This week, Secretary of Education Arne Duncan announced competitive grants to encourage states to increase their college graduation rates, with a goal to add eight million college graduates by 2020.

Sure, there are plenty of legitimate policy-related reasons why we might want to increase the number of college graduates. After all, as Secretary Duncan put it, “We all know that the best jobs and fastest-growing firms will gravitate to countries, communities, and states with a highly qualified work force.”

But, for those who can’t imagine Obama doing anything without an ulterior motive, consider the graphs below that show that increasing the number of college graduates might also increase the number of Democratic voters and reduce the number of Republican voters.

The first graph shows the state-level relationship between the percentage of individuals identifying as Democrats (data from Gallup) and the percentage of individuals with bachelor’s degrees. There’s a clear, statistically significant relationship that explains 28 percent of the state-level variation in Democratic identification. For every one percentage point increase in college graduates in a state, the percentage of Democratic identifiers increases by 0.75 percent.

The second graph shows the state-level relationship of Republican identifiers and college graduates. As you’d expect, it’s pretty much the reverse. For every one percentage point increase in college graduates, the percentage of individuals identifying as Republicans decreases by 0.76 percentage points. This simple regression explains 30 percent of the variation.

Now let’s look at the relationship of state-level education to state-level liberals and conservatives. Here the relationship is even more significant:

For every one percentage point increase in state-level college graduates, the percentage of liberals also increases by 0.75 percentage points. Impressively, education level explains 66 percent of variation in state-level percentage of liberals.

By contrast, for every one percentage point in college graduates, there is a 0.88 percentage point decline in the share of conservatives, and this by itself explains 58 percent of the state-level variation in the number of conservatives.

Does this mean that there is a simple causal story that education makes people more liberal either because (in the conservative telling) it turns them into elitist snobs, or (in the liberal telling) it gives them enough knowledge to understand how the world works?

Maybe. Maybe not. Perhaps liberal, Democratic states invest more in education, which is why those states have more college graduates. It’s also important to note that 1) these are state-level, not individual-level relationships, and 2) this is a static relationship, not a time series.

Nonetheless, the graphs are quite telling. The more college graduates, the more Democratic (and especially more liberal) the state. The fewer college graduates, the more Republican and (and especially more conservative) the state. There’s clearly something going on here, and I’m actually quite curious to hear how conservatives would respond.

Increasing the number of college graduates by eight million would bring the number of college graduates in the United States from approximately 83 million (27 percent) to 91 million (about 29 percent in 2020). That’s two percentage points, and if the relationship between state-level education and voting is indeed causal, it would mean a 1.5 percentage point increase in the share of Democratic identifiers and a similar decline among Republican identifiers. This could tip some states.

Well, now I’ve given conservatives an argument against increasing the number of college graduates.