Relevance & Research Question
Digital data donations in online panels have become an additional avenue for researchers to collect health and fitness tracking data, social media data, and web search and viewing histories from panel participants. Initial studies investigating the general willingness to donate data reported encouraging hypothetical consent rates, underscoring the potential of this method. Nevertheless, actual participation rates in donation studies generally remain low and systematically biased across demographic groups, raising concerns about representativeness and generalizability. In contrast, data donations within online panels have demonstrated consistent participation rates, indicating that online panels might offer favorable conditions for successful data donations. In our study, we explored the conditions and participant characteristics that influence data donation success in online panels compared to other data donation studies conducted within survey contexts.

Methods & Data
We conducted a systematic review of studies that implemented digital data donation protocols in the context of surveys using comprehensive searches across major academic databases (see preregistration: https://osf.io/nad2y). Studies meeting eligibility criteria were included in a quantitative meta-analysis assessing key indicators for donation success. For this exploratory part of the study presented here, we assessed whether indicators for donation success differed between donation studies in online panels and those in cross-sectional recruitments.

Results

In general, donation probabilities varied largely across studies and were strongly influenced by study design and contextual factors, such as donation method, the requested data type, and the recruitment strategy. Studies with a lower mean age generally showed higher donation probabilities. However, online panels displayed more consistent donation probabilities compared to cross-sectional recruitments. Potential moderating factors, such as participants’ familiarity with online data collections and trust in the institution receiving the donation, warrant further consideration.

Added Value
Our results offer meta-analytic insights into key indicators of data donation probabilities for studies conducted in online panels compared to cross-sectional recruitments.

Relevance & Research Question

Data donations are increasingly discussed as a valuable source for social science research. However, little probability-based evidence exists on what drives individuals’ hypothetical willingness to donate personal communication data. We examine three components that could influence consent behavior: motivational framing (societal benefit, personal benefit, no benefit information), privacy (consent from chat partners required vs. consent not required), and the requested data type (aggregated metadata vs. full chat content).

Methods & Data

The study was fielded in the May 2025 wave of the German Internet Panel (GIP), a probability-based online panel of the German adult population. Respondents were randomly assigned to one of 18 conditions in a 3×2×2 experimental design. The dependent variable was a binary measure of hypothetical willingness to donate WhatsApp chats, and we controlled for characteristics such as WhatsApp usage intensity and perceived data sensitivity.
Results

Overall, 11% of respondents (approx. 350 individuals) reported willingness to donate their chat data. Willingness was higher when societal benefit was emphasized (13%) compared with personal benefit or no benefit information (each 10%). Requiring consent from chat partners showed no significant effect (12% vs. 10%). Participants were more willing to donate when full chat content was requested (15%) rather than aggregated metadata (8%). Multivariate analyses indicated that willingness increased among heavy WhatsApp users and decreased with higher perceived data sensitivity. No significant interaction effects across experimental factors were found.

Added Value

This study provides one of the first probability-based examinations of willingness to donate WhatsApp chat data—an especially sensitive and understudied data type. The results indicate that, in this hypothetical context, variation in content sensitivity influenced stated willingness less than expected..These findings offer empirically grounded guidance for implementing future data donation infrastructures and highlight which informational cues and design choices may reduce barriers, increase trust, and support responsible integration of chat-based digital trace data into social research. For example, trying to limit the content sensitivity of a data donation request may not be as promising to increase data donation rates as simply emphasizing the research’s societal benefit.

Relevance & Research Question
Data protection regulations in the EU, Brasil, and California give users the right to access the information online platforms hold about them. Data donation studies capitalize on this requirement by asking web survey respondents to request and donate their data at the end of the survey. However, these studies often suffer from modest donation rates. In this study, we experimentally test whether using different motivational appeals in the data donation request and persuading respondents who initially decline the request increase participation.

Methods & Data
In summer of 2025, we conducted an experiment involving over 2,000 participants from a German online access panel. Panel members were asked at the end of a web survey to donate data from LinkedIn, Instagram, or YouTube. Here, we randomly varied the motivation appeal, either (1) emphasizing respondents' ability to quantify their online platform behavior (personal appeal), (2) emphasizing respondents' ability to learn what online platforms know about them (prosocial appeal), and (3) no appeal. Those who declined were asked for their reason for not donating, followed by a persuasion message that either provided (1) an argument tailored to their reason for declining or (2) a random argument not tailored to their reason for declining.

Results
We found that the motivational appeals did not significantly affect respondents' willingness to donate data or their actual donation rates. However, the inclusion of motivational appeals had some effects on donation bias, for example, the prosocial appeal led to an overrepresentation of respondents with high trust in science among the donors. While providing a persuasion message led about 7% (60 out of 876) of decliners to reconsider their initial decision, only 3 additional respondents actually donated their data after receiving the persuasion message. The type of persuasion argument (tailored vs. random) did not affect participation.

Added Value
Our study shows that basic motivational appeals and persuasion messages have little effect on the participation behavior in data donation studies, but can influence donation bias. As our persuasion message was rather static, future studies investigating such adaptive survey designs could adopt a more tailored approach based on conversational agents.