Keywords: data donation, self-reports, digital behavioral data, smartphone use

Digitalization has produced a vast range of new data types. Researchers are actively discussing the prospects of using various trackers - pedometers, fitness bracelets, and digital activity trackers - as new sources of information about human behavior. For example, data from smartphones can be used as a measure of media consumption or social media usage. However, obtaining this kind of data requires developing new theoretical approaches, methodological standards, and solving a number of technical, legal, and ethical problems. One possible way to obtain such data is through the method of “data donation”, which allows researchers to collect digital behavioral data directly from respondents without installing special tracking applications. In this approach, respondents share with researchers the data their mobile devices have collected (for example, screen time data).
Thus, this study aims to evaluate the concurrent validity of self-reported data on smartphone usage using digital behavioral data obtained from respondents via the “data donation” approach. We conducted an experiment on cross-validating self-report data on smartphone use and digital behavioral data among iOS and Android users (n=200, non-probability sample). Respondents were asked to estimate: 1) average phone usage time; 2) average usage time for the current day; 3) average time over the past week; and 4) typical week. Respondents also ranked applications based on the volume of use over the last week. Data from screen time sections of phones were attached to the questionnaires. Survey data and mobile tracker data were linked at an individual level.
Hypotheses tested in the research:
H1: The accuracy of estimating smartphone usage time depends on the length of the time interval. The shorter the period, the more accurate the assessment.
H2: Smartphone use is routine and usually typical, so the average score for a day, for a current day, and for a past day does not significantly differ.
H3: The accuracy in estimating the use of apps depends on the volume of their use. The more an app is used, the more accurately it can be estimated.
Currently, statistical analysis is in progress.
The results of the previous experiment conducted by the authors showed a moderate positive correlation between retrospective self-report data and digital behavioral data on mobile usage. The results of the analysis of variance with repeated measures (Repeated Measures ANOVA) showed that there is no statistically significant difference in the accuracy of estimates of smartphone usage for a short (one day) and for a longer period (the current week). That is, the accuracy of self-reported data does not vary significantly within the current week. Therefore, the level of concurrent validity of self-reporting on smartphone usage can be described as moderate. However, digital behavioral data may be considered more preferable for assessing smartphone usage, due to the ease of data collection, the absence of significant additional costs, and presumably, a shorter response time and reduced burden on respondents.