Generally I prefer not to blog about my academic work, partly because I'm convinced that this site is going to get me fired someday but largely because what I do isn't very interesting to…humans, I think, unless they are political scientists. But in the process of collecting data for a new project, I came across some interesting stats on 2010.

One thing that I find amazing about the literature in political science is how little we know about non-voters. We have a decent-plus understanding of voters' preferences, ideological tendencies, and so on, and the same cannot be said about the substantial portion of the potential electorate that doesn't participate. Things like exit polls, for example, are integral to post-election analysis (both popular and academic) and they exclude non-voters by definition. There is some new interesting work on the voter/non-voter divide (Diana Mutz has a book coming out soon) but for decades we've been assuming, or arguing with varying degrees of success, that the preferences of the two groups are essentially the same. I have always found this patently stupid.

Lately I've been doing a lot of research into creative ways to show evidence of differential turnout – the idea that overall turnout doesn't change much from election to election but different groups of people are voting each time. For example, Obama Mania and the lameness of the McCain campaign made a lot of Republicans stay home in 2008, whereas in 2004 Republicans were highly energized while the Democrats had a hard time rallying turnout behind John Kerry. I've always considered this the most plausible explanation for changes in election outcomes over time. And in theory this narrative explains 2010 very well; Democratic voters got all fired up and turned out in droves in 2008. Then two years of Obama being a big ol' letdown to the left and a hideous monster to the right led us to 2010, when Democratic-leaning voters were apathetic and the GOP/Tea Party brimmed with enthusiasm. Makes sense, but does it match the data?

The Cooperative Congressional Election Study (CCES) is a new, very large sample survey that has been a real boon to voting behavior research. In 2010 the sample was over 55,000 whereas the old stand-by (American National Election Study) has about 1200. Briefly setting aside the valid question about whether people report voter turnout honestly on surveys – hint: they don't – here is a breakdown of self-reported ideology among voters and non-voters in the 2010 CCES sample. Click any graph to embiggen:

It is hardly surprising that a much larger percentage of non-voters consider themselves "middle of the road", as apathy toward participation is often a by-product of apathy toward politics in general. But note how a considerably larger portion of voters consider themselves "conservative" or "very conservative" – more than double the share of non-voters in each category. Compare that to 2006 – when the ideology question had only five categories instead of seven, but you get the picture:

In the last midterm the ideological distribution of voters and non-voters is quite similar, with voters only slightly more conservative than non-voters. Even more interesting, here is the ideological placement of Barack Obama among voters and non-voters. In other words, survey participants are also asked to identify how they would describe the ideology of major political figures like the President, their reps in Congress, and so on:

Wow. I checked these number about five times because they are so lopsided. Of people who reported voting in 2010, more than half described Obama as "very liberal" whereas barely 25% of non-voters described him that way. The question is subjective, of course, but that's the point. The people who voted in 2010 appear to be disproportionately drawn from the ranks of people who think Barack "The Eisenhower Republican" Obama is the commie-libro-marxisocialist child of Fidel Castro and ACORN. Unfortunately the same question is not available from 2006 re: George W. Bush. They did ask a generic presidential approval rating question, though:

Again we see a roughly similar distribution – voters exceeded non-voters at the extreme points, i.e. they were more likely than non-voters to either really like or really dislike Bush. What we don't see is the heavily skewed distribution we see in the 2010 Obama Ideology question (although it bears re-emphasizing the differences in the two questions).

This small slice of the CCES data from 2010 supports the hypothesis that voters and non-voters were ideologically dissimilar and had very different perceptions of the incumbent President and Congress. It suggests that finding a way to recreate the enthusiasm for participation that Democratic-minded people showed in 2008 – or even 2006, when voters and non-voters were essentially even – is a necessary component of Obama and his party avoiding disaster in 2012. Something tells me that slogans and slick marketing won't be enough this time.