there is also only a ~25% difference between the groups re: hostility. and the authors address the lack of their confidence in their criteria::
While the inter-rater reliability for hostility was
good, there is a risk that we may not have captured the full spec-
trum of responses, as we specically excluded comments that
consisted solely of threats, insults, or ridicule. As such, although
we cannot say with certainty that conventionalist comments are
more hostile on average than conspiracist comments, we can say
with some condence that this is true among comments that also
contained some amount of persuasive content.
they had to strip out a lot of the data, several thousand comments boiled down to a few hundred.
|