“Election Day” by Richard Yuan on Flickr, used under CC BY-NC 2.0
This certainly was an interesting semester to teach intro statistics. My students analyzed poll data, linked to in detail on realclearpolitics, to see if jumps were statistically significant, explored the correlation between the way different states vote, and analyzed the dependence of different demographic variables or responses to opinion poll questions on preferred candidate in polling crosstabs. We talked about why online polls are unreliable, how to check the accuracy of pollsters, and how important it is to look at averages and long-term trends in polling. And after the election, once the dust settled, we talked through some of the reasons why things didn’t go the way a lot of people had expected. Those discussions continue every week or so as new information appears.
I harped on my view of the importance of voting in all my classes. I showed how few young people vote, and why it’s important for them to be registered in their new home state. I linked them to the voter registration form (we can do it online in Maryland), voting guides, ways to find their polling place, early voting hours, everything I could think of. More than a few of my students said they don’t vote despite all this, but I know I got a lot of others registered and engaged for the first time.
Through it all, I tried my best to be impartial, not even disclosing who I intended to vote for. But c’mon, look at me up there: I’m a hipster-y woman in my 30s with a Ph.D. It doesn’t take a social scientist to take a pretty good guess as to my political allegiances. One student asked who I was voting for in a contentious local school board election, and I told them, along with my reasons for my choices, but never went further than that.
I’m not sure exactly why I felt the need to act so impartial. I even start my stats classes every semester with an activity, adapted from Dave Kung, where the students take one data set and use it to argue for multiple different conclusions. I emphasize the inherent uncertainty of statistical analysis all the time: pretty much everybody’s got an angle when they’re trying to prove something, and if you dig hard enough you can almost certainly find numbers to back up your side, even if they don’t hold up to scrutiny. I teach my class how to p-hack, so they’ll know how easy it is. And they know that type I and II errors are always lurking, and you’ll never know when they’ll come out to bite you. But I do believe that I am in a privileged position there at the front of the class, and I’d like for my influence to come only through my math instruction, and not overt moralizing.
In Tuesday’s class though, I did have to lay my cards on the table a little bit: between the allegations of election fraud from the left, and voter fraud from the right, I felt obligated to dig in. I went through the results of Nate Cohn’s regressions on Wisconsin, showing that there’s no evidence of a difference between paper and electronic ballots once education level and race are accounted for. I said that I thought it was irresponsible to even appear to allege electoral fraud with no evidence. And it was unconscionable for a president elect to claim voter fraud with no evidence, and terrifying that he would propose stripping of citizenship for speech. I know I got a little passionate during this part of class, and a few of my students did too. I’m not sure if any of them thought I went overboard – guess I’ll find out in my evals.
In tomorrow’s class we’ll unpack the study cited for the prevalence of non-citizen voting and maybe talk about the effect of education level on precinct results. I’m sure I’ll keep it more buttoned-up tomorrow. But like my colleague said so eloquently, sometimes you’ve got to say something.