Not a long post, just dropping a few screenshots from a study of internet trolls:

Note: the asterisks show that the result was statistically significant, as in, the relationship definitely exists and it’s very unlikely that they got the results they got out of pure chance. Here you can note that all these factors are definitely linked with trolling, and barely if at all linked to the others (aside from a subset of people that primarily debate online having a mildly statistically significant relationship with vicarious sadism, those small numbers could be just random errors). You can get a better idea of it by looking at the original graph and seeing where the values and those brackets marking the range of error are.