The Iowa caucuses became the leadoff presidential nominating event essentially by accident in 1972, and they have been institutionalized to the point where it is hard to imagine the nominating race starting anywhere else. Once again in 2012 the voting for president kicked off in Iowa. But unlike 2008, when the contests for both political parties were wide open, only the Republicans had a contested nomination. Democratic President
Barack Obama, who was running unopposed for a second term, won more than 98% of votes from those who attended the Democratic caucuses.' Although the outcome for the Democrats was clear, a great deal of uncertainty surrounded the Republican results, when on caucus night Mitt Romney was declared the winner by eight votes over Rick Santorum, only to have that overturned two weeks later when the party announced that Santorum had won the Iowa caucuses after all (Fahrenthold and Wilgoren 2012). Overall, around 121,000 Republicans were reported to have attended, equal to about 19% of registered Republicans in Iowa (Iowa GOP 2012). This turnout was up only slightly from four years prior.
It is not our purpose here to get into the weeds of what actually happened in the 2012 Iowa Republican Caucus. Suffice it to say that the challenges of counting the ballots--which are cast in over 1,700 precincts generally on paper and counted by local caucus leaders--provided new fodder for those who see Iowa's role as illegitimate. After all, given the intense media focus on Iowa and the stakes that are perceived to exist, the narrative about Iowa may well have been different had Santorum been declared the winner on caucus night instead of Romney. As we show elsewhere (Redlawsk, Tolbert, and Donovan 2011), media attention to the candidates shifts based in part on how well they do in the caucuses versus expectations set by the very same media. This shift then affects voters in states with later nominating events (primaries and caucuses). Thus, although winning Iowa may not be a direct path to the nomination, failing to meet or beat media expectations is potentially a path to failure (Donovan and Hunsaker 2009). In this article we update our (Redlawsk, Tolbert, and Donovan 2011) analysis of the impact of Iowa and New Hampshire on the rest of the nomination process, using new data from the 2012 nomination campaign. We find that these new data, if anything, strengthen the argument that Iowa matters. Whether this is a good thing or not, we leave to another time.
In this article we build on two aspects of our earlier work. First, we examine Iowa voters using data from a University of Iowa Hawkeye Poll which was conducted in late November and early December 2011. We use these data to compare people who said they would attend the Republican caucus in 2012 to Iowa Republicans who said they would not. Second, we model results in Iowa to predict the 2012 contest so we may assess who did better or worse than would be expected based on early media expectations about the candidates. We also update our analysis of the effect of changes in media attention to candidates after Iowa and New Hampshire, and test how this predicts the outcome of the lengthy nomination process. We find evidence that Iowa and New Hampshire continued to play influential roles after controlling for other factors, and even with the uncertainty that was rampant in the 2012 Republican nomination process. (1)
Iowa Caucus Goers, 2012
Among the many critiques of the Iowa caucuses has been the complaint that those who show up are not even representative of voters within Iowa, never mind the (straw person argument) that they do not represent the country as a whole. (2) The conventional wisdom about caucus goers was for many years best encompassed in this paragraph from Stone, Abramowitz, and Rapoport (1989):
Caucus attenders and other nomination activists are not typical of the larger populations they may be presumed to "act for." They are better educated, older and have higher incomes. They are more committed to and active in their party than the average citizen. They are less likely to be indifferent on the issues and are more "extreme" in their ideological commitments. (44) But as detailed in Redlawsk, Tolbert, and Donovan (2011), the 2008 caucuses on both sides seemed to have challenged this wisdom. As we showed through a series of surveys of likely and actual caucus goers, by caucus night 2008 those attending their caucuses were quite similar to the rest of their partisan cohort. On a range of demographic measures--partisan strength, education, income, age, and marital status--actual 2008 caucus goers tracked very closely to a broader Iowa registered voter sample (Redlawsk, Tolbert, and Donovan 2011, tbl. 6.2), in part because of high levels of mobilization by the Obama campaign. The one demographic difference we found then was religious preference, especially among Republicans, where evangelicals were more likely to caucus than those who did not characterize themselves in this way. Ultimately the intensity of the 2008 caucus, which drove attendance to record highs for both parties, led to caucuses that were broadly representative of the voters who identify with the two parties in Iowa. (3)
So, to what extent was 2008 sui generis? After all, there were two extremely intensive campaigns that blanketed Iowa for more than two years before the 2008 caucuses. Perhaps the level of intensity was such that even Iowans who were normally marginal caucus goers were mobilized. There is some evidence of this, since more than half of 2008 caucus goers were first timers (Redlawsk, Tolbert, and Donovan 2011). It is certainly possible that 2012 was different, with only one contest and somewhat less intensity as a result. Indeed, CNN entrance polls show that 60% of 2012 Republican caucus goers said they had caucused before. (4)
We investigate the 2012 Iowa Republican contest with Hawkeye Poll data collected in late November/early December 2011 (Boehmke 2011). This telephone survey identified likely caucus goers for both parties, though given the uncontested nature of the Democratic caucus, there were relatively few of them in the sample so they are not discussed here. To be classified as a likely caucus goer, the individual had to report being very or somewhat likely to caucus and to be able to name which political party they would caucus for. Overall, 277 respondents said they were likely to caucus with the Republicans. This included 184 who self-identified in the survey as Republican, plus an additional 93 who identified as independent or who revealed no party preference. The same survey included another 199 self-identified Republicans who said they were not likely to caucus. This allows us to compare three groups--Republican identifiers unlikely to attend, Republicans likely to attend, and all voters (Republicans and others) who were likely to attend the GOP caucuses--to get some sense of whether caucus goers in 2012 were again broadly representative of the party.
A demographic comparison of Republicans likely to caucus and not caucus is presented in Table 1. A quick look at Table 1 suggests that in many ways, the basic demographics of those who were likely caucus goers in 2012 were not dramatically different from the sample of Republicans who said they would not caucus. Across gender, race/ethnicity, marital status, and church attendance, there are only small differences between 2012 GOP caucus goers and self-identified Republicans who were not likely to caucus. This is not, however, to suggest they are exactly the same. Likely caucus goers were more middle aged than Republican nonattenders, with both the oldest and youngest cohorts slightly underrepresented. But the difference is not substantively meaningful, and it is quite similar to the difference we report from 2008 (Redlawsk, Tolbert, and Donovan 2011).
As they were in 2008, likely caucus goers in 2012 were slightly more male, but in 2012 they were similar to nonattending Republicans in ideology. Unlike 2008, Republican caucus goers in 2012 and Republican identifiers not caucusing were nearly identical in religious denomination. The 2012 GOP caucus goers were slightly wealthier and more highly educated than Republicans who were not likely to attend. These differences are not unique to the Iowa caucus; they match demographic patterns observed in studies of political participation (Brady, Verba, and Scholzman 1995). Table 1 also illustrates that Republican caucus goers were much more likely than other Republicans to report attending church at least once a week and to call themselves strong conservatives, but much of these differences dissipate in the sample of all caucus goers (which includes self-identified Republicans likely to attend, plus independents).
As with 2008, it appears that relatively high turnout in a caucus--and 2012 was another record for Republican turnout--does move the aggregate demographics of the caucuses toward the general population of party identifiers who do not caucus. Thus, what we found in 2008 does not appear to be a one-off fluke. One difference from 1988, when Stone, Abramowitz, and Rapoport (1989) were writing the paragraph quoted above, is that turnout in 2008 and 2012 (Republican) was substantially higher than in the prior record turnout year of 1988. That year, about 120,000 Democrats and 80,000 Republicans caucused. In 2008, twice as many Democrats and 50% more Republicans attended, with the 2012 Republican turnout increasing slightly more. The point is, at a certain...