As a college educated white woman, I was rather interested to see that my people have been making headlines lately. It all started with a Steve Bannon interview in Vanity Fair where he said “The Republican college-educated woman is done,” he said. “They’re gone. They were going anyway at some point in time. Trump triggers them. This is now the Trump movement.”

This quote launched quite a few op-eds and poll questions, which culminated in this chart from the Washington Post (source here):

Now I was pretty interested in this chart, since Andrew Gelman has taught me to be deeply suspicious of any reportedly wild swings in voting preference. However, looking a little closer I noted that he had the expanding gender gap as one of his 19 takeaways from the 2016 election, so I decided to take a closer look. What does it mean if a party loses a key demographic like that?

Well, it doesn’t entirely cause the drop you might think. I took a look at Gallup and Pew, and noted that Pew doesn’t show a dramatic drop in preference through 2017: 

Note: preference is not the same as people saying “I’m a Republican” or “I’m a Democrat” (both of which are down), but what people end up actually voting.

Anyway, I was curious why a massive drop-off in support from white college educated women wouldn’t translate in to a big jump in that graph, when it occurred to me that the underlying numbers are probably pretty different. Sure enough, per the Census Bureau, there are about twice as many Americans without a college degree (about 68%) vs with (about 32%). I can’t find exact breakdowns that take more than 2 factors in to account, but the same report shows that men and women have Bachelor’s degrees at about equal rates, and white people as a group have Bachelor’s degree at around a rate of 36%. The Washington Post puts the number of white people with college degrees at around 50 million, with the number without around 90 million. In other words, college educated white women are likely around 10% of the voter population, whereas non-college educated white men are likely around 20% of the total. A drop in one is made up for by the rise in the other.

So what does this mean for elections? Well, it’s complicated. Unlike gender, educational attainment varies wildly by state. According to Wiki, my state (Massachusetts) has the highest percentage of people with Bachelor’s Degrees in the country (about 40%), almost double the rate of West Virginia at about 20%. Going just off the Wiki list, the top 14 states with the highest levels of Bachelor’s degrees actually all voted for Hillary in 2016. Assuming those degrees are somewhat evenly distributed by gender, this may mean the loss has already been felt and that any further change will just be driving the states further blue. Since the electoral college system is in play, this could make the impact on national elections even smaller.

To complicate things in the other direction though, people with a Bachelor’s degree are much more likely to vote than those without one. For some reason the Census Bureau page that used to report this is down, but this site makes it look like the gap is pretty decent (80% for those with a Master’s or higher, vs 50% for a high school diploma). Again though, if the votes are unevenly distributed that could lead to less of an impact than you’d think.

I’ll be interested to see what the mid-term elections bring for updates to these trends.