Advanced quantitative analysis often controls for variables that aren’t of central interest. But what does it mean to “control for” a variable? XKCD offers a fun example:

So, do subscribers to *Martha Stewart Living* live alongside furries? Probably not. In any case, these maps don’t offer any evidence in favor of this conclusion. This is because of a variable that hasn’t been controlled for: population density. XKCD captions the cartoon:

To control for population, one would have to divide the number of subscribers/furries by the total population. This would give us the *percentage* of the population that is described by both proclivities, instead of the sheer number of devotees. Then the maps would actually show variance in the proportion of the population instead of variance in the population itself.

In other words, we would have controlled for population in order to get a closer look at what we’re really interested in: furries, of course.

*Lisa Wade is a professor at Occidental College and the co-author of Gender: Ideas, Interactions, Institutions. Find her on Twitter, Facebook, and Instagram.*