What the California Primary Tells Us About November

What the California Primary Tells Us About November

By Sean Trende - June 19, 2014

A few days ago, Daily Kos Elections’ David Nir noted a SurveyUSA poll finding Rep. Scott Peters trailing Republican Carl DeMaio, 51-44, but observed:

In California's recently concluded primary, Peters took 42 percent of the vote. If he only performs 2 points better in November, that would be a remarkably small increase compared to the kinds of jumps in Democratic performance we saw between the primary and the general in 2012. Yes, the fall surge will be smaller for Democrats in a midterm year, but the smallish 5-point bump in CA-52 last time was at least partly due to the heavily contested fight between Peters and fellow Democrat Lori Saldana, something we didn't have this time.

I thought the observation about the relationship between Democratic primary voting in 2012 and the final margin was interesting, since we see a similar effect in Washington state. Democrats basically perform a fairly consistent two points better in November than they do in their “top-two” primary, which is similar to California’s (e.g., everyone runs on the same ballot, and the top two vote-getters proceed to the fall election).

I hadn’t really paid attention to the relationship, if any, between California’s primary and the final results. I’d just assumed that because California’s primary is so much earlier than Washington’s, the relationship would be weak.

As it turns out, that assumption was incorrect. The following scatterplot shows the relationship between the total Democratic share of the votes cast in the primary and the Democratic primary winner’s vote share in November (races where two Democrats or two Republicans proceeded to the general election are excluded, and perhaps more controversially, Bill Bloomfield is treated as a Republican):

The r-square here is a pretty nice .926.

So the question is, what else can we do to improve the fit? It seems that Democratic vote share in the primaries might be inflated by multiple Democratic candidates and reduced by multiple Republican candidates. Therefore, the amount of “increase” might be reduced if there are more Democrats, and vice versa. Likewise, a strong independent effort could reduce the number of votes available for Democrats and result in a larger Democratic vote share in the fall. Finally, a high Hispanic population could have resulted in a shift toward Democrats, as Hispanics became more engaged in the fall election.

So I performed a regression analysis with the final November vote share for Democrats as my main variable, while controlling for total Democratic vote share in the fall, total number of Democratic and Republican candidates on the primary ballot, the Hispanic Citizen Voting Age population in the district, and the percentage of the vote won by independent candidates in the primary.

We get an r-square of .9545, with the variables pointing in the expected directions (negative for the number of Democratic candidates in the primary, positive for the rest). The only significant outlier is the result from California’s 21st district, where a relatively weak Democrat won the primary and then lost badly in the fall.

So, now some predictions, and a huge caveat.

If we take the results from 2012 and superimpose them over the 2014 results for the races that RealClearPolitics currently rates as at least somewhat competitive, we get the following:

This would suggest very little change in the California delegation. It would also suggest that the race in the 52nd District should continue to tighten.

Now for the huge caveat. Unlike Washington State, where we’ve seen Democratic vote shares increase at a pretty steady rate from primary to general over the course of two decades, we only have a single observation in California. In particular, the presidential year results may not be applicable to midterm elections in a state like California. Moreover, Democratic prospects improved generally in 2012 from June to November, especially after the Democratic Convention, which seemed to lift Democratic candidates from the top to the bottom of the ticket.

The original model suggests that all other things being equal, things should improve about 10 points for Democrats from the primary to the general election. What if California isn’t like Washington State, and increases from primary-to-general elections aren’t constant across years? Let’s assume that, instead, the constant for midterm elections in California is five points. We’d see something like this:

This would suggest real vulnerabilities for California Democrats in the fall.

Overall, I’d say this: California Democrats, including Scott Peters, have reason for optimism in the fall given the primary results and the results from 2012. But it should be tempered by the reality that the Democratic base is harder to energize in off-year elections, and this may make the lessons of 2012 inapplicable to 2014. 

Sean Trende is senior elections analyst for RealClearPolitics. He is a co-author of the 2014 Almanac of American Politics and author of The Lost Majority. He can be reached at Follow him on Twitter @SeanTrende.

UC's New Motto: Take the Money and Run
Debra Saunders · December 7, 2014
A Reply to Rush Limbaugh
Mona Charen · December 9, 2014
Against the Wave: Three Winning House Democrats
David Byler · December 14, 2014
Obama, Congressional Dems Show Cracks in Unity
Jim Kuhnhenn · December 7, 2014

Sean Trende

Author Archive

Follow Real Clear Politics

Latest On Twitter