OSD2F: An Open-Source Data Donation Framework

Abstract

The digital traces that people leave through their use of various online platforms provide tremendous opportunities for studying human behavior. However, the collection of these data is hampered by legal, ethical and technical challenges. We present a framework and tool for collecting these data through a data donation platform where consenting participants can securely submit their digital traces. This approach leverages recent developments in data rights that have given people more control over their own data, such as legislation that now mandates companies to make digital trace data available on request in a machine-readable format. By transparently requesting access to specific parts of this data for clearly communicated academic purposes, the data ownership and privacy of participants is respected and researchers are less dependent on commercial organizations that store this data in proprietary archives. In this paper we outline the general design principles, the current state of the tool, and future development goals.

Linking Surveys and Digital Trace Data: Insights From two Studies on Determinants of Data Sharing Behaviour

“…As a consequence, attitudes and values only affect sharing behaviour to a small degree. Hence, factors that directly relate to the data sharing difficulty, such as the data sharing method (Araujo et al, 2021; Boeschoten et al, 2020) should be tested and optimised for research that seeks to combine surveys and digital trace data. Providing survey respondents with additional incentives (Keusch et al, 2019) also seems beneficial for increasing the willingness to share data from digital platforms.…”

“…Another option is to ask participants to export (parts of) their digital trace data themselves, which is a functionality that most platforms and services offer, and then share these ‘data download packages’ (Boeschoten et al, 2020) with the researchers, for example, by uploading them through a web tool. This relatively new approach is often called ‘data donation’ due to the active role of the respondents within the data sharing process (see, e.g., Araujo et al, 2021). Notably, those different data sharing procedures vary with regard to task difficulty and respondent burden, which translates into perceived costs.…”

“…Notably, those different data sharing procedures vary with regard to task difficulty and respondent burden, which translates into perceived costs. Specifically, more active data sharing procedures, such as data donation, are usually more burdensome for respondents than passive ones, such as providing consent and a username (Araujo et al, 2021; Keusch et al, 2019). At the same time, more active data sharing procedures provide more control, which can increase the willingness to share data (Juga et al, 2021).…”

Privacy-preserving local analysis of digital trace data: A proof-of-concept

“…Alternatively, Araujo et al. 6 developed OSD2F, which does not allow the local processing to completely take place before uploading and donating but does allow participants to inspect their data in order to let them decide which parts to share. The software introduced in this paper is unique in the sense that it allows the researcher to choose which DDP is used, which data formats are processed, which types of features are extracted, and how the data are summarized.…”

Digital data donations: A quest for best practices

“…As research making use of data donation begins to gain traction in the social sciences with different initiatives and solutions covering a diversity of use cases, 5 it becomes increasingly important for the field to establish best practices. These practices should ensure, on the one hand, that projects are designed and executed in a responsible, transparent, and privacy-protecting manner, thus being respectful of individuals participating in such research.…”