> The risk of bias in this research seems rather high. Of the form of "our believes are correct and everybody who disagrees is underinformed".

It's not circular reasoning like this. They asked a bunch of middle-school science level questions in related fields.

The interesting bit isn't that people who disagrees are misinformed. It's that they assess they will do better on the test, and bet more that they will do better, than people who don't.