Would you ever hand over your private messages for the sake of scientific research? These days, so much of our social lives happen through texts and social media messages. These interactions are all rich sources of data that social psychologists and other researchers would love to study to better understand real-world social behavior.

As a psychology student, I’ve participated in plenty of research studies, but if one of them asked me to upload my entire messaging history?! That’s not something I’d be very willing to hand over. No one needs to see the 27-text rant I recently sent to my best friend except her!

Using donated digital data is becoming increasingly popular in research, but as you can imagine, many people, like me, have privacy concerns about sharing something as personal as their messaging history. How can we trust that researchers are handling such sensitive data safely?

To address this privacy concern, a research team including Olya Hakobyan, Paul-Julius Hillmann, Florian Martin, Erwin Böttinger, and Hanna Drimalla (pictured below) developed Dona – a privacy prioritizing data donation platform where people can donate messaging data from WhatsApp, Facebook, and Instagram for the purpose of scientific research.

What makes Dona unique is that it never collects the actual content of your messages. Instead, it only gathers metadata like message length and timestamps. The content itself never leaves your device, so there’s no chance of anyone, including the researchers, reading what you actually sent. In fact, even the metadata undergoes de-identification on your browser before it’s donated, so your data won’t ever be associated with your name.

So, if I were to donate my messages, the researchers wouldn’t see my 27-text rant. They’d only know that a participant sent 27 messages of various lengths to someone else in 5 minutes.

Can such minimized data still be valuable? In their recent paper published in Behavioral Research Methods, the developers of Dona explain how their novel approach works to protect participant privacy while still allowing for meaningful insights for both the participants donating their data and the researchers using it.

How Dona works

There are three main steps to using Dona (see details in the figure below):

• Data request – Participants download their chat history files directly from WhatsApp, Facebook, and/or Instagram onto their own devices.
• Data de-identification – All usernames are replaced with unique IDs, and the timestamp and word count of each message are extracted.
• Visual feedback – Participants get to visualize their messaging data, including intensity (based on word count), active hours, and response times.

Was the visual feedback useful to participants?

Yes! The researchers designed and ran a study to understand how the visual feedback could benefit the participants. Participants rated the visual feedback as highly informative, and there was moderate agreement that it improved their understanding of their messaging behavior Participants were less convinced that the feedback would change their messaging habits, such as how often they send messages (Panel A in the figure below).

Most participants were also able to recognize changes in their messaging behavior over time (panel C in the figure below). Notably, most of those participants were even able to link those changes to real life events (see the dark blue bar in panel C in the figure below).

Because participants could connect their messaging patterns to personal experiences, the authors point to potential implications of utilizing the visualization of messaging habits for mental health. As they explain:

“Imagine going through a tough time, not immediately noticing that your messages to close friends have become less frequent. Seeing that pattern laid out can bring clarity over how life events are shaping our interactions and how we can act upon it”.

Does the minimized metadata still reflect known aspects of social interactions?

Yes! In the same study, the researchers used several measures of features of human communication (interaction balance, heterogeneity, and burstiness) to summarize the messaging data and check whether it captures patterns consistent with prior research.

Interaction balance looks at how evenly conversation partners contribute to a chat. In this study, it was measured by comparing the number of words each conversation partner sent. Most participants had balanced conversations, sending roughly equal numbers of words. But some chats were skewed, with one person sending or receiving more messages than their conversation partner (panel A in the figure below).

Interaction heterogeneity captures how attention is spread across different contacts (i.e., some contacts may get most of our messages, while others may only hear from us occasionally). This was measured using the Gini index, a measure where higher values indicate a more unequal distribution of communication across contacts (i.e., only a few people get most of your messages). The Gini index median was around 0.44 (on a scale from 0 to 1), showing that people’s interactions weren’t equally spread across all their contacts, but also weren’t just focused entirely on just one person (panel B in the figure below).

Burstiness captures the timing of our conversations. An interaction is considered “bursty” when it involves periods of rapid back-and-forth messaging, followed by periods of silence. Measured on a scale from -1 (very regular) to 1 (very bursty), most chats fell in the positive range, showing conversations were somewhat “bursty” (panel C in the figure below).

These findings highlight how even simple data, like how often we send messages or how much we write, can still uncover meaningful patterns in social interactions. Dona is an innovative platform that, especially when combined with questionnaire data, offers exciting opportunities to study how communication habits relate to important psychological factors.

Personally, a platform like this makes me feel much more comfortable with the idea of donating my own data. I’m also pretty curious to see what my visual feedback would reveal! If you’re interested in donating your own data to Dona, you can try it out here.

Featured Psychonomic Society article

Hakobyan, O., Hillmann, P. J., Martin, F., Böttinger, E., & Drimalla, H. (2025). Development and evaluation of Dona, a privacy-preserving donation platform for messaging data from WhatsApp, Facebook, and Instagram. Behavior Research Methods, 57(3), 1-17. https://doi.org/10.3758/s13428-024-02593-z