This really doesn’t surprise me at all. Had I been told of the design of the experiment my predictions would have been pretty much exactly what their results showed.
Far from not accepting Trump style debate and body language from a woman, audiences liked it just fine. Gender expectation are never so simple as third wave pseudo-feminists like to pretend. Women can incorporate lots of stereotypical male style and be liked and admired for it, and men can incorporate approximately no elements of stereotypical female style without being hated.
But beyond that, switching the sexes of the participants brings out the core features of each candidate. It’s not a Republican vs a Democrat or a Man vs a Woman, it’s a Straight Shooter vs a Bean Counter.
These are broad cultural stereotypes.
In short, a Straight Shooters say they mean and mean what they say. If you’re uncomfortable around them then that’s your problem. Straight Shooters are open and genuine and liable to say what everyone else is thinking.
Bean Counters, on the other hand, know the price of everything and the value of nothing. There is no spontaneity or genuine feeling, just a dry recitation of dry facts and if you don’t like it you’re being irrational.
American voters will always go with a Straight Shooter over a Bean Counter.
In the 2016 election Sanders was also a Straight Shooter while Jeb was a Bean Counter (please applaud).
I think the method is kind of genius and in the future parties should do just this kind of experiment to figure out the electability of various candidates.