An advantage that political science has over journalism is the ability to revisit an issue as new data comes in. Under the relentless pressure of daily deadlines, journalists rarely if ever have an opportunity to come back to a story unless circumstances are such that it becomes newsworthy again. Political scientists, in contrast, can revisit an issue endlessly until they come to some common understanding regarding its main features.
A case in point is the recent off-year elections for governor in New Jersey and Virginia. For the most part, journalists are already moving on as new stories – the health care vote, the Ft. Hood attacks – take the headlines. But political scientists are still crunching the data, trying to determine the causes and significance of the Republicans victories. Toward that end, Charles Franklin published a column yesterday suggesting that rather than diminished turnout, what primarily drove the election results was a shift in voter preferences away from supporting Democrats and toward supporting Republican candidates.
As Franklin wrote, “The shifts in outcomes between the 2008 presidential and 2009 gubernatorial elections in New Jersey and Virginia were driven far more by shifts in voting preferences among groups than by changes in turnout across those groups. Only age groups show consistently substantial changes in relative share of the electorate. Vote preference, in comparison, shows quite large shifts between election years. While one narrative of the 2009 election was changing turnout motivation, this turns out to be substantially false. Instead, changes in candidate preference drove the Republican wins in both New Jersey and Virginia.”
In my posts leading up to and after the election, I made a slightly different claim: without dismissing the impact of changing preferences, I argued that the results were largely due to lower turnout, particularly among the president’s party supporters, that is characteristic of these off year elections in both states going back almost 40 years. Franklin, however, downplays more than I the impact of lower turnout for a very simple reason: exit polls in both states indicate the proportion of particular groups who voted last Tuesday did not differ enough from their proportions in the 2008 presidential elections in both states to explain the different results. Using exit polls from both elections, Franklin looks at nine different ways to categorize voters, including partisan affiliation, race, gender, and age. His results are captured in this chart.
Note the arrows – the longer the arrow, the bigger the shift in that group’s preferences from 2008 to 2009. (A purple color indicates that the group as a whole shifted support from one party to the other.) Although the details are hard to see, what it shows in a nutshell is a significant shift in voting preferences away from the Democratic candidate in key voting blocs, including 18-29 year olds, those without a college degree, independents, rural voters and males. The most significant shift, however, is among Independents. In Virginia, 49% of independents supported Obama in 2008, but that dropped to 33% who voted Democrat in the gubernatorial election. In contrast, the preferences of neither Republican nor Democrat partisans shifted as much across the two elections. The story is similar in New Jersey, as the following table indicates.
Here 51% of independents supported Obama in 2008, but only 30% supported Corzine (the Democratic candidate) in 2009. Support among Democrats for the two candidates in 2008 and 2009, in contrast, didn’t change as much, while “Republicans came home to their party a bit, from a 14 percent defection rate for Obama to just 8 percent defection to Corzine.” And, as in Virginia, Franklin found “substantial movement” in voter preference among other key groups, including 30-44 year olds, moderates, whites, Hispanics and males.
How do I respond to Franklin’s argument? He is a first rate scholar who has presented an interesting (and provocative) argument. In the end I stand by my initial claim: that the results reflect a combination of lower turnout and a change in voter preferences linked to local issues and candidates in both states, rather than any referendum on Obama. As Franklin acknowledges, turnout in the two gubernatorial races decreased by almost half from what it was in 2008. As the following two tables show (source here), this is not unusual; turnout is always lower in the off-year elections than it is in presidential years in both states.
Franklin argues, however, that the lower turnout is not the primary explanation for the shift in voter preference from Democrat to Republican in 2008 to 2009. But I think it explains some of it. Note that exit polls show that there is a decline in the share of the electorate who are Democrats, and an increase in the proportion who are Republican. Moreover, the exit poll data show that there is a shift among the various age groups as well. In Virginia, the 18-29 year olds dropped 11 points, from 21 to 10 percent of the electorate, from 2008 to 2009. Similarly, the 30-44 age group also declined, from 30 to 24 percent. In contrast, the proportion of 45-64 year olds and those aged 65 and up increased. A similar shift by age occurs in New Jersey, with 18-29 and 30-44 year olds showing a decline as a proportion of the electorate, while those aged 45 and above took an increasing share in 2009 over 2008. So at least some of the results reflect a shift in the voter pool. So I remain convinced that different levels of turnout within key subgroups explain some of the shift.
Franklin suggests, however, that these shifts are not the biggest factor in the different results from 2008 to 2009. Instead, he notes that in both states there was a significant shift in support, from Democratic to Republican, among Independents. In thinking about his argument, I do worry about the “squishiness” of the independent category. Because the relative proportion of the electorate who calls themselves “independent” did not change much from 2008 to 2009, but their support for Democrats declined significantly, Franklin argues that preference changes primarily drives the results. But we know from other studies that independents include a substantial number of “leaners” who typically prefer one party to the other. My worry is that the depressed turnout in 2009 reflects the absence of a substantial number of “leaners” who are predisposed to vote Democratic. That is, the pool of independents is weighted more heavily to Republican leaners in 2009 compared to 2008.
Nonetheless, in the end I don’t necessarily disagree with Franklin’s argument that a shift in voter preferences explains a good deal of the decline in support of Democrats across the two elections. Indeed, I argued as much in my earlier posts – but I suggested the shift in preferences was not because of changing opinions toward Obama, but because local issues overrode national concerns in these off-year elections. And in fairness to Franklin, nowhere does he suggest that the shift in preferences reflects changing attitudes toward Obama – he is only suggesting that voters, particularly independents, were less likely to vote Democrat in 2009 than they were in 2008. That shift in preferences might be entirely due to local factors, including issues and the candidates on the ballot.
In short, I remained convinced, in the absence of additional evidence, that the two off-year elections in New Jersey and Virginia cannot be viewed as a referendum on Obama or his policies. I don’t, however, disagree with Franklin’s argument that the results reflect both turnout and a change in voter preferences, although we might disagree on the relative weight to place on both factors.
I hope this gives you a better sense of how political science works. Believe it or not, we do try to – or at least I try to – base our arguments on the available evidence, rather than on our own political predispositions. When additional data comes out that suggests I am wrong, I will be the first to post it. I hope that’s why you come to this blog.