"Merely asking the question about spousal income led to enormous shifts in men’s preferences in the upcoming presidential election. Men who weren’t asked about spousal income until late in the survey preferred Hillary Clinton over Donald Trump in a hypothetical general election matchup by a 16-point margin; men who were asked about spousal income only a few questions before being asked about the Clinton-Trump matchup preferred Trump by an eight-point margin — a 24-point shift in preferences. The conclusion that this is about gender is reinforced by the fact that the spousal income question had no effect at all on a matchup between Trump and Bernie Sanders. Men who had been primed to think about a threat to their masculinity preferred Sanders by four points; unprimed men, by three."
From an article in Harvard Business Review by polisci prof Dan Cassino, published in April. (I'm posting this because a colleague posted it on Facebook and I got into a long back-and-forth about it.)