This is a preprint, so any feedback and comments are welcome. This has been an effort to put together, so thanks are very much in order. Thanks to all my co-authors, and especially Ralf and @alantump.bsky.social :) 15/n

Here is an overview of the results. On top we have discrimination ability. For example, older adults have higher discrimination ability. On the bottom we have response bias, older adults have a false news bias, they tend to be more cautious. 14/n

We also looked at some further meta-questions. Experiments on MTurk have better discrimination ability than on Lucid. And headlines displayed with a source lead to better discrimination ability. 13/n

Our biggest effect is familiarity. If you’ve seen a news headline before, you’re much more likely to treat it as true. 12/n

Partisan bias, aka ideological congruency, is a very reliable effect. People treat ingroup news to be true more often and disregard outgroup news as false more often. This again is worrying. The effect holds across partisan lines and is ready for interventions. 11/n

We find that analytical thinking (CRT) is reliably associated with higher discrimination ability and a false news bias (caution). It’s not yet fully clear what the CRT is measuring, unlocking this would help to build better interventions. 10/n

Confirming a worrying trend in other studies, we find that Democrats show a higher discrimination ability than Republicans, and a false-news bias (caution). Seen in light of polarisation, this hints towards Democrats and Republicans inhabiting different perceptions of truth. 9/n

A null effect of education on discrimination ability. Higher education is also linked with treating news as true more often, a kind of naivety (a small effect). Resonates with research suggesting current educational practices need to better equip people for the digital world. 8/n

Older adults have a higher discrimination ability than younger adults and also a false news bias (they are more cautious). This is quite interesting, especially as older adults are known to share the most misinformation. (Note: each line is a study, split by true and false). 7/n