The Washington PostDemocracy Dies in Darkness

Do newly registered voters actually vote? In California: Yes.

Placeholder while article actions load

As voter registration deadlines before the November general election begin to pass, new disputes over voter registration efforts have erupted. In the wake of the shooting of Michael Brown in Ferguson, Mo., Democrats launched a voter registration effort, angering some local Republicans. More recently, a registration effort led by a Democratic Georgia state representative -- which has turned in a number of new voter applications -- is being investigated by the Republican secretary of state.

Every vote counts, as they say, but one longstanding bit of conventional wisdom in political campaigns is that newly registered voters aren't likely to vote. In part, that mentality stems from the fact that, with limited resources, it makes more sense for campaigns to target voters who they know have voted before. Voting is, after all, a habit.

But in light of this year's elections, we thought the whole thing was worth investigating.

Political Data is a California-based ... well, political data firm. The firm's Paul Mitchell was generous enough to compile data on the voting history of recent California elections, based on people currently registered to vote in the state. We parsed the data to figure out how much new voter registration efforts actually correlate to turnout, in California at least.

This is the very top line, comparing all voters in the state with people registered since the last statewide election. Interestingly, you can see that turnout among the newly registered voters matches or even exceeds the turnout for the voting population as a whole -- in general elections. They are substantially less likely to vote in primaries.

But in pulling the numbers Mitchell included a set of data that relates to something the firm had noticed in its work. People registered to vote shortly before Election Day were more likely to vote than people who were registered at other times during the year. They're marked as "late registration" on the graph below, reflecting people registered after January (for a primary) or after September (for a general).

There's still the dichotomy between generals and primaries, but in general elections, late registrants have been significantly more likely to vote than the voter pool on the whole.

There's one demographic set where this difference between new and established voters is particularly noticeable: age. (When we say "established voters" here, we really mean "all voters," which is mostly voters who have been registered for a long time.) This is what turnout has looked like overall for the last eight statewide elections in California, by age.

Older voters are consistently more likely to vote -- and the age/turnout gap is wider in primaries than in generals.

When you look at the difference in turnout between newly registered voters and the voter pool on the whole, though, the pattern flips. Younger voters who are newly registered are more likely to vote at higher rates compared to their age group than are older voters.

In the first graph, you can see a spike in the 18 to 24 age group compared to the 25 to 29 and 30 to 34 age groups, particularly. That, too, is flipped when you compare new and all voters: the bulge is in the age groups that turn out least heavily in general. That bulge is in part because the groups on the whole don't turn out as much. With lower turnout in those age groups and lower turnout in primaries, it makes sense that the biggest increase would be in those age groups in those elections.

In Georgia and Missouri, there are two demographic differentiations that are probably of the most concern to Republicans: party and race. (Given that more black voters correlates very strongly with more Democratic votes.)

It's tricky to extrapolate from California to other states on party, but it's clear, when comparing new and overall voter turnout by party, that new independent voters outperform their peers on the whole. (California has a number of third parties and allows for an option of "decline to state" a party when registering.)

In looking at race, it seems that black and Latino voters follow the pattern we saw in the first graph: new voters vote more frequently than the overall pool in generals, but not primaries.

The number one takeaway here is that California is not Georgia and is not Missouri. But the number two takeaway is that there is at least some evidence that voters registered between now and Election Day are more likely to vote than voters on the whole. Which makes a spike in turnout understandably something that makes members of the other party nervous.