Relevance & Research Question

The growing influence of social media has sparked increasing concerns about its impact on digital well-being, particularly among young people. A survey from 2023 revealed that adults are highly concerned about the effects of social media on their children's mental health (Knight Foundation, 2023). Our literature review highlighted key areas of existing research, such as the relationship between parents' social media sharing and children's privacy (Ong et al., 2022), the adverse effects of digital media on toddlers (e.g., Barr, 2022), emotional developmental delays, anxiety, and depression among children (O’Riley et al., 2018; Primack et al., 2017), and the short-term benefits of social media breaks (e.g.,  Brown & Kuss, 2020). Building on these findings, the current research aims to (1) explore adolescents’ (10-14 years) social media behavior and (2) identify different social-media-user segments based on their social media behavior and sociodemographic variables.

Methods & Data
To answer the research question, a quantitative online survey was conducted. A convenience sample of 297 (Austrian) respondents (parents of at least one child aged 10-14) completed the questionnaire for small renumeration. To identify social-media-user segments, we collected information about adolescents’ consumption behavior of social media (i.e. use of different platforms, duration of use, social-media-activities, active vs. passive social-media-usage, different social-media-rules set by parents) evaluated by their parents. Finally, we collected sociodemographic data (age, gender, type of school of adolescents; age, gender, highest education of parents). We then conducted a two-step cluster analysis (applicable for nominal and metric data) to derive segments based on social media behavior of adolescents.

Results
We are currently refining and deepening the results and implications of the cluster analysis which will – hopefully – be discussed at GOR 2025.

Added Value

Exploring adolescents’ social-media-behavior is critical from both a scientific and a practical perspective by providing insights into how different segments of adolescents engage with social media, enabling tailored interventions for digital well-being, education, and digital literacy. Additionally, it supports policymakers in creating ethical, segment-specific strategies for communication, product design (e.g. apps), and online safety regulations.

Relevance & Research Question

Smart surveys combine surveys with smart elements from sensors, for example, the use of the smartphone camera for receipt scanning in a household budget survey and the use of geolocation tracking to identify activities in time use surveys. At this point, relatively little is known on how to best implement smart surveys in the general population for official statistics, and what influence the different features of smart surveys have on participation behavior.

Methods & Data

In 2024, fieldwork experiments were conducted in Norway, Belgium, and Germany to test various options in how to design and field smart surveys as part of national household budget surveys and time use surveys. The experiments in the three countries varied several design features to test their effect on recruitment and participation rates to smart surveys. In Norway, the use of different platforms from which the data collection app could be accessed and the use of CATI interviewers for recruitment and follow-up was tested. In Belgium and Germany different recruitment protocols were tested, including the use of different appeals in the invitation letters focusing on features of the smart survey (e.g., use of the camera to scan receipts) and secondary data collection modes (e.g., paper questionnaires instead of app).

Results

We find that recruitment and participation rates vary across countries, and that the differences between within-country experimental conditions are relatively small. The poster will present results on differential recruitment and participation rates and nonparticipation bias in the three countries.

Added Value

This research is part of the Smart Survey Implementation (SSI) project, funded by EUROSTAT, which aims to enhance data collection for official statistics across Europe through digital innovation. This experiment specifically addresses recruitment challenges in app-based surveys and evaluates the potential of mobile technology to streamline participation in official household budget and time use surveys.

Relevance & Research Question

Digital skills have become important for navigating in today’s information society. While prior digital inequality research has mostly focused on studying general internet uses and skills, research on smartphone-specific inequalities is still scarce. In addition, existing measurement instruments mostly rely on survey-based self-reports or small-scale laboratory-based performance tests that are susceptible to measurement and representation errors. In this study, we examine the feasibility of using novel scenario-based measures to evaluate the level of smartphone skills in the general population. Scenario-based measures evaluate smartphone skills by assessing how well respondents perform a set of smartphone activities described in a hypothetical situation.

Methods & Data

Data were collected in the German Internet Panel, a probability-based online panel of the general population aged 16-75 in Germany, in March 2022. Respondents were asked to answer three scenario-based questions and rate their general smartphone skills. The scenario-based questions asked respondents to correctly order a set of steps to carry out smartphone activities, such as buying a train ticket with an app that is not yet installed on their device. We examine response distributions and correlations between the scenario-based and self-reported measures. We also assess whether predictors of smartphone skills differ between the two measures.

Results

The scenario-based and self-reported measures are significantly positively correlated and measure the same underlying construct as determined by an exploratory factor analysis. Compared to self-reports, the scenario-based measures, however, have substantially greater rates of item-nonresponse. Older and less educated smartphone owners are significantly less likely to respond to the scenario-based questions. The predictors of smartphone skills differ by respondents’ sociodemographic characteristics across the two measures. Older, female, and more educated respondents are more likely to underreport their smartphone skills in the self-report compared to the scenario-based questions.

Added Value

Methodologically, we demonstrate the feasibility of using scenario-based measures of smartphone skills in an online survey. Substantively, we contribute to the growing body of research on the second-level smartphone divide.

RELEVANCE & RESEARCH QUESTION

Data donation is a relatively new, user-centered approach to
collecting digital trace data, increasingly relevant in various research fields (Haim et al., 2023). It
offers significant potential to understand online behavior, especially amid growing API restrictions, as
it can help validate and enrich survey data. However, little is known about who participates in these
studies (Keusch et al., 2024). Previous results hint that participation is non-random (Welbers et al.,
2024). Our study aims to identify correlates of participation, moving beyond sociodemographics to
consider political preferences and vote choice.

METHODS & DATA:

Wave 60 participants from the GLES Tracking pre-election survey (n ≈ 2,000, CS,
CAWI, online access panel) are invited after the 2025 German Federal Election to complete a follow-
up survey on their vote choice, followed by a request to donate their Google search histories from six
weeks before to one week after the election. Besides search terms, we collect URLs of clicked search
results and the order of searches and clicks. We experimentally vary the invitation framing to
evaluate how different descriptions of the collected data (“web services”, “data collected by Google”,
“data collected by Google Search”) influence participation rates. The digital trace data will be linked
to both survey datasets to analyze participation across sociodemographics, political attitudes,
electoral preferences, and vote choice. Our design captures multiple participation stages, from initial
opt-in to final data donation.

RESULTS:

Data is collected one week after the 2025 German Federal Election (February 23, 2025). At
the GOR conference, we will present results on participation rates by invitation framing,
sociodemographics, political preferences, and vote choice, differentiating between respondents who
opted in, consented, processed, and ultimately donated their data.

ADDED VALUE:

This study provides insights into how sociodemographic characteristics relate to
participation in data donation studies. Additionally, we examine political preferences and vote choice,
as well as how the specificity of the data requested in the invitation affects willingness to participate.
Since Google search data is highly private, our findings will inform optimal framing for requests
involving sensitive data. Our study advances methodological approaches in digital behavioral data
research and provides practical guidelines for designing effective data donation studies.