New preprint: psyarxiv.com/v45bk/. Are sharers of "fake" news ignorant or motivated by political goals? We combine individual US Twitter & survey data (N > 2300). We find no evidence that "fake" news sharers are ignorant. The best predictor is hatred of the other party. [1/8] 
We test whether sharing of links to "fake" news websites reflects (1) ignorance; (2) disruptive sentiments (e.g., trolling); and (3) partisan motivations. We find no evidence for #1. Some evidence for #2. And massive evidence for #3. [2/8]
This is the key figure. Sharers of "fake" news sources on Twitter are not ignorant. They are knowledgable, digitally literate and not low in cognitive reflection. They are not trolls but cynics. Most importantly: They are partisans with strong outparty animosity. [3/8]
We also observe two forms of ideological asymmetries. First, Republicans are more frequent sharers of "fake" news. For Republicans, there are strong associations between sharing "real" partisan news and sharing pro-Republican "fake" news. Less so for Democrats. [4/8]
Second, Democrats are more partisan in their overall sharing patterns. Democrats overwhelmingly share centrist and pro-Democratic "real" news. Republicans, in contrast, share more broadly. [5/8]
Why do we see these ideological asymmetries? It is difficult to tease apart supply and demand. But using sentiment analysis, we found that the only news source type more positive towards Republicans than Democrats was pro-Republican "fake" news. [6/8]
A potential interpretation: Democrats motivated by partisan goals can easily find ammunition in centrist news stories against their outparty. Republicans with similar motivations, in contrast, need to move to fringe news sites in order to consistently find such ammunition. [7/8]
In sum: People who link to "fake" news are not ignorant but motivated by strong partisan goals. While Republicans are more frequent sharers, this does not necessarily reflect stronger partisan goals. Democrats can satisfy these goals by linking to the mainstream media. [8/8]
