Social media research data: Facebook lets science hang

The question of who shares which posts, liked or commented when and when, can influence election outcomes today. Scientists are interested in the evaluation of such data, but especially on Facebook, the exchange is more than slow.

Anyone who travels on social networks knows the price for the seemingly free use: the users pay with their data, with which the companies take a lot of money. But for another group of people, too, the data is invaluable: scientists.

With the data from the social networks, they can track social processes and reduce them to certain causes. At the Chair of Political Data Science at the University of Applied Sciences Munich, political scientists investigate the interactions between social media and actual events. One of the central questions: “What effects did Desinformation have on the voting decision of citizens in the German Bundestag election 2017?”

12 research teams should have access to a huge dataset

As Mark Zuckerberg in the wake of the scandal involving the analysis company Cambridge Analytica, which had illegally evaluated over 85 million Facebook profiles, had to testify before the US Congress, not only exulted the scientists in Munich.


At the hearing, the Facebook boss had announced as a kind of compensation the “Social Science One” project, in which twelve international research teams from Italy via Chile to Germany should have access to a comprehensive data set. Everyone is concerned with the question of whether and, if so, how disinformation and false reports in the past have helped to manipulate voters.

More on the subject: New verdict – off for Facebook fanpages?

Few data that are hardly meaningful

Meanwhile, but disillusionment has occurred among the researchers. Morteza Shahrezaye of the TU Munich told LEAD: “Unfortunately we noticed that the data are not of good quality.” They are incomplete and therefore unsuitable for an analysis including valid results.

Facebook would have provided only a fraction of the links, which had been announced in the order of about one billion. Shahrezaye says, “We may have gotten one or two percent of that data.”

Few very active users and bots can control the discussion

What Facebook delivered, are mainly URLs from public user pages. For their research, the researchers not only need posts with political content, but also need to know how the users respond. Whether they share her like, if – and if – how they comment on her.

“With such data, for example, false news and hate speech can be identified and analyzed,” says the political scientist. With the data that they have now received from Facebook, they could do virtually nothing. “We can not say anymore if someone comments on a post positively or negatively, we can only see how many Likes got a post in total, but no longer, which user liked which post.”

Exactly that would be interesting, for example, to recognize bots. Or to examine the role of so-called hyperactive users who, according to Shahrezaye, are able to manipulate the political discussion without the activation of bots.

More on the subject: The hunger for medical data

After the scandal surrounding Cambridge Analytica, Facebook has closed open interfaces

Particularly annoying is the current lack of information because scientists were able to obtain such data from Facebook in the past. This was possible through an extra service of the websites that had specially set up for developers. As a developer, you could create an account at the so-called developer website and got a password to download the data.

After the Cambridge Analytica scandal, Facebook closed these open programming interfaces in 2018. Unlike large companies, researchers now have no opportunity to request the release of certain data, says Shahrezaye. This results in the almost absurd situation that also the research teams, to which Facebook data from the “Social Science One” project were promised, now have less data than before available.

The researchers of the TU Munich have meanwhile informed Facebook about the inadequacy of the data and urged the corporation to deliver the promised data. “We understand that the data for Facebook are valuable,” says the Munich political scientist. “But they are also on a political level, which is why we need them for our analysis.”

More on the subject: Digital consumer protection in practice

Legally, the corporations are not obliged to anything

The tech companies are not required by law to provide the data of their users for research purposes. However, politicians such as Tankred Schipanski, the CDU / CSU spokesman on digital policy, have already said that in the future, it may not be enough to free up the platforms for voluntarily publishing the data to researchers.

And also the Green politician Anna Christmann called in the “Süddeutsche Zeitung” on the federal government to “at European level for a reliable access to social media data for research purposes.”

Facebook claims to want to work for science

At the Technical University of Munich, the hope that Facebook will publish the data even without a previously adopted law has not yet been buried. The SZ quotes Facebook social media campaign manager Chaya Nayak as saying, “Facebook’s leadership believes it’s very important to share data with the scientific community.”

In addition, Facebook is working on the development of differential privacy technology. This should ensure that data shared with scientists can not be deanymized. Facebook could then also provide additional information such as the age or the region of the users and at the same time avert objections from privacy advocates.

Leave a Reply

Your email address will not be published. Required fields are marked *