The next session at the 5th Data Donation Symposium at the Weizenbaum-Institut in Berlin starts with Olya Hakobyan, whose focus is on the question of how to scale up data donation studies. The first challenge is how to motivate and recruit participants, which is centrally a question of trust in the researchers; the second is dropouts of participants due to technical difficulties; the next is managing the influx of data donations, especially if studies have been able to recruit a substantial number of participants; and finally studies also need to confront the challenge of varying data quality in donations.
Olya’s interest is in assessing social interactions and their association with psychological variables like loneliness and resilience; this connects self-reported and observed data, especially also around major life events, using the data donation platform Dona. This allows data donations from various social media platforms, but in practice most donations come from WhatsApp and Instagram.
Dona engages in extreme data minimisation by removing all content automatically; what remains are timestamps, word counts, media object counts, reaction types, and anonymised user IDs. Such reduced data can still be used very productively for studying social interactions, even if the content of those interactions is no longer known.
Accessing such data is difficult; there are different procedures for WhatsApp (where personal briefings are required to walk participants through the data request procedure) and Instagram (where data downloads are self-directed). Platform responses can take days to arrive, so further follow-ups after several days are required here. Some 41% of WhatsApp users ended up submitting donations in the end; this was also motivated by small payments to participants.
Data influx was managed via participation tokens that enabled such compensation payments and avoided the processing of duplicate donations. The data are also tested for volume and quality, as well as for statistical dependence (which could result from multiple participants submitting data that describe direct social media interactions between them). All of these measures help in the scaling up of these data donation efforts.
