The next speaker at the Indicators of Social Cohesion symposium is the great Cornelius Puschmann, presenting work from the excellent POLTRACK project on polarisation and individualised online information environments, which has been conducting a longitudinal panel study as well as tracking participants’ online activities in Germany over a period of 20 months since March 2023.
This presentation focusses especially on participants’ search practices, with search of course representing a key pathway to political information. This project focussed especially on searches for information on German political actors, and was especially interested in the role of affect towards those actors, and of pre-existing party affiliations. The dataset contained some 10 million total searches, but only some 13,000 of these (or 0.12%) searched for political entities (i.e. leading politicians or parties), even though 25% of all participants engaged in such political searches.
The largest number of participants searched for leading SPD and Greens politicians, and for the far-right AfD party (though neither of this indicates inherent support for these); participants with greater political interest and education searched more for politicians and parties, while older people, women, and people with lower news interest searched less. Searches for individual politicians were driven largely by stronger positive as well as negative sentiment towards them, but not by party affiliation; political interest was also a strong driver, and women searched slightly more than men for them. (This also varied strongly between individual politicians, though.)
The search data also uncovered the influence of a (likely Russian-origin) conspiracy theory about the supposed Nazi affiliations of leading German politicians’ ancestors, interestingly.
Future research needs to explore this further, and consider the full search process more generally. Clearly, affect is very important here, and requires additional investigation. How all this is further changed by the embedding of AI into search engines must also be investigated further as this happens more fully.
