I will grant, my big bugaboo on this at the moment is various statistical studies that are similar to what I see in economics and the worst kind of behavioral theorists. Now I am not opposed to statistical studies per se. But i became deeply suspicious of them when I got into designing them and debunking them starting with the national broadband plan in 2009. While it is totally possible to do a good statistical study on user preferences, user exerience or other things, it is just much much easier to do a really bad study and then dramatically over emphasize its meaning.
Sometimes this is deliberate. "Tell me the outcome you want and I will design the study" is a fairly well known industry trick. But a lot of time it proceeds from a rather simplistic idea about express preferences, implied preferences, and P values. And then discounting anything else that disagrees with your conclusion.
In fairness, however, much of the distortion also goes to press reports who find "nuance" too newsie.
Now lets stipulate that human beings are a tricky bunch, and not super good at self-reflection. People who proudly think of themselves as not racist or sexist are often a bundle of absolutely racist and sexist assumptions. What is less understood is how powerful these may be in any indiividual and what other connections may or may not counteract them. Additionally, we understand that exposure to different groups tends to reduce negative assocaitions -- but not always. People can, and do, make exceptions in their mind.
Importantly, as all these caveats indicate, it's a complicated phenomena. What I find more frustrtaing is the contradiction between discussion of racism and sexism (and even race and gender) as a social construct, while simultaneously treating them as immutable charatceristics.
Which brings me to the point of this rant. Vox's coverage of this study on Trump voters (again):
Racism and Sexism Predict Support For Trump Much More Than Economic Disatisfaction
Part of my issue here is a long-standing debate about the general maleability of people. I saw with my own eyes people say things in 2016 that they found offensive in 2015 without even realizing they had changed their mind. Why? Because racisim and sexism are indeed learned behaviors. If you hook up with Trump for one reason, and stay for others, you will start speaking the language and internalizing the culture.
It's also annoying because no one is saying that racism and sexism weren't involved. Of course they were. So this study finding that more racist or sexist you were the more likely you were to vote for Trump, whereas the "economic distress" category is less good as a predictor, should not be terribly interesting. (I also have problems with the measure of economic distress, which is all over the map on the post-mortem. There are lots of people who are employed at jobs with decent wages who voted Trump not because they are currently in distress, but because they fear the trends of offshoring and automation and believe that "they" are getting advantages "they" don't deserve.) As a political organizer, I don't find this information terribly useful or interesting because it fails to get at what's important to me -- how many people do I have to flip to win next time, and how do I flip them conssitent with my actual morals and goals. If the answer were "get them to be less racist or sexist," I'd be happier. But no one who focuses on race or gender in the political race seems to have an answer to the "how do you make them less racist or sexist" question other than "keep telling them what racists and sexists they are." As the author of the Vox article linked to above wrote in an earlier article, there is good scientifc evidence that while you can work to diminish racism and sexism, telling people how racist and sexist they are is not one of them.
All of which gets me back to "political scientists seem to know less and less about how politics works." The study is, at best, confirmatory that if you were racist and/or sexist, you gravitated toward Donald Trump, and being racist or sexist was a stronger indicator than mere conomic distress. But it doesn't tell us anything useful about combination, change over time, or why voters who voted for Barack Obama in 2008 and 2012 suddenly became more racist.
I want more complex studies and better reporting, dammit.