Posted by: sweeneyblog | August 17, 2012

Election Analysis: What the Primary tells us about the General

Hello Loyal Readers,

Have I mentioned that I love charts? My passion for charts and spreadsheets is only surpassed by my love of nerd references, cooking and my lovely wife. There is nothing more reassuring than cold hard data. Actually, before Bryna and I decided to get married, we made a spreadsheet plotting out key points in our lives to look for variance and possible divergence. It was surprisingly romantic.

But back to politics. I took a look at the results from the primary to see what I could extrapolate about the general election. I looked at geographic areas for places where Democrats or Republicans are losing or gaining ground. I looked at whether or not the Buys Bounce was still in effect, I looked at a number of factors, but in the end I kept coming back to one incontrovertible fact:
Not very many people voted. Really, it was a lousy turnout this year. So instead of looking at who the electorate voted FOR, I decided to examine just who voted. First, the totals over the last ten years:

Total Votes Cast

As with all the images on this site, you can enlarge them by clicking on them. Now the data I used is not perfect, I’m pulling from a variety of spreadsheets, so it does not perfectly line up with the Auditor’s totals, however it is internally consistent, so I rely on its findings.

As you can see, interest in the primary comes and goes. Whatcom County was very fired-up about their Tea Party candidates in 2010, and it showed. However, 2012’s turnout is nowhere near as strong. So who did vote? Let’s break it down by gender:

Gender of Electorate

Part of the reason 2010 was such a conservative year is that women, who generally vote more Democratic, did not turn in their ballots at the same rate as men. This year, however, it looks like we are returning to a gender gap similar to 2006 or 2008.

Another oft repeated point is that primary elections skew toward older voters. This is true, check out the breakdown over the last five primaries:

2002 Primary

2004 Primary

2006 Primary

2008 Primary

2010 Primary

2012 Primary

Age over Time

As you can see, the age ratio is staying pretty steady from one primary election to another. However, this is not the ratio of people voting in the general election. If you look at the general elections of 2006 and 2008, you find there is a different mix of people voting.

2006 General

2008 General

It becomes clear that the primary electorate is not the same universe of voters as the general. Okay, water is wet Riley, tell me something useful.

Even though the primary electorate looks vastly different than the general, if we find statistics on another primary that resembles the one in 2012, it should give us a clue as to what the general in 2012 will look like. Will it be the red wave of 2010, or the Obama block party of 2008? If we look for indicators in the primary data, we can match it up with the appropriate general election data.

So reviewing the charts I had already built, it looks to me like this election reflected somewhere between 2006 and 2008, with regard to gender, age breakdown, and total turnout. What about City vs. County?

Total votes in Bellingham vs the County

Again, this seems to just drive home how much of an anomaly 2010 was. The county turnout shot up that year, topping even the 2008 electoral turnout. However, in 2012 it seems to have returned to 2008 levels, if not a little below.

What about partisan identification, can we look at that? Yes, with a huge caveat. To get these stats, I am digging through the Whatcom Democrats database, and the Whatcom Dems have been very good about identifying Democrats, but not as successful ID-ing Republicans. Consequently, when I pull the numbers, it looks like Dems have a HUGE advantage over the Republicans, but in actuality, we just have better data on the Dems. So, please look at the differences from year to year, not necessarily the totals.

Voter ID over time

Would you look at that? Democrats and Republicans are voting at levels a little below 2008, but a little more than 2006.

So what does that tell us? Using the assumption that the primary reflects the general, we can predict, based on this primary, that our general election turnout will be something similar to 2006 or 2008. What does that mean? It means if I was Rep. Jason Overstreet, or Rep. Vincent Buys I would be looking over my shoulder.

I ran the numbers on the new 42nd legislative district with the vote totals of 2006. Now, 2006 was a pretty unique year because tons of energy and effort was poured into the 42nd races, with Brandland and Ericksen working desperately to hold onto their seats. That said, if you averaged out Kelli Linville, Jasper MacSlarrow and Jesse Salomon (the three Democratic candidates that year) against the averaged totals of Dale Brandland, Doug Ericksen and Craig Mayberry, the Democrats win in the NEW 42nd district 51.4% to 48.6%.

In 2008, it was just Linville and Ericksen running for reelection with token challengers on either side (sorry Mark!). When you run the averages there, you find Republicans scraping by with 50.7% and the Dems coming in with 49.2%, a difference of around 1,000 votes.

We have already determined that 2012 will look like a mix between 2006 and 2008, which means I bet the 42nd race is going to come down to the wire. Throw in a couple ballot measures boosting Democratic turnout and you have the makings for a very interesting election night.

What patterns do you guys see in the numbers? Predictions? Analysis? The floor is yours, loyal readers.


Ask and ye shall receive. In the comments, the extremely knowledgeable Jim Fox and Bill McCallum both asked about turnout as a percentage of total registered voters. They contend that while the turnout was low, it was not unusually so. I also got an email from Whatcom Democrats supervolunteer, and coiner of the “many hats” theory Steve Schuck making very much the same point.

So here it is, voter turnout as a percentage.

Voter turnout as a percentage of total registered

I kept the zero data point in there just to show the context that most primary returns are within a narrow range of 32-46%. Bill and Jim’s point holds true, while the turnout percentage is low, there have been other years where the turnout has been low, if not lower. Here is the same graph without the contextual point, so you can see it a bit closer.

Percentage Voter Turnout zoomed in

This seems to buttress my point that the best comparable model for turnout in 2012 is the election of 2006, where we had a similar level of returns. What do you guys think?



  1. I think we’ll see a pretty big surge in voters this year. Nobody that isn’t a political nerd was interested in the primary. EVERYBODY will be ready to vote in November. Although I don’t think very many people in our part of the world are excited for Romney, adding Paul Ryan made things very interesting. It really got people talking.

  2. I personally voted in the primary this year, but I confess, like a lot of people, I don’t always vote in the primaries. However, I know a lot of people who definitely won’t miss the general election, especially with the referenda that are on the ballot this year — same-sex marriage and marijuana legalization — which will bring out more people. Lots of people I know are voting for Obama, not because they love him, but because they’re afraid of Romney.

    Incidentally, do you think the Libertarian lawsuit to get Romney off the ballot will succeed?

  3. You are amazing!

  4. The 2012 primary turnout was not lousy. It was slightly below average. Not all the ballots are counted for the primary but the number of ballots returned was 40.3 percent. I have the primary turnout numbers for for the last 10 presidential years in Whatcom County. The average turnout for 1972 through 2008 was 41.3 percent. The lowest turnout year was 1996 at 33.9 percent and the highest year was 1972 at 46.9 percent.

    People getting ballots delivered to their home does not appear to encourage turnout. In the 1972 primary, 6 percent voted by mail.

    • I think this year is lousy in compared to the last couple of years. I appreciate the historical perspective but I only looked at the last five primaries because there are so many other factors involved, mail-in voting being the most obvious.

      Since you are not the only one who has asked about this, I will be adding some charts with percentage of voters shortly.

  5. Buys and Overstreet win, Mckenna and Dunne win, Romney/Ryan wins.

    • I think you mean Reagan Dunn, who got a measly 30 something percent in the primary, right? You can make a case for Buys, Overstreet, McKenna and even Romney/Ryan but Dunn? I have a hard time seeing how he makes it this year.

  6. The ol’ perfesser here (50 years and counting in the social and cognitive sciences) thinks you’ve done a very nice job. With all the appropriate caveats noted, it does look like 2010 was an anomaly. If so, we should see a regression to the mean. Perhaps — one can only hope. Get your checkbooks my friends, just to make sure.

  7. I was going to make Bill McCallum’s point: Our turnout data put’s this year’s county turn-out (heading towards 40.3%) very close to average of Registered Voters (1972-2008).
    See: County vs State
    Max 46.9% 48.0%
    Avg 41.3% 43.4%
    Min 33.9% 40.8%

    However returns definitely felt slow and delayed, with more than usual confusion, corrections and write-ins for major races.

    I caution at reading Received Returns dates too closely; as there are built-in process load-leveling backlogs. “Received” is when Returns are actually signature-checked and logged into the process-stream — not when Returns are dropped into ballot or mail-boxes, or enter the courthouse building. (Election Day “Returns” are not counted “Received” until Wednesday). There may also be a day or more lag between Received and Counting.

    Note: Sec of State county summary uses “Turnout% = Counted/Reg-Voters” rather than Received, or Total Voting-Age Population.

    I’m estimating, of final cast results, 65% were Received and 50% Counted by Election Evening (8/7), with 99% Received and 92% Counted by Friday (8/10).

    I’d try to get rid of your initial (zero, 2000) data-points.

    Here are some national stats: National Voter Turnout in Federal Elections: 1960–2010 —

    Thanks for some excellent and interesting work.

    • I’m making a bonus chart with the turnout percentages since there seems to be a great deal of interest in this. I added the zero data point because otherwise the charts lack contrast, the scale starts somewhere near the lowest point and visually you don’t “see” how close some of these data point are to each other. That’s why I included it.

  8. […] Election Analysis: What the Primary tells us … – The Political Junkie Go to this article […]

  9. […] initiatives, campaigns and millions in advertising dollars, I figured I would take a break from the insanity of electoral […]

  10. […] There, I have thoroughly doomed my electoral future. I can see the attack ads now. Anyway, I often build charts of data, or thoroughly mock terrible charts from other […]

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s


%d bloggers like this: