and overall well-being.
  • Passive Sensing as a Valid Tool: The study validates passive sensing (keyboard and app use) as a reliable and relatively accurate method for measuring adolescent smartphone use, offering a valuable alternative to self-report.

Key Results

  • Reliability: Keyboard and app-use measures were found to be reliable and positively correlated with each other (r = 0.33).
  • Convergent Validity: Passive sensing measures showed moderate positive correlations with self-reported use (rs = 0.21-0.35).
  • Daily Usage: Participants averaged 5 hours of daily smartphone use via passive sensing, compared to approximately 3 hours via self-report.
  • Gender Differences: Female participants showed significantly higher average daily app use, keystrokes, and self-reported smartphone use compared to males.
  • App Category Usage: Social apps were the most used category across both passive sensing measures and self-report.
  • Late-Night Activity: 80% of Android users had app usage between midnight and 5 AM on weeknights.

Limitations and Future Work

  • Sampling Bias: The study acknowledges potential sampling biases, as participants who opted into passive sensing may differ demographically from the broader ABCD cohort. Privacy concerns were a significant barrier to participation.
  • Missing Data: Substantial missing data was observed, particularly for iOS keystroke data. This could be due to technical issues, participants disabling the app, or the cumbersome nature of the third-party keyboard. Missingness itself was also found to predict lower reported and passively measured usage.
  • Hawthorne Effect: Participants might have altered their behavior due to the awareness of being monitored, potentially affecting the external validity of the findings.
  • Cross-OS Comparability: While efforts were made to harmonize app categories, differences in app store categorization and third-party app data collection limitations (especially for iOS) mean that direct comparisons between operating systems require careful consideration.
  • Future Work: The authors emphasize the potential of combining this rich passive sensing data with the broader ABCD dataset for longitudinal studies on health and neurodevelopment. They suggest future analyses could explore specific apps and their psychiatric correlates, consider intersectionality characteristics (e.g., LGBTQ+ youth), and investigate the impact of late-night usage on sleep and mental health. The development of more robust methods for controlling for missing data and improving cross-OS comparability is also highlighted.

Summary updated

Abstract

Youth screen media activity is a growing concern, though few studies include objective usage data. Through the longitudinal, U.S.-based Adolescent Brain Cognitive Development (ABCD) Study, youth (m<sub>age</sub> = 14; n = 1415) self-reported their typical smartphone use and passively recorded three weeks of smartphone use via the ABCD-specific Effortless Assessment Research System (EARS) application. Here we describe and validate passively-sensed smartphone keyboard and app use measures, provide code to harmonize measures across operating systems, and describe trends in adolescent smartphone use. Keyboard and app-use measures were reliable and positively correlated with one another (r = 0.33) and with self-reported use (rs = 0.21-0.35). Participants recorded a mean of 5 h of daily smartphone use, which is two more hours than they self-reported. Further, females logged more smartphone use than males. Smartphone use was recorded at all hours, peaking on average from 8 to 10 PM and lowest from 3 to 5 AM. Social media and texting apps comprised nearly half of all use. Data are openly available to approved investigators ( https://nda.nih.gov/abcd/ ). Information herein can inform use of the ABCD dataset to longitudinally study health and neurodevelopmental correlates of adolescent smartphone use.
BibTeX
@inproceedings{alexander2024passively,
  author = {Jordan D. Alexander and Janosch Linkersdörfer and Katherine Toda-Thorne and Ryan M. Sullivan and Kevin Cummins and Rachel L. Tomko and Nicholas B. Allen and Kara Bagot and Fiona C. Baker and Bernard F. Fuemmeler and Elizabeth A. Hoffman and Orsolya Kiss and Michael J. Mason and Tam T. Nguyen‐Louie and Susan F. Tapert and Calen J. Smith and Lindsay M. Squeglia and Natasha E. Wade},
  title = {{Passively sensing smartphone use in teens with rates of use by sex and across operating systems}},
  booktitle = {Scientific},
  year = {2024},
  doi = {10.1038/s41598-024-68467-8},
}
Figure