First, what sort of sampling bias is introduced by studying only men who are willing to view such materials? It seems highly implausible to me that this effect is zero.

Would this have an effect on the difference between homophobes and non-homophobes? Intuitively, it should have a uniform effect across the board so that the comparison of differences is still valid (though what Unnamed mentions in response to the parent undermines this), though this is hard to know without checking.