Not only does it take "a bunch of research scientists creating a bunch of avatars to determine that women don't find flailing arms attractive," it even still isn't established having done that! This is just one data point and in future we might find a different way to interpret it. You are confusing on one hand having an expectation (we've all got that) and on the other hand having confirmation of that expectation (much more difficult).
E.g. if I showed you a study saying that being cold causes you to have a cold, would you say that nothing new had been discovered? If you said yes, you'd be wrong (and for argument's sake suppose you did say yes, even if you already knew this), because in fact the research has been done and being cold does not cause a cold! That people tend to get more colds in winter is due to other factors that are indeed caused by the colder weather, but not by way of cold directly making you more susceptible to getting infected. So avoiding being cold will do nothing to help you not get a cold. So in this case your expectation was shown to be wrong (we are assuming for the sake of argument), and in this instance we can agree that you learned something new. Now what if the study had shown that being cold causes a cold? Since it is abundantly clear that you didn't actually know that being cold causes a cold (since in reality it doesn't), you would have learned something new from this information - you would have learned information that increases your confidence in what you already suspected to be true. So when research comes out saying something you already suspected, you will now know, I hope, that you are still learning something new.
Back on the concrete topic: asking women what they find attractive in male dancing is a perfectly reasonable thing to also do. It is a good way to generate hypotheses about the subject, and it can corroborate information gained in other ways. Yet what if what women say don't match what this experiment comes out with? I know that I certainly place much more stock in this kind of research than I would a questionnaire. The point is to approach a matter in as many different directions as possible, because in that way you can get a much higher confidence about what is actually true, instead of just discovering what people's expectations are. That's good science. It's not common sense, because common sense is about just assuming that your expectations are true. Surely being cold causes colds, right? It is common sense not to investigate "obvious" things, and that is why science trumps common sense.