The final speaker in this ECREA 2018 session is Pablo Jost, who begins by discussing the limitations of surveys (expensive, reactive, increasingly unable to reach representative panels of participants) and raises the question of whether digital trace data may be able to be used as an alternative. Twitter might be a poor substitute, as it remains an elite medium in many countries; Facebook use, however, is far more widespread, and could therefore be seen as considerably less inherently biased.
The present project sampled news media posts on Facebook in the lead-up to the 2017 German federal election, and in these identified the presence of relevant political actors. It then captured the positive and negative Facebook reactions to these posts, and matched this with a longitudinal survey of party and politician popularity. Do these indicators align, then?
It explored this first on an aggregate level, for regional media posts across 13 German states. This showed a negative correlation between public approval rates for parties and politicians, and the likes received by articles about them. This does not look promising for the use of Facebook data as an alternative to surveys, then.
However, controlling such patterns for the average activity of party supporters on Facebook appears somewhat more promising; the more positive approval is for politicians in a given state, the more Facebook reactions (of any type, positive and negative) articles about such politicians receive. The lack of distinction between positive and negative reactions here may indicate that any type of reaction on Facebook points to the salience of a candidate or party, not simply to approval. Of course, these reactions are in the first place to the articles that cover them, not to the politicians themselves – so there could be negative reactions to an article about a politician from supporters of that politician, too.