How Much Time Do You Spend on Your Smartphone? Examining the Inaccuracies of Self-Reported Estimates of Smartphone Use and the Implications for Detecting Effects

Background: The increasing ubiquity of mobile device and social media (SM) use has generated a substantial amount of research examining how these phenomena are associated with various aspects of well-being and psychopathology. However, many of these studies relied upon self-reported estimates to measure amount of use, which can be inaccurate when compared against more objective measures of use. Errors in self-reported estimates can have significant consequences when attempting to detect effects on well-being and may lead to type I and/or type II errors.

To date, there has been limited research into the psychosocial characteristics—such as depression, loneliness, and life satisfaction--that may predict inaccuracies in estimates of mobile device and SM use.

Thus, utilizing Apple’s recently released “Screen Time” application to obtain actual iPhone and SM use data, which automatically tracks usage on iPhones—this study aimed to: 1.) examine the accuracy of self-reported estimates of use; 2.) examine how inaccuracies in estimates of use bias correlations with psychosocial variables; 3.) examine how usage amount and psychosocial factors predict the amount of inaccuracy in estimates of use.

Methods: Utilizing a cross-sectional online survey design, we recruited 393 iPhone users from Amazon Mechanical Turk from 1/3/2019 to 2/2/2019. The sample was mostly white (80%) and male (58%) with a mean age of 33. Participants first provided estimates of their iPhone and SM use over the past week, then transferred their actual usage information from the Screen Time application, and also completed validated measures of depression, loneliness, and life satisfaction.

We used paired-sample t tests to examine mean differences between estimated and actual use and Pearson correlations to examine relationships among the use and well-being variables. We took the absolute difference between estimated and actual use measures to create “inaccuracy scores” for weekly overall iPhone use and weekly SM use. We used multiple regression to examine whether usage amount and psychosocial factors predict amount of inaccuracy.

Results: Overall, self-reported estimates of use were highly inaccurate: participant estimates of weekly iPhone use and weekly SM use were off by an average of 22.1 and 16.6 hours, respectively. Participants significantly over-estimated their weekly and daily SM use. Additionally, we found that the error in self-reported estimates of use led to inflated correlations with the well-being variables when compared against the actual use variables. Results of multiple regression analyses found that depression and amount of actual use were the strongest predictors of inaccuracies in estimated weekly overall iPhone use and weekly SM use.

Implications: Findings suggest that relying upon participant estimates to measure amount of iPhone and SM use may not be reliable and may lead to biased correlations with important psychosocial characteristics, potentially leading researchers to conclude that a stronger relationship exists than is warranted. Furthermore, participants who are more depressed and/or have greater amounts of use may be providing the least reliable estimates of use, suggesting that estimates of use may be prone to systematic error that is fundamental to the relationship being investigated.