@Setanta,
We know that the manipulation between the studies was the use of a pronoun, and this manipulation had the effect of making women react differently, depending on whether the pronoun was gender exclusive or not. It should follow that the universal "he" pronoun was having a negative effect in line with the detriments that sexism is the cause of.
Psychology and sociology are soft sciences, and we must be even more careful when interpreting the data of these disciplines compared with the hard sciences. But that's no reason to dismiss them out of hand. A favourite phrase in psychology is "this suggests that" -and I think we could do worse when discussing the nature of things in the social sphere.
With regard to what you say in your edit, I'm pretty sure I actually wrote an essay on that very subject for my degree. I'm not at liberty to dig it out now, but what I will say is that the situation you describe where evidence can be drafted in to either side of a debate is not unusual in social science, or any other scientific discipline. Progress towards a refined understanding of ___ is made by opposing strands of research being argued over, within a scientific framework (analysing each other's data, using new and improved methods to examine old findings) so your objections aren't completely damning, and indeed it would require a closer look at literature than (presumably) any of us have time (or inclination) to carry out, in order to say what's right, one way or the other.
I see quite a bit of wriggling around here in response to the studies that sozobe has found. Methodological concerns are important, but unless someone can do better than just say "Well, our knowledge of this study is less than exhaustive.", I'll remain confident that they support the sexist pronoun contentions some of us here are putting.