Despite what academics might have you believe, science and especially social science has gotten increasingly political these days. It’s a pretty big problem that throws into question the validity of social science results.

A common algorithm taught in Statistics 101 is how to demonstrate a statistically significant association between two variables x and y. The first step the algorithm is to assume the null hypothesis: that there is no association between x and y.

A lot of my students are understandably confused by this step. If we are trying to show an association, why do we start by assuming the opposite?

The idea is to show that a world in which the correlation appeared by chance is so improbable that there must be a genuine assocation between the two variables: a proof by contradiction, if you will. The algorithm makes a convincing argument because it gives the other side a fighting chance, and lets it collapse under the weight of its own contradictions.

And the same holds true on a more meta-level. If you fervently believe hypothesis X to be true, and you end up showing hypothesis X is true, then with all due respect I’m not that impressed. I’m much more interested in the cases when the evidence doesn’t corroborate your deeply-held beliefs. A good exercise is to think of two hypotheses that you would love prove true but for which the evidence is inconclusive at best. I’ll go first.

Here are two hypotheses which I find personally unpleasant but statistically compelling:

1. The gender disparity in many fields will probably never disappear.
   I would love to have more female peers in math, but the evidence suggests that the gender gap in STEM is largely due to genetic factors.
2. School choice might not produce significantly better outcomes for students. More generally, increasing the quality of our educational system might not be a cure for all of our society’s ills.
   In theory I favor free market solutions to economic issues, but the evidence supporting charter schools and other school choice systems is currently pretty thin. I think this is part of a larger issue that there isn’t a ton of evidence supporting the popular belief that increasing the quality of our schools leads to markedly better outcomes for students. I’m passionate about learning and teaching, so accepting the limits of education is a tough pill for me to swallow.