Here's Feser's $.02:
But of course there is another obvious way to interpret the results in question [He's speaking of the results of the Phil Papers survey that revealed that the majority of professional philosophers lean towards or accept atheism whereas the majority of professional philosophers who specialize in philosophy of religion lean towards or accept theism] – as clear evidence that those philosophers who have actually studied the arguments for theism in depth, and thus understand them the best – as philosophers of religion and medieval specialists naturally would – are far more likely to conclude that theism is true, or at least to be less certain that atheism is true, than other philosophers are. And if that’s what the experts on the subject think, then what the “all respondents” data shows is that most academic philosophers have a degree of confidence in atheism that is rationally unwarranted.
There's lots of interesting stuff to think about here. Should the confidence of non-experts reflect the attitudes of experts? Shouldn't this depend, in part, upon the size of the 'knowledge gap' between expert and non-expert? Suppose there's a gap. (Plausible). Is that gap anything like the gap between global warming deniers and climatologists? I don't think so, but that's still perfectly consistent with the idea that non-experts ought to be less confident in their attitudes upon learning what we've learned when the results were released.
Here's something that I think matters but I don't know what to make of it.
*Suppose the majority of the experts agree that a certain argument for the non-existence of X (electrons, phlogiston, fairies, objective moral standards, heaven, a justification for intentionally terminating a pregnancy) fails.
*Suppose that this is based on the widespread conviction that there's some adequate reason or other to believe in X.
This is all perfectly consistent with widespread disagreement amongst experts on two points:
(CP1) what the adequate reasons are for believing in X;
(CP2) what's wrong with the arguments for the non-existence of X.
So, some what-iffing based on next to nothing.
What if the experts were evenly divided in the following ways. We divide the experts into the A team and B team if we look at their attitudes concerning (CP1). We divide the experts into the C team and D team if we look at their attitudes concerning (CP2). The members of the A team thought that the reasons that the members of the B team had for believing in X were inadequate and poor for reasons readily available in the literature. The members of the B team thought that the reasons that the members of the A team had for believing in X were inadequate and poor for reasons readily available in the literature. The members of the C team thought that the members of the D team failed to neutralize the arguments for the non-existence of X because those responses rested on false premises that were shown to be false/unwarranted in the literature. The members of the D team thought that the members of the C team failed to neutralize the arguments for the non-existence of X because those responses rested on false premises that were shown to be false/unwarranted in the literature.
I can imagine some epistemologists saying that if the experts had a high degree of confidence in the hypothesis that X existed, that would be misplaced confidence given some principles about the weight of peer opinion and some evidentialist assumptions (which, admittedly, might be hard to rationally accept as a package given the principles about the weight of peer opinion which themselves are problematic given contingent facts about what opinions are floating around). I can imagine some epistemologists saying that when expert or ('expert') opinion is known to be not warranted by the evidence, the gap in confidence between expert and non-expert (which is really the gap between specialist and non-specialist) does not entail that the attitudes of non-specialists/non-experts are unwarranted/unreasonable/epistemically impermissible.
At any rate, I think the issue is a bit more complicated than some people have assumed. Indeed, I fear that I've oversimplified things. You've been good to read this far. Enjoy some Elvis Costello, you deserve a treat: