Restrictions on social media data harming research, says paper

Restrictions placed on researchers’ ability to access data by social media companies including X and TikTok is a “threat to open science”, according to a new paper.

Changes made in 2023 to application programming interfaces (APIs) – used by researchers to measure the behaviour and attitudes of social media users – have included restricting access or introducing charges.

Previously this data had been highly useful to academics in their research, a paper published in Nature Human Behaviour says.

“For example, these data can be used to examine where conflict is likely to occur, where to allocate aid in the event of natural disasters, how online polarisation or misinformation is impacting voting patterns,” the paper says.

It warns that changes to API access will make this type of research harder to conduct.

X (formerly Twitter) was criticised earlier this year for introducing charges for its API that many researchers said would force them to abandon their work on the platform.

In August 2023, it granted academic researchers permission to distribute an unlimited number of tweets “if they are doing so on behalf of an academic institution and for the sole purpose of non-commercial research”.

However, the authors argue that this is too restrictive because it still prevents access to the raw data. Furthermore, the terms of X state “one cannot infer anything on an individual level regarding, for example, health, political stance or demographics”, the paper notes.

“Currently, X’s API is both expensive and restrictive regarding data collection, sharing, and thus impeding replication attempts, especially with large datasets”, it adds.

TikTok initially allowed access to its data only to US academics, but in July expanded its research API to Europe. However, the researchers believe its terms remain too restrictive to be compatible with research, stating that scholars must “refresh research API data at least every 15 days, and delete data [that is no longer available]”.

“The changes are adversely affecting academics who want to study the impact of social media on mental health, on misinformation, political views and so on,” said one of the paper’s authors, Brit Davidson, senior lecturer in the University of Bath’s School of Management. “It also inadvertently impacts app developers that have built their service on this source of information.

“It’s critical that research on people and society can access these large-scale datasets as there can be policy implications and far-reaching consequences if we get it wrong. Over time, we have many cases of where the lack of open science impacts our ability to verify and check for science credibility. We’ve seen science discredited, which causes concern as to whether work can be reproduced or replicated.”

Some of the changes to APIs have been necessary following the Cambridge Analytica scandal in 2018, said co-author Joanne Hinds, associate professor in management, information, decisions and operations at Bath.

“This led social media platforms to implement strict measures to prevent third-party users from gaining access to personal data without consent. They then enabled users to revoke app permissions, which gave users more control over their data to protect user privacy,” she said.

“However, this wave of changes is pushing researchers to abandon projects or to consider gathering data outside official means,” she added, “and that will, unless addressed, mean that we just simply can’t study important questions about these platforms which are used by millions of people every day.”

tom.williams@timeshighereducation.com