Determinants of willingness to donate data from social media platforms

ABSTRACT

Social media data donation through data download packages (DDPs) is a promising new way of collecting individual-level digital trace data with informed consent. Nevertheless, given the novelty of this approach, little is known about whether and how people would share their data with researchers, although this could seriously affect selection bias and thus, the outer validity of the results. To study the determinants of data-sharing and help future data donation studies with detecting the conditions, under which the willingness is the highest, we pre-registered two vignette experiments and embedded them in two online surveys conducted in Hungary and the US. In hypothetical requests for donating social media data via DDPs, we manipulated the amount of the monetary incentives (1), the presence or lack of non-monetary incentives (2), the number of requested platforms (3), the estimated upload/download time (4), and the type of requested data (5). The results revealed that data-sharing attitude is strongly subject to the parameters of the actual study, how the request is framed, and some respondent characteristics. Monetary incentives increased willingness to participate in both countries, while other effects were not consistent between the two countries.

KEYWORDS:

• Data donation
• data download package (DDP)
• data-sharing
• digital trace data
• linking survey data

Acknowledgements

Author contributions: Conceptualization Z.K., Á.S., J.S., E.O., J.K.; theoretical background: Á.S., D.D., A.K., E.P.; methodology Z.K., Á.S., J.S. and J.K..; formal analysis, Z.K., Á.S., J.S.; data curation, Z.K. and J.S..; writing – original draft preparation Z.K., Á.S., J.S., D.D., A.K., E.P., E.O., J.K..; writing – review and editing, Z.K., Á.S., J.S., D.D., A.K., E.P., E.O., J.K.; visualization, J.S..; funding acquisition, Z.K. and S.A.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Data is available in the repository of CSS-KDK: https://openarchive.tk.mta.hu/584/. doi: 10.17203/KDK584.

Ethics approval

Ethical review and approval were waived for the Hungarian study because data was anonymized within the fieldwork of the study. The results do not allow identification of the individuals involved in the Hungarian study. The authors and the fieldwork agency managed all information collected in accordance with the General Data Protection Regulation (GDPR). The US study has been approved by the Harvard Institutional Review Board (IRB22-0942).

Notes

1 The other main reason was privacy concern.

2 Countries were chosen on a practical basis and opportunity to collect data. Nevertheless, we believe that collecting data from different countries allows us to test the robustness of our findings and assess whether cultural factors influence attitudes and behaviors towards data donation. There are several reasons why comparing these two countries is relevant. For instance, compared to Hungary, the US is a highly developed economy with a strong focus on technological innovation, which may impact how citizens react to digital requests. Internet and social media usage (or platforms) and digital literacy are also different in the two countries. These factors can influence people's understanding of data donation and their readiness to participate in such initiatives. Trust in science, trust in institutions or interpersonal trust may all affect willingness to donate data. The levels of trust also show some, although not large differences between the two countries. Lastly, data privacy regulations are markedly different in Europe compared to the US which can lead to differing levels of public awareness and trust in how data is handled and protected. At the same time, identifying cultural differences was not the main focus of this study, thus we did not develop any comparative hypotheses. Conducting data collection across two countries enhances the robustness of our study by introducing a more diverse sample, thereby increasing the generalizability and reliability of our findings.

3 The pre-registration for the Hungarian study is available here: https://osf.io/r4kxm; and for the US study is here: https://osf.io/tvefj

4 For a robustness check of the results, we calculated the standard deviation of the willingness probability for the first and second six vignettes. A smaller standard deviation might have been a sign of fatigue for the respondent. Based on Barlett's test, we did not find differences between the standard deviations (p = .39) of the two sets.

5 For robustness check, we re-ran this multilevel model with the first six and second six vignettes separately (see Table A2 in the supplementary). There were some differences between the evaluations of the first and second six vignettes. Still, the incentive had the most pronounced positive effect in the regressions fitted to both vignette groups. For the second six vignettes, fewer variables have a significant impact which may indicate fatigue and less attentive evaluation.

6 In the US sample, 7% of respondents said they belonged to Other gender category or did not answer the gender question. We imputed data of these respondents in the main models, but we also run an alternative model where the Other and DK/NA categories are combined and included as a third category in the analysis. The results of these modelling runs are presented in Table A5. The three-category variable performed very similarly to the two-category variable, with female having higher willingness to participate.

7 Our models were also tested by including the variable measuring the incentive as a factor in the model rather than continuously. The explanatory power of the models was not higher in the alternative runs, and the B values of the incentive variable indicated that the effect of the variable of interest was linear.

Additional information

Funding

The research was funded by the Eötvös Loránd Research Network within the framework of Flagship Research Projects: KÖ-32/2021. The work of Julia Koltai was funded by the Lendület ‘Momentum’ grant of the Hungarian Academy of Sciences. The work of Zoltán Kmetty was supported by the Bolyai Scholarship, grant number: BO/834/22.

Notes on contributors

Zoltán Kmetty

Zoltán Kmetty, PhD, is a senior research fellow at the HUN-REN Centre for Social Sciences, CSS-RECENS research group; and an associate professor at the Eötvös Loránd University Faculty of Social Sciences, Sociology department. He has diverse research interests, including political sociology, network studies, and suicide research. He is an expert in methodology, survey design, and quantitative analysis [email: [email protected]].

Ádám Stefkovics

Adam Stefkovics, PhD in sociology, is a research fellow at the Centre for Social Sciences in Budapest and a visiting researcher at the Institute for Quantitative Social Sciences at Harvard University. His research interest includes political sociology and survey methodology.

Júlia Számely

Júlia Számely, PhD candidate at the Department of Network and Data Science at the Central European University. Her thesis explores the personal characteristics associated with misinformation vulnerability in Hungary and their effects on misinformation spreading dynamics, using a donation based dataset relating digital footprint to survey data on the individual level. She was trained in economics, sociology, and social data science.

Dongning Deng

Dongning Deng, PhD candidate at the Doctoral School of Sociology at Eötvös Loránd University, Budapest. Currently, her doctoral research focuses on computational sociology and studies the social activity patterns of diverse social groups in digital platforms. Her specific interest is in the social network and temporal patterns and their related social implications. She also has a background in public health.

Anikó Kellner

Anikó Kellner, Completing her MA studies in Sociology at Eötvös Loránd University, Budapest. Her research interest includes sociology of religion and research methodology. She holds a PhD in History from Central European University.

Edit Pauló

Edit Pauló, PhD candidate at the Doctoral School of Sociology at Eötvös Loránd University, Budapest. Her doctoral research focuses on the digital experiences of older adults. She was trained in economics and sociology.

Elisa Omodei

Elisa Omodei, PhD in Applied Mathematics for the Social Sciences, is an Assistant Professor at the Department of Network and Data Science at the Central European University. She is an expert in network science and data science. Her research interests span from socioeconomic vulnerability to misinformation diffusion and political polarization.

Júlia Koltai

Júlia Koltai, PhD, is a senior research fellow at the HUN-REN Centre for Social Sciences and an associate professor at Eötvös Loránd University, Faculty of Social Sciences. She is a sociologist and statistician by training and has worked on computational social science related problems in recent years. She is the head of two research labs, ‘MTA–TK Lendület “Momentum” Digital Social Science Research Group for Social Stratification’ and ‘Social Science group of the National Laboratory for Health Security’.