Here's an interesting study dug up by bspencer over at Lawyers, Guns and Money, who right now is embroiled in a massive sprawling multi-thread discussion on what it means when men complain about feminists being dismissive of them.
On that more general point, I don't want to say too much (short version: there is a big difference between recognising some people can use "check your privilege" as a way to shut down male comments and thinking it's so major and commonplace a problem as to make it worth mentioning in your introductory talk for a secular conference for women, because they'll totally never have thought about that), but the secularist issue has spawned some interesting tangents, including bspenser's suggestion that, per the link above, Americans dislike atheists more than they do Muslims.
Again, this isn't something I want to dwell on - there's all sorts of reasons why labelling of a hypothetical person as a Muslim does not translate into how actual Muslims (for which one can frequently simple read: brown people) are considered, and of course there's a greater distance still between how people feel about a minority and how oppressed that minority is (I should note that bspenser isn't claiming otherwise on that score).
I'm mentioning all of this because the suggestion that atheists are more reviled than Muslims struck me as so ludicrous on the face of it that the Patheos link proves strikes me as very important (though it really demonstrates a lack of trust than hatred, I think). Even with - or maybe because of - the amount of time I spend reading up on American culture and politics, I have to continually remind myself how very differently atheism is viewed in (some or much of) the United States to my own experiences in Britain, especially since it's so completely counter to how one would expect things to be given which of those two countries has Christianity as its official state religion.
Let's break down the Patheos piece a little. Firstly, a few sentences on the conjunction fallacy, which basically goes like this: "a figure bounds out of a TARDIS and saves Earth from the Daleks. Is it more likely our saviour is a) a Time Lord, or b) a Time Lord and the Doctor". Much is made of the fact that most people will choose b), even though by its very nature, a) is more plausible, since if b) is true, a) is true as well.
To simplify significantly, the idea here is that the fallacy is more seductive the more the question seems to describe the subset of people in the second group. If the above was re-written to have answers "a) a Time Lord" and "b) a Time Lord who is under 170cm tall", people would be more likely to realise what's going on. There is, in other words, a correlation between people's inherent assumptions - and by extension their prejudices - and how likely they are to commit the fallacy. Whether this is because the added assumptions make it harder for someone to not read option a) as say, "a Time Lord but not the Doctor", or if people's grasp of probability is being overloaded, I don't know .
So, here the idea in this research was to see how often the fallacy was applied when considering different types of people. If the question is re-written as "Somebody does some pretty shitty things, is he a) a teacher, or b) a teacher and some quality X", where X is Christian, Muslim, rapist, or atheist.
You can see the results at the link. Even when you offer the idea that the man is a rapist - that is a man who's already committed criminal acts worse than the ones described in the study - fewer people make the intuitive leap. I suppose one could make the argument that it's easier to spot the fallacy when given other unrelated crimes than when considering a persons' (lack of) belief system, but even so, when "just because he's a rapist doesn't make him a thief" carries more rhetorical power than "just because he's an atheist doesn't make him a thief", something strange is definitely going on. 
All of which is a sobering reminder of how lucky I really am. Not only am I a heterosexual white cis man with a middle class background and a decent job, but I also get to be an atheist in a country and in a societal circle in which the only reason people would have to think me likely to commit acts of grotesque immorality is having met me. Thank the lack of God for that!
 There's an alternative version of the question that goes like this: "Emma is a sexually confident, assertive woman, who likes to wear high heels and tops that show off her cleavage. Is she most likely to be a) a librarian or b) a stripper? The idea here is that there are (allegedly) fewer strippers in the UK than librarians, and therefore if the other information offered to us is irrelevant, a) is the correct answer. You can learn a great deal about people by seeing how much of the additional information they will claim isn't irrelevant at all.
 Admittedly, that's just from looking at the means. Using the confidence intervals, we can only say that people consider lawbreaking to be no less common a feature of atheists than they do of rapists.