Comparing Caucus and Registered Voter Support for the 2008 Presidential Candidates in Iowa David Redlawsk University of Iowa

Department of Political Science Publications

1-1-2008 Comparing Caucus and Registered Voter Support for the 2008 Presidential Candidates in Iowa David Redlawsk University of Iowa

Daniel Bowen

Caroline Tolbert

DOI: https://doi.org/10.1017/S1049096508080219

Copyright © 2008 American Political Science Association. Used by permission. http://journals.cambridge.org/action/ displayJournal?jid=PSC PS: Political Science & Politics, 41:1 (2008) pp. 129-138. https://doi.org/10.1017/S1049096508080219

Hosted by Iowa Research Online. For more information please contact: [email protected]. Comparing Caucus and Registered Voter Support for the 2008 Presidential Candidates in Iowa

David Redlawsk, University of Iowa Daniel Bowen, University of Iowa Caroline Tolbert, University of Iowa

s the 2008 presidential nominating the first nominating events are now the 2007, with a margin of error of ϩ0Ϫ3.0 A process got underway, Iowa’s coveted starter’s gun in a 50-meter dash rather percentage points for the full sample.2 status as first-in-the-nation appeared in- than a mile run, who gets off the starting The second survey used a split sample creasingly in jeopardy, as states engaged blocks first may well matter even more. design of 907 registered statewide regis- in aggressive frontloading throughout As Hull ~2007, 66! argues, Iowa’s impact tered voters and a sample of 787 poten- 2006 and 2007. In the past, late March on New Hampshire and the national nomi- tial caucus attendees, 555 of which were primaries in large states like Florida, New nation process is a “wild, wired one.” In deemed likely to attend. The survey was York, and California were irrelevant to the this rapid sea of a changing nomination in the field from July 29 through August electoral outcome. To avoid a repeat in process we take a close look at the Iowa 5, 2007.3 Results for statewide registered 2008, Florida moved its primary to Janu- electorate, both statewide registered vot- voter samples were weighted to match ary 29 and California moved to what is ers and a subset of likely caucus attend- gender to statewide levels using the Cur- now being called “super duper Tuesday” ees, to shed light on the underpinnings of rent Population Surveys. Because gender on February 5 when nearly two dozen support for the presidential candidates in distributions are unknown in the cau- states will hold primaries. Under pressure the early stages of the 2008 campaign, cuses, caucus samples were unweighted. from extra-early voting in Florida and using unique rolling cross-sectional data Student callers conducted both telephone other front-loading states, as we write this to track opinion change over time. surveys, averaging about 14 minutes the Iowa caucuses are to be held on Janu- Because Iowa has been a swing state each, with a completion rate of approxi- ary 3, two days after New Year’s. It seems in recent presidential elections, statewide mately 24% of contacts.4 possible that as a result of the nominating attitudes towards presidential candidates season becoming more condensed, there should be relatively consistent with na- may be an increase in the importance of tional opinion. However Iowa caucus Iowa and New Hampshire, the opposite of attendees represent only a small subset Iowa Registered what the states moving earlier wanted. If of registered voters—generally 10–15%; Voters—March and 20% at best.1 Scholars have argued that July/August Polls primaries and caucuses create ideologi- We begin by examining the results David P. Redlawsk is associate profes- cally polarized party nominees because among our overall registered voter sam- sor of political science at the University of the composition of the electorate ex- ples in March and August, and then turn Iowa. He is author of How Voters Decide: cludes independents and moderates and to likely caucus goers specifically. While Information Processing in Election Cam- varies demographically from general most of the country in March 2007 was paigns (with Richard R. Lau) and editor of election voters ~Fiorina, Abrams, and giving little thought to the 2008 general Feeling Politics: Emotion in Political Informa- Pope 2004; Donovan and Bowler 2004; election, in Iowa it was already a hot tion Processing. He is also past chair of the but also see Norrander 1989!. We draw Johnson County (Iowa) Democratic Party, topic with less than a year until the lead- on two unique statewide University of off Iowa caucuses. Registered voters, where he oversaw 57 precinct caucuses in Iowa Hawkeye Poll surveys conducted in 2004. whether likely caucus goers or not, were late March 2007 and late July through asked who they would support in No- Daniel C. Bowen is a graduate student early August 2007, including representa- vember 2008. As seen in Table 1, at that in political science at the University of Iowa. tive samples of Democratic and Republi- point all three of the top Democrats ~Hil- His research interests are American state can likely caucus attendees. We examine lary Clinton, John Edwards, and Barack and local political institutions, public opin- how the population of caucus attendees Obama, in that order! led all Republi- ion, and representation. differs from statewide registered voters, cans. Following Obama were Rudy and whether the Iowa caucuses are repre- Caroline J. Tolbert is associate profes- Guliani, John McCain, and Mitt Romney. sentative of the state as a whole as we No other candidate received more than sor of political science at the University of examine the demographic and issue Iowa. She is author of Educated by Initia- 2%. It is not that no other candidate was tive: The Effects of Direct Democracy on Cit- bases of candidate support in Iowa. running an active campaign—Republican izens and Political Organizations in the candidates made 33 visits to Iowa in the American State (with Daniel Smith) and University of Iowa first quarter of 2007; only seven of Digital Citizenship: The Internet, Society and Hawkeye Poll which were by Romney and McCain Participation (with Karen Mossberger and combined and with no visits by Giuliani. Ramona McNeal). She is editor of Democ- Our data are from two statewide ran- On the Democrats’ side there had been racy in the States: Experiments in Election Reform (with Bruce Cain and Todd Dono- dom sample surveys collected through 18 visits, eight of which had been by van). Her published work focuses on voting, the Social Science Research Center at non-top tier candidates ~Democracy in elections, and representation in the United the University of Iowa. The first survey Action 2007!. But voters, perhaps led by States. She can be reached at: caroline- was of 1,290 registered voters and likely the media, focused only on the top three [email protected]. caucus attendees between March 19–31, of each party.

PSOnline www.apsanet.org DOI: 10.1017/S1049096508080219 129 August sample said they were Table 1 not satisfied with their choices, “If the 2008 presidential election were held today, who would you an increase from March. This is vote for president? Just tell me the name.” [Open Ended quite consistent with the idea Question—All Names Given by Respondent, Percents] that this relatively small group of pure independents may be Partisans Including Party Leaners unsatisfied in general with the All Respondents Democratic Republican two-party system ~Donovan, Parry, and Bowler 2005; Dono- March August March August March August van and Bowler 2004!. N = 1290 N = 873 N = 600 N = 433 N = 479 N = 366 Don’t Know 35.28 31.09 27.56 22.75 37.72 36.25 Candidate Support in Clinton (D-NY) 12.88 18.14 24.42 29.93 .95 4.13 the Iowa Caucus Edwards (D-NC) 12.41 9.07 23.08 16.31 1.20 1.90 Gore (D-TN) .69 1.42 Of course only caucus attend- Obama (D-IL) 10.75 14.08 16.56 20.67 3.62 7.07 ees get to choose nominees in Richardson (D-NM) .96 3.56 1.56 5.57 .51 1.94 Iowa. Measuring opinions of Brownback (R-KS) .55 .45 1.37 1.06 this small subset of voters is an Gingrich (R-GA) 1.65 .48 3.92 1.13 art as well as a science, and Giuliani (R-NY) 7.48 4.15 .42 17.65 9.27 commercial polls use a variety Huckabee (R-AR) .34 .91 .86 2.16 of techniques to find them. We Hunter (R-CA) .07 used three screening questions: McCain (R-AZ) 7.32 1.03 1.21 15.29 2.20 whether the respondents had Romney (R-MA) 5.19 9.17 1.00 .44 10.41 20.41 participated in the 2000 or 2004 Tancredo (R-CO) .99 2.35 caucuses; whether they said they Thompson, F. (R-TN) 2.29 .25 5.14 were likely or very likely to Thompson, T. (R-WI) .86 .83 1.97 2.97 attend the 2008 caucuses; and Other 3.57 2.52 2.76 2.15 4.53 3.00 whether they could name the 6 100.00 100.00 100.00 100.00 100.00 100.00 party where they would caucus. Individuals who reported they were very or somewhat likely to attend but could not name the party were removed from the By August, Iowa voter preferences four in each party ever got much press likely caucus sample since one chooses a shifted significantly. Clinton’s support or voter support throughout 2007, limit- party to actually caucus. increased over 5 percentage points in ing voters’ awareness of alternatives, Caucus attendees are believed to be March to more than 18%. Following even though many voters were looking. more partisan than non-attendees— Clinton was Obama ~14%!, who passed In particular, the top Republicans did not disproportionately coming from the Edwards, while Romney and Edwards seem to be a good fit for a conservative strongest party members. Our results in were virtually tied with just over 9% of Iowa Republican party. Table 2 tabulates Table 3 confirm this. While strong Dem- the respondents supporting each candi- satisfaction with the then current choice ocrats make up 24% of our overall March date. Following them came Giuliani at of candidates by respondent partisan- sample, they make up more than 36% of only 4%, Richardson with 3.5%, and ship.5 It is immediately clear that Repub- likely caucus attendees. For Republicans, Fred Thompson with just over 2%. Since licans were generally much less satisfied the difference is not so stark—strong Re- March, support for Republican candi- with their choices than Democrats. While publicans are 19% of the total March dates had shifted substantially with Mitt there was improvement for Republicans sample, and 21% of all likely caucus at- Romney the main beneficiary. Both from March to August, even many strong tendees. But in both cases, strong parti- Giuliani and McCain fell from March to Republicans remained unhappy with their sans are overrepresented, while all other August, with McCain’s support all but options. On the other hand, the data sug- groups ~except weak Democrats! are un- disappearing, as is evident in Table 1. gests that the most partisan of Democrats derrepresented. We see the same in Au- The Republican results were also af- were comfortable that they could find a gust, with both strong Democrats and fected by talk surrounding Fred candidate from among those already de- strong Republicans accounting for a Thompson’s potential entrance into the clared, while those less partisan may larger percentage of likely caucus attend- campaign. The August data show the line have been more open to alternatives. All ees than their percentage of registered between the early top tier candidates and Republicans, on the other hand, appeared voters. Embedded in these data is more the rest of the field was blurring. For the open to other options. support for the idea that Republicans Democrats, Bill Richardson made gains Perhaps the most interesting point of were less content with their options—and to become a fourth major candidate as comparison between March and August the political environment—than were Edwards’s support declined. On the Re- levels of satisfaction is among pure inde- Democrats. Overall, Republicans of all publican side, March “also ran” candi- pendents. While we focus on the nomina- stripes represent 42% of the overall dates Mike Huckabee and Tom Tancredo, tion process, it is interesting to examine March sample, but only 37% of all likely along with potential candidate Fred those in our sample who will probably caucus attendees. Democrats, on the other Thompson, all received greater support have no say in the nomination for either hand, are 42% of the overall sample, but among Republicans than did McCain. party. Pure independents were even more 62% of all caucus attendees. The typical While there were some 18 declared likely than Republicans to show a lack of Democrat was far more likely to claim he candidates across the two parties at vari- satisfaction with the available choices. or she would attend the caucus than the ous points in time, only the top three or Over half of pure independents in the typical registered Republican voter.

130 PS January 2008 However, within both parties’ individ- completion of the August survey, Repub- ual caucuses, strong partisans dominate in licans held their Straw Poll in Ames, roughly equal percentages. Of all Repub- which Romney won, but Mike Huckabee lican likely caucus attendees in the March came in a surprisingly close second. Yet, data, 57% are strong Republicans; for the Huckabee had not even registered 2% Democrats the share is virtually the same, support in the August survey. This points 57%. Results from the August data show out one of the severe difficulties of poll- strong Democrats as 62% of the likely ing the Iowa caucuses; so much depends Democratic caucus attendee sample, an on hidden “on the ground” efforts of the increase over the March results, while campaign to get believers out to events strong Republicans make up 56% of the like the Straw Poll and the caucus itself.7 likely Republican caucus attendee sam- Perhaps counter-intuitively, the per- ple, a slight decrease from March. centage of likely Democratic and Repub- lican caucus attendees who did not know The Bases of Candidate which candidate they would support in- Caucus Support creased from March to August. These numbers, however, were still much lower While registered voters preferred Clin- than the percentage of registered voters ton in March, the story was different for who did not know who to support. Two likely caucus attendees, for whom Clinton things are clear. First and not surpris- was not the Democratic leader, and ingly, caucus attendees are more likely to Giuliani shared his support with McCain, have an early candidate preference than as is clear in Table 4. In March, Edwards are those who are less likely to attend. led among likely Democratic caucus goers Second, those who did not have a general even though Clinton led among all regis- election preference but who were likely tered Democrats, and Romney led among caucus attendees primarily supported Ed- Republicans. There was a great deal of wards ~for Democrats! or Romney ~for volatility between March and August. For Republicans!. When those voters were the Democrats, while Edwards held a asked to focus on the more immediate nearly seven-point lead over Clinton in caucus rather than the November 2008 March, with Obama seven points behind general election, they did have a prefer- Clinton, by August he had declined as a ence, and that preference shook up the frontrunner, falling from 35% to 26.7%, horserace, especially for the Democrats.8 leaving him nearly tied with Clinton. Clinton also showed a small drop in support, from 29% to 25%. The beneficiary Comparing Registered of these declines seemed to be Richard- Voters and Caucus son, whose support surged by more than Attendees 8.5 points. Caucus attendees are clearly a Survey marginals tell us that at a par- different electorate, at least in their candi- ticular time there was a particular distri- date preferences—while Clinton’s support bution of support for candidates—but among registered Democrats increased that by itself is not the whole story, since from March to August by nearly 6%, her campaigns are dynamic events that carry caucus support slipped by 4%. More sub- the possibility of change all the way stantially, Edwards was polling 10% bet- through to Election Day. However, our ter among likely caucus attendees than Iowa surveys included questions that among registered Democrats in August.

Democrats Independents Republicans allow us to go much further. First, we On the Republican side, McCain and can examine the underlying bases of sup- Giuliani were nearly tied in March with port for candidates in terms of strength Romney third among likely caucus at- of partisanship, which should provide tendees. However, while “don’t knows” some sense of each candidate’s core sup- received more support from registered porters. Second, because we examined voters than any frontrunner ~38%!, cau- samples of likely caucus attendees as cus attendees were more likely to have well as registered voters, we can draw a preference ~only 24% were “don’t Strong Weak Leaning Leaningcomparisons between Weak likely caucus Strong at- knows”!. The August data show how tendees and registered voters as a whole, things were changing for Republicans. March August March August March August March August March August March August March August letting us consider how caucus voters Both Giuliani and McCain’s support may differ from other voters in Iowa. shrank considerably from March, with Giuliani falling by 10 percentage points and McCain all but collapsing to just Partisan Strength and General over 3%. Romney emerged as the clear Election Presidential Preference frontrunner with 28%. Fred Thompson, Sam Brownback, and Tom Tancredo Table 5 examines our general election were all beneficiaries of Giuliani and preference question using the March 2007 McCain’s drop in support, though by data by the strength of partisanship of Table 2 Satisfaction with Current“Given Choices the of people Candidatessatisfied, who by somewhat now Respondent’s satisfied, say Partisanship or they not are satisfied?” running for president in 2008, how satisfied are you with your choices. Are you very Very SatisfiedSomewhat SatisfiedNot SatisfiedTotal 56.18 37.83 49.76 43.85 5.99 61.88 19.99 64.50 6.40 100 22.22 66.96 18.13 100 18.97 60.66 13.28 24.70 100 49.76 14.08 13.29 42.58 14.64 100 5.36 56.49 36.95 100 12.28 58.92 52.05 100 7.35 62.16 31.23 61.78 100 33.73 7.78 58.98 30.06 7.08 100 58.43 31.14 11.40 100 29.62 21.96 100 19.61 100 100 100 100 smaller amounts. Within days of the registered voters on the standard 7-point

PSOnline www.apsanet.org 131 Table 3 Partisan Strength and Likely Caucus Attendance

% of All Likely %ofOwn % of Own Party Likely % of Sample Caucus Attendees Party Sample Caucus Attendees March August March August March August March August Strong Democrat 24.1 22.4 36.2 35.4 46.5 43.6 56.7 61.6 Weak Democrat 12.6 12.5 13.7 10.0 24.2 24.3 21.2 17.3 Independent—Leaning Democrat 15.2 16.5 11.8 12.1 29.3 32.3 18.4 21.1 Independent 6.3 9.3 1.5 2.4 n/a n/a n/a n/a Independent—Leaning Republican 10.3 10.2 6.4 6.4 24.5 26.0 16.1 15.9 Weak Republican 12.9 11.7 9.0 11.4 30.9 29.9 21.3 28.5 Strong Republican 18.6 17.4 21.4 22.3 44.6 44.1 56.9 55.6 For March survey, N = 1170 for entire sample, N = 489 for likely caucus sample. For August survey, N = 844 for entire sample, N = 555 for likely caucus sample.

Table 4 “If your caucus was TODAY, who would you support for president at your caucus? Just tell me the name.” [Asked of Likely Republican and Democratic Caucus Attendees Only]

Likely Democratic Caucus Percent Likely Republican Caucus Percent Attendees March August Change Attendees March August Change Edwards 35.34 26.67 −8.67 Giuliani 21.61 11.87 −9.74 Clinton 28.72 24.72 −4.0 McCain 21.19 3.28 −17.91 Obama 21.18 21.60 +0.42 Romney 18.12 28.37 +10.25 Richardson 1.06 9.63 +8.57 Thompson, F. 8.13 +8.13 Don’t Know 11.88 14.42 +2.54 Brownback 2.08 4.08 +2.0 Others 1.82 2.94 Tancredo 5.59 +5.59 Thompson, T. 5.10 2.42 +2.68 Gingrich 3.27 .86 −2.41 Don’t Know 23.52 27.61 +4.09 Others 5.13 7.79

Table 5 Partisan Bases of Support for 2008 Presidential Candidates, March Data [Percents]

Clinton Edwards Obama McCain Giuliani Romney Don’t Know Partisan Strength Strong Democrat 43.98 35.47 29.01 5.51 7.14 9.53 14.02 Weak Democrat 16.41 18.76 15.21 7.50 7.44 6.27 9.36 Independent—Leaning Democrat 18.54 21.57 20.12 8.95 4.59 13.43 12.82 Independent 3.13 6.50 7.11 3.79 3.98 7.44 15.54 Independent—Leaning Republican 5.09 5.90 8.35 22.51 16.83 11.30 8.16 Weak Republican 7.20 7.52 9.68 21.99 16.86 15.73 21.08 Strong Republican 5.65 4.28 10.52 29.77 43.17 36.31 19.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00

scale9—but from an unusual perspective. instance, we learn that Hillary Clinton election support came from voters who Instead of examining which candidates got nearly half ~44%! her support from are not Democrats. More than one-third are supported by strong Democrats, strong Democrats, far greater than either of all his support ~35.6%! came from strong Republicans, etc., we look at the of the other top Democrats. Equally people who cannot stand up for him bases of each candidate’s support. So, for striking is how much of Obama’s general in the Iowa caucuses ~unless they are

132 PS January 2008 willing to register as Democrats! while Obama did better among the most edu- caucus samples! and Obama ~in caucus about 24% of Edwards supporters and cated caucus attendees than among regis- samples! gained more support ~compared 21% of Clinton supporters were not Dem- tered voters with over 53% coming from to the baseline! from non-married voters. ocrats. While Obama could rightfully those with at least a college degree, Age. Likely caucus attendees are claim to be the Democrat who drew the while only 13.6% came from those with slightly older than registered voters over- most crossover support in Iowa, crossover a high school diploma or less, far below all, but this difference is not at all large. voters are not the ones who choose nomi- the 26.7% caucus attendee baseline. In Clinton and Edwards drew similar sup- nees. Clinton, on the other hand, clearly the end the demographics of the caucus port from all age groups—similar to mobilized strong Democratic partisans, electorate dramatically reduced the im- each other and close to the baseline for while John Edwards fell somewhere in portance of the least educated for Clinton both registered voters and caucus attend- between Clinton and Obama. ~Ϫ16.77! and Obama ~Ϫ10.48! but in- ees. But they both drew less support We turn now to a comparison between creased it for Edwards ~ϩ4.22!. from the youngest group. Obama appar- likely caucus attendees and registered Income. As with education, caucus ently picked up that support, receiving voters in order to examine the extent to attendees are skewed towards higher in- disproportionately more from the young which the two differ on dimensions other come levels. For registered voters, just ~ϩ3.5% for registered voters and ϩ8.1% than partisanship and candidate prefer- under one-third report household incomes for caucus attendees! and less from the ence. We are interested here in the “in- of $75,000 and over, while about a quar- old ~60 and up category!. While 18–29 ternals” of support for the candidates, ter report less than $30,000. But nearly year olds make up only 6.8% of likely that is the degree to which their bases 41% of likely caucus attendees are in the caucus attendees, they were nearly 15% came from different demographic and highest income group, while 16% are in of all Obama caucus supporters. An im- issue groups. Thus in Tables 6 and 7 the the lowest group. The deviation from the portant caveat was evident for Obama— column percentages sum to 100%, allow- registered voter baseline was not large for while young people might say they are ing us to compare support for each can- any of the candidates. But among caucus likely to caucus, much would depend on didate to the overall distribution of attendees, we again see differences that the effectiveness of the Obama campaign demographics and issues within the reg- could account for the relative change be- in actually getting them to show up on a istered voter and likely caucus attendee tween Clinton and Edwards among likely cold January night.11 Parallels between samples ~these baselines are shown in caucus attendees. While Clinton got over Obama and Howard Dean’s 2004 cam- the first two data columns of both 27% of her registered voter support from paign in terms of mobilizing the young tables.! Table 6 reports support for the those making less than $30,000 a year, were evident throughout 2007. top three Democrats while Table 7 shows less than 14% of her caucus support Religion. Likely caucus attendees do the Republican candidates using the came from that group ~about 2.3% below not differ much from registered Demo- March data. Given space limitations we the caucus baseline!. But she saw an in- crats on basic religious preference.12 will highlight only the most interesting crease among the wealthiest, with sup- Clinton’s registered voter support base differences in demographic and issue port nearly 6% over the baseline. was more dependent on Protestants than importance support for each candidate in Edwards and Obama, whose income were the other two; 68.3% of Clinton March 2007—respondents were simply bases were similar among registered vot- supporters were Protestant compared to asked how important particular issues ers, diverge when we look at caucus at- 58.2% for Edwards and 60.2% for were to their presidential vote in 2008. tendees. Obama’s support became heavily Obama. Obama, however, relied heavily upper income; nearly 50% of all his sup- on support from Protestant caucus at- Bases of Support for porters make over $75,000. Edwards be- tendees while Edwards appeared to be Democratic Candidates came more dependent on middle income the favored Democrat among Catholics; voters; he was disproportionately sup- he was the only candidate to receive a Education. About one-third of regis- ported by those with $30–75,000 in- higher percentage of support from Catho- tered Democrats in Iowa have no more comes, about 8% more than Clinton and lics than the baseline. This is stronger in than a high school diploma, while just 15% more than Obama. the caucus sample, where Edwards got over 39% have at least a college degree. Gender. Not surprisingly, Clinton’s almost a quarter of all his support from But those who say they will caucus are base was heavily female—about two- Catholics, an increase of 5.1% over the better educated—nearly half hold a col- thirds of her supporters in both samples sample population. lege degree or higher, while only a quar- were women, while women were only None of the Democratic candidates ter are high school graduates or less. 57% and 58.6% of the two baselines drew much support from born-again and Here we have our first indicator of why samples respectively. Obama, on the evangelical Christians, but of course most Clinton led among registered Democrats other hand, drew disproportionately from Democrats do not identify themselves as but not among caucus attendees. Among men, with 55% of his caucus support such.13 Edwards’s numbers jump out. He registered Democrats, 41% of Clinton’s base made up of male voters, well above received just 12.9% of his support from support came from the lowest education the 41.4% baseline. Edwards’s support in born-again and evangelical Christians levels ~a deviation of 7.3 over the base!, March 2007 was almost perfectly in line among registered Democrats ~Ϫ7% ver- a group less likely to caucus. Only 30% with the baseline of both registered vot- sus the baseline! and only 6.6% of his of Edwards’s base came from this group, ers and caucus attendees. caucus support from them ~Ϫ10.22%!. as did 24% of Obama’s. At the same Marital Status. Marital status, like In summary, while there were interest- time Clinton’s support relied less on the gender, seems to have little impact on ing differences between the three top best educated while Edwards’ support the likelihood of attending the caucus— Democrats in their demographic bases of was heavily dependent on this group both samples have roughly the same support, we find relatively little differ- ~31% compared to 47%; baseline of distributions on this demographic. Inter- ence between registered Democrats state- 39%! as was Obama’s ~49%!. Yet, estingly, Edwards drew disproportion- wide and likely Democratic caucus among likely caucus attendees, Clinton’s ately from the married subpopulation, attendees on most demographics, with the support was evenly balanced and mirrors especially among caucus attendees.10 exception of education and income. While the actual baseline for this group. But Both Clinton ~in registered voter and we saw earlier that those who caucus are

PSOnline www.apsanet.org 133 Table 6 Support for Democratic Candidates: Registered Democrats vs. Likely Democratic Caucus Attendees, March Data, Column Percents

Sample Baseline Clinton Edwards Obama Registered Caucus Registered Caucus Registered Caucus Registered Caucus Education H.S. Diploma or Less 33.81 26.74 41.14 24.37 30.05 34.27 24.07 13.59 Some college (4 Year 26.83 23.72 27.71 27.51 22.65 15.07 27.00 32.97 or Technical) Bachelors 21.89 24.43 18.27 25.38 26.16 24.54 24.66 23.53 Post-Bachelors 17.47 25.10 12.87 22.74 21.15 26.12 24.27 29.91 Total 100 100 100 100 100 100 100 100 Income Less than 30k 24.43 16.33 27.04 13.94 20.61 13.39 20.73 17.62 30k to under 50k 28.06 26.76 30.87 29.09 26.07 27.08 26.74 20.62 50k to under 75k 14.88 15.95 10.37 10.12 17.23 20.70 13.37 12.26 75k and up 32.63 40.96 31.72 46.85 36.09 38.83 39.15 49.50 Total 100 100 100 100 100 100 100 100 Gender Female 57.08 58.62 64.81 67.11 56.94 57.99 52.49 44.77 Male 42.92 41.38 35.19 32.89 43.06 42.01 47.51 55.23 Total 100 100 100 100 100 100 100 100 Marital Status Not Married 28.72 28.26 30.09 36.59 23.57 19.40 27.58 32.88 Married 71.28 71.74 69.91 63.41 76.43 80.60 72.42 67.12 Total 100 100 100 100 100 100 100 100 Age 18–29 5.07 6.81 4.46 6.33 2.93 2.14 8.63 14.95 30–44 16.35 11.19 16.57 7.96 18.31 13.34 18.29 10.71 45–59 38.55 39.84 36.19 40.86 38.41 40.33 39.03 39.25 60 and up 40.03 42.16 42.79 44.86 40.35 44.20 34.05 35.09 Total 100 100 100 100 100 100 100 100 Religion Protestant 62.62 63.06 68.33 66.05 58.23 57.56 60.16 67.33 Roman Catholic 22.24 19.45 20.82 18.37 23.43 24.51 21.91 16.72 Other 15.14 17.49 10.85 15.58 18.34 17.93 17.94 15.95 Total 100 100 100 100 100 100 100 100 Born Again Yes 19.98 16.80 19.64 14.57 12.91 6.58 16.22 19.43 No 80.02 83.20 80.36 85.43 87.09 93.42 83.78 80.57 Total 100 100 100 100 100 100 100 100 Issues (% who say issue is “very important” for their presidential vote)* Immigration 41.29 40.13 41.13 41.96 45.32 40.80 35.54 39.14 Iraq 38.52 42.16 36.95 33.96 43.29 49.34 39.63 46.39 Abortion 29.88 32.17 30.36 37.63 33.56 31.01 20.87 18.29 Gay Marriage 14.52 15.04 15.55 15.92 13.68 14.84 15.62 15.80 Economy 11.42 9.71 10.85 4.20 9.12 12.02 13.34 8.58 Terrorism 2.12 2.62 4.19 5.83 1.34 .87 1.35 1.44 *Issue marginals (last six rows) are the result of multiple questions, thus the categories do not add up to 100%. Immigration, abortion, and gay marriage percentages are the result of two questions and should not be compared directly to totals on Iraq, economy, or terrorism. All cross-tabs are statistically significant with a 95% confidence interval using a Chi-Square test.

clearly stronger partisans, there are few Policy Issues Democrats, immigration was considered other differences between these two very important by more than 41%, more groups of Democrats, perhaps laying to We now look at a brief examination of than any other issue. Iraq followed with rest some of the concerns that caucus six issues voters might consider impor- 38.5% and abortion at 29.9%. Gay mar- goers are substantially unrepresentative of tant to their presidential vote ~see Nor- riage, the economy, and terrorism were Iowa partisans, at least for Democrats. rander 1986!.14 Turning first to registered well behind, at 14.5%, 11.4%, and 2.1%

134 PS January 2008 Table 7 Support for Republican Candidates: Registered Republican vs. Likely Republican Caucus Attendees, March Data, Column Percents

Sample Baseline McCain Giuliani Romney Registered Caucus Registered Caucus Registered Caucus Registered Caucus Education H.S. Diploma or Less 32.97 29.15 33.04 21.96 24.15 28.97 26.85 25.69 Some college 27.20 27.85 33.5 27.45 25.74 20.51 28.24 35.93 Bachelors 27.50 27.85 20.25 32.94 37.11 32.31 28.08 28.13 Post-Bachelors 12.33 15.15 13.21 17.65 13.00 18.21 16.83 10.24 Total 100 100 100 100 100 100 100 100 Income Less than 30k 15.89 14.16 15.77 14.37 12.59 11.91 21.24 18.22 30k to under 50k 27.00 21.35 21.35 12.98 26.47 13.17 23.30 5.42 50k to under 75k 21.93 24.36 25.43 19.47 20.65 23.67 26.08 40.13 75 to under and up 35.18 40.13 37.45 53.17 40.29 51.25 29.38 36.23 Total 100 100 100 100 100 100 100 100 Gender Female 46.75 38.33 38.60 44.44 43.61 32.69 37.08 31.19 Male 53.25 61.67 61.40 55.56 56.39 67.31 62.92 68.81 Total 100 100 100 100 100 100 100 100 Marital Status Not Married 17.92 17.28 20.57 24.18 18.21 25.56 16.55 6.42 Married 82.08 82.72 79.43 75.82 81.79 74.44 83.45 93.58 Total 100 100 100 100 100 100 100 100 Age 18–29 5.29 6.47 6.21 4.44 5.57 10.77 4.70 3.82 30–44 18.59 15.81 19.05 26.41 17.12 16.15 10.29 14.07 45–59 33.93 32.60 38.18 36.21 36.09 24.74 32.38 47.55 60 and up 42.19 45.12 36.56 32.94 41.22 48.33 52.63 34.56 Total 100 100 100 100 100 100 100 100 Religion Protestant 75.53 81.79 70.16 70.54 77.16 81.99 80.97 83.66 Roman Catholic 16.48 12.13 20.68 21.49 16.73 13.19 11.91 8.17 Other 7.99 6.08 9.16 7.97 6.11 4.82 7.11 8.17 Total 100 100 100 100 100 100 100 100 Born Again Yes 43.45 43.64 37.40 25.14 33.89 41.06 45.70 38.79 No 56.55 56.36 62.60 74.86 66.11 58.94 54.30 61.21 Total 100 100 100 100 100 100 100 100 Literal Bible Interpretation Yes 63.52 61.75 60.73 44.22 60.21 63.23 65.49 63.07 No 36.48 38.25 39.27 55.78 39.79 36.77 34.51 36.93 Total 100 100 100 100 100 100 100 100 Issues (% issue is “very important” for their pres. vote)* Immigration 56.02 63.09 55.39 40.37 64.12 79.49 54.28 66.67 Abortion 44.38 41.64 41.73 28.63 42.42 35.51 52.68 50.00 Gay Marriage 38.89 41.60 40.93 36.21 32.98 31.15 41.24 44.80 Terrorism 19.15 26.71 15.94 9.93 31.23 50.51 21.64 25.69 Iraq 13.21 11.80 13.67 18.69 10.04 6.41 12.19 7.65 Economy 12.42 10.03 12.00 2.22 12.61 19.36 14.09 16.67 *Issue marginals (last six rows) are the result of multiple questions, thus the categories do not add up to 100%. Immigration, abortion, and gay marriage percentages are the result of two questions and should not be compared directly to totals on Iraq, economy, or terrorism. All cross-tabs are statistically significant with a 95% confidence interval using a standard Chi-Square test.

PSOnline www.apsanet.org 135 among Democrats. For those who parties are quite similar at the high and extra support from older voters ~60 and planned to actually caucus, the important low ends; about 16% of Democrats and up category!, who provided 52.62% of issues remained the same—the top three 14% of Republicans who plan to caucus his base. Interestingly, older registered were Iraq, immigration, and abortion, at report incomes below $30,000, while Republicans were a smaller part of 42.1%, 40.1%, and 32.2% respectively. around 40% of both groups report in- McCain’s base, 5.5% less than the base- Likely caucus attendees do not seem es- comes over $75,000. In this sense the line, despite the fact that he is by far the pecially different on issues from Demo- likely Republican caucus attendee is ac- oldest of the three candidates. Yet, when crats overall in Iowa. tually more representative of all regis- we look at likely Republican caucus at- The issue of the war in Iraq somewhat tered Republicans than is a Democratic tendees, age effects shift substantially for differentiated the Democrats in March caucus attendee relative to other Romney, whose strength among caucus 2007. Among registered voters who sup- Democrats. attendees was in the 45–59 group, 15% ported Edwards, 43.4% considered Iraq Across the Republican candidates, above the baseline, while his support to be very important, about 5% over the there was relatively little variation when among the oldest group dropped to baseline. The difference was larger comparing registered voters. But as with 34.5%, nearly 10% below. No other can- among caucus attendees where 49.3% of education, there were some large differ- didate showed such a dramatic shift, al- Edwards supporters felt this way, 7.2% ences when we examine likely caucus though Giuliani’s caucus base was over the baseline. For Obama, 46% of attendees. Both McCain and Giuliani somewhat older than his registered voter his caucus supporters considered Iraq to drew heavily on wealthy voters; McCain support. be very important, but for Clinton only received over 53% of his caucus support Religion. McCain was the candidate 34% of her supporters did. The flip side ~13% over the baseline! from those with least reliant on Protestants for his sup- was terrorism, which few Democrats higher incomes, while Giuliani got about port among both registered voters and considered very important. Clinton sup- 51% of his caucus support from this caucus attendees. In turn, McCain got porters, however, were much more likely group ~ϩ11.1%!. Romney was the least substantially more support from Catho- to cite terrorism—nearly 6% of her cau- reliant on support from the wealthy, lics. Giuliani’s support more or less mir- cus supporters called it very important, drawing disproportionately from the rored the proportions overall for both compared to 1.4% for Obama and less lower income voters. registered voters and likely caucus at- than 1% for Edwards. Clinton did very Gender. The gender gap was alive and tendees. Romney was the most reliant on well with those few Democrats who fo- well among Iowa Republican partisans, Protestants, with more than 80% of his cused on terrorism generally, as opposed as it is nationwide. Among registered support in both samples from this group. to the war in Iraq. Iowa Republicans, 53.3% are male while The Republican samples have double 61.7% of likely caucus attendees are the percentage of born-again and evan- Bases of Support for male. So where Democratic caucus at- gelical Christians than Democrats. Republican Candidates tendees slightly over-represent Demo- Among registered Republicans, Romney cratic women, the Republican likely led with 45.7% of his support from this Education. Just under 40% of regis- caucus attendee sample is heavily male. group. The caucus sample, however tered Republican voters in Iowa have at Both Giuliani and Romney relied heavily paints a different picture, with Giuliani least a college degree, while about a on men for their caucus support, both most reliant on them, at 41.1% of his third have no more than a high school about 6–7% above the baseline. Women base. Although this was the highest per- diploma. Likely caucus attendees are formed a larger share of McCain sup- centage among the three candidates, it skewed a bit towards the better educated, porters relative to the others, more than was still below the sample baseline, sig- but not nearly so much as the Demo- six points over that caucus attendee nifying that many had yet to settle on crats. But there were clear differences baseline. any of the leading candidates and were among the candidates in their bases. Marital Status. Among registered vot- disproportionately undecided. McCain Only 33.7% of the registered Republi- ers, married Republicans form about the got less of his support from born-again cans who supported McCain have a col- same support base for each of the candi- and evangelical Christians than did the lege degree or higher, 6.1% below the dates, none of which differ much from other two, 18.5% below the baseline for baseline. Giuliani and Romney, on the the baseline. And both married and un- caucus attendees. A similar story is found other hand, both drew a better educated married Republicans say they will attend among those who believe in a literal in- base among registered voters—Romney a their caucus in equal proportions to their terpretation of the Bible, who make up little more than 5% above the baseline representation among registered voters. more than 60% of both Republican and Giuliani more than 10% above. Sup- However, among Romney caucus sup- groups.15 Among registered voters Rom- port shifted when looking at only likely porters, more than 93.5% were married, ney had the greatest support here, while caucus attendees. McCain and Giuliani well above the 82.7% caucus attendee Giuliani did the best among likely cau- drew more caucus support from the bet- baseline. Unmarried Republicans were cus goers. While McCain did better than ter educated ~both about 8.5% above the simply not supporting Romney, relative he did among born-again Christians in baseline! while Romney’s caucus support to the other candidates. the registered voter sample, he still got was based more among those less Age. As with Democrats, Republican the lowest share of his caucus support educated—he received the lowest level likely caucus attendees are not substan- from this group ~44.2%, which is 19% of support from the most educated. This tially different in age than registered Re- lower than Giuliani and 18.8% lower highlights how divergent caucus support publicans overall, they may be just a than Romney!. It is probably safe to con- appeared to be from registered support little bit older. Age did condition support clude McCain was not the favored candi- for the Republican presidential candi- for the candidates however, both among date among conservative Christians. dates, at least in early 2007. registered voters and caucus attendees. Summarizing the demographic bases of Income. As with the Democrats, Re- Among registered voters, McCain’s sup- support for the candidates, we can con- publican caucus attendees have higher port was more reliant on middle-aged clude that likely Republican caucus at- incomes than registered Republicans voters between 45–59 ~38.18%, 4.25% tendees, who like Democrats are stronger overall. Yet the caucus samples for both over the baseline!, while Romney got partisans, are slightly better educated,

136 PS January 2008 wealthier, more likely to be male, and a ference was heightened among likely While Clinton and Giuliani emerged little bit older than registered Republicans caucus goers, with over 50% of as frontrunners among statewide Iowa as a whole in Iowa. We find that the Re- Giuliani’s supporters focused on terror- voters early in the season, the analysis publican candidates did in fact rely on ism. Giuliani supporters also considered reveals why Obama, Edwards, and Rom- different bases for their support, espe- immigration to be very important—a full ney could mount credible challenges in cially in the caucus, more so than the 79.5% of his caucus supporters reported Iowa, with Obama gaining support Democrats. While Democratic caucus this, compared to 63% of all caucus at- among the most educated and wealthy attendees differ only slightly from regis- tendees. In the minds of many, these two Democrats, as well as the young and tered Democrats, Republicans who cau- issues probably hung together well, as those opposed to the Iraq war. However, cus do not look as much like registered they might well have considered illegal the young may be the least likely to turn- Republicans statewide. immigration might to be a national secu- out at the caucuses, which may hurt rity issue. Obama relative to Edwards and Clinton. Policy Issues Turning to the social issues—abortion Romney led among social and religious and gay marriage—Romney supporters conservatives and earned support across While Iraq and immigration were tied considered these to be very important to a income and education groups. Edwards for Democrats as very important issues, much greater degree than supporters of gained support from the middle of the for registered Republicans immigration either of the other two candidates. This income and education spectrums. Only was cited by more than 56% as very im- was especially true among the likely cau- one candidate in each party gets to “win” portant, followed by abortion ~44.4%!, cus attendee sample, where 44.8% of the Iowa caucuses, yet more than one gay marriage ~38.9%!, terrorism ~19.2%!, Romney voters called gay marriage im- candidate may do “better than expected” Iraq ~13.2%!, and the economy ~12.4%!. portant ~compared to 41.6% overall! and and in doing so propel him or herself to Given that Republican candidate rhetoric 50% considered abortion to be a very im- the nomination. downplayed Iraq and played up terrorism portant issue ~versus 41.6% overall, and The empirical evidence also revealed ~an especially potent issue for Giuliani! only 28.6% of McCain voters!. Giuliani similarities and differences between the this is not surprising. Immigration was seemed to be capitalizing on “tough guy” preferences of likely caucus attendees ~and is! a hot button issue in Iowa where issues—drawing heavily on terrorism and and Iowa registered voters. On policy the farm economy and the meatpacking immigration. Romney appeared to be rid- issues, likely caucus attendees map industry rely on Latino immigrants, and ing social and moral issues like gay mar- closely to statewide voters within each abortion and gay marriage are key issues riage and abortion. McCain, though, was party. However, the divergence among for the conservative Republican base. different. The only issue he seemed to Democrats—with Edwards leading in the Likely Republican caucus attendees were draw an abnormally high amount of sup- caucus while Clinton led among all reg- even more focused on immigration as a port from was the Iraq war. istered Democrats—provides fodder for very important issue, with over 63% cit- critics that the existing system produces ing it. These voters were also much more Conclusion polarized candidates. The analysis also likely to see terrorism as very important highlights some socioeconomic and par- than registered Republicans Our intent in this paper is to illumi- tisan biases of caucus attendees relative The candidates’ bases of support by nate some of the differences between to the Iowa electorate; in particular indi- issues tracked quite well with their gen- voters who are likely to attend the Iowa vidual wealth and education matter and eral overall approaches to the nomination caucuses and registered voters statewide, affect one’s likelihood of caucusing and campaign. McCain focused more on and to compare who supports the presi- presidential candidate preferences. The Iraq—albeit generally in support of the dential candidates for both parties over current presidential nomination system war—than did the other two and more time. These data provide a window into requires that each candidate compete in a than 18% of McCain’s caucus supporters opinion change and candidate prefer- handful of early nominating states ~Iowa, called Iraq very important, versus under ences over the course of a campaign, and Nevada, and New Hampshire!, win the 12% of all likely caucus attendees. On also allow us to compare across catego- favor of political activists, and be elect- the other hand, McCain supporters were ries of voters ~politically active caucus able in the general election. In the midst much less likely to think any of the other attendees to regular voters!. The Iowa of aggressive frontloading, the 2008 elec- five issues were very important, com- caucuses play a key role in the presiden- tions may be the last before the states pared to the caucus attendee baseline. tial nominating process ~Squire 1989; default to some pseudo-national primary. Among those who considered terror- Adkins and Dowdle 2001; Stone, Rapo- We share the findings of the Iowa elec- ism very important, Giuliani was win- port, and Abramowitz 1992!, and as such torate as a marker in the midst of rapid ning hands down. Among registered it is useful to understand both the basic change of state election laws governing voters, 31.2% of Giuliani supporters said nature of those most likely to caucus and the presidential nomination process with terrorism was very important, a differ- the wellsprings of support for candidates larger more comprehensive election re- ence of almost ϩ10% compared to the as they began the long haul towards the forms looming on the horizon. other two leading Republicans. This dif- nomination.

Notes * The University of Iowa Hawkeye Poll is all these people along with the team of graduate tended; the number was much smaller for Re- co-directed by David Redlawsk and Caroline and undergraduate students who worked with us publicans who did not have a nominating Tolbert. It was administered with the support of on the survey project. contest that year. the University of Iowa Social Science Research 1. Those who are not registered party 2. Of the 1,290 poll participants 32% were Center, Director Kevin Leicht, and funded by the members may choose to register at their caucus, Republican, 36% Democrat, and 32% indepen- University of Iowa Office of the Provost and the so they may participate that evening. In 2004, dent, which closely matches the actual party reg- College of Liberal Arts and Science. We thank approximately 20% of registered Democrats at- istration of Iowa voters. Nearly three-quarters

PSOnline www.apsanet.org 137 were married. Women made up 62% of the raw responses including “yes”, “no” and “does not 9. “In politics today, do you consider your- sample. Of the full sample, 508 were identified know0cannot remember.” self a Republican, Democrat, or Independent?” as likely caucus attendees. Of these, 310 planned Question 2: “Many people find it difficult to Follow-up questions asked “Do you consider to attend the Democratic caucus while 198 attend the caucuses because it takes a lot of yourself a strong Republican or not a strong Re- planned to attend the Republican caucus. The time. How likely are you to attend your Caucus publican?” and “Do you consider yourself a margin of error for all likely caucus attendees is in January 2008? Are you very likely, somewhat strong Democrat or not a strong Democrat?” ϩ0Ϫ4.4%, 5.5% for the Democratic sub-sample, likely, not very likely, or not at all likely?” Independent leaners were measured by the ques- and 6.5% for the Republican sub-sample. Question 3: “Which party’s caucus do you tion “As of today, do you lean more to the Re- 3. Of the statewide registered voter sample, plan to attend, the Democratic Party Caucus or publican Party or more to the Democratic 29.1% of the participants were Republican, the Republican Party Caucus? Democratic Cau- Party?” 34.9% were Democrat, and 36% were inde- cus, Republican Caucus, or does not know.” Re- 10. “What is your marital status? Are you pendent. Nearly 70% were married. Women spondents providing a valid response to Question married or with a committed partner, separated, made up 57% of the raw sample. In the caucus 1, who noted “very likely or “somewhat likely” widowed, or have you never been married?” sample, 542 were identified as likely caucus at- to attend the caucuses in Question 2 and who 11. As we write this in early November tendees. Of these, 319 planned to attend the knew which parties caucus they would attend, 2007, one of the major unknowns is what will Democratic caucus while 223 planned to attend question 3, were coded as likely caucus happen with college student attendance. In the the Republican caucus. The statewide poll had a attendees. past caucuses were held while classes were in margin of error of ϩ0-3.25 percentage points. 7. As we write this we are analyzing results session at most colleges; in 2008 virtually every The margin of error for all likely caucus attend- from our third survey, in late October 2007. student will be on winter break, and most likely ees is ϩ0-4.4%, 5.5% for the Democratic sub- Huckabee’s momentum coming off the Straw in their home community rather than their col- sample, and 6.5% for the Republican Poll is evident, in the most recent survey he was lege town. In addition, at the University of Iowa, sub-sample. at just under 12%, tied for second behind Rom- more than 40% of undergraduates are from out 4. The poll survey was administered using ney at about 36%. By the time this is published of state and will presumably not be in Iowa on WinCATI, a computer aided telephone interview- we will know if his momentum continued to January 3. ing system. Paid student callers were used for caucus day. 12. “What is your religious preference?” The the August survey while student callers received 8. In March, while 49% of the Democrats response was recorded and then summarized into class credit for their involvement in the March who initially could not choose a general election the broad categories used here. survey and were trained ahead of time to follow candidate remained undecided in their caucus 13. “Would you describe yourself as a “born a specified script. The pedagogy of this project preference, Edwards received support from again” or evangelical Christian, or not?” is interesting in itself, but space precludes dis- nearly 60% of the previously undecided who 14. “How important is @terrorism# to your cussion here. name a caucus preference, while Clinton and vote for president in 2008? Is it very important, 5. “Given the people who now say they are Obama split most of the rest. All three candi- somewhat important, or not that important?” running for president in 2008, how satisfied are dates retained more than 92% of their general Identical question wording was used for Iraq you with your choices. Are you very satisfied, election supporters as well. On the Republican War, economy, abortion, gay marriage, immigra- somewhat satisfied, or not satisfied?” side, all three major candidates maintain at least tion, health care, energy policy, environment, 6. Question 1: “As you know, the Iowa cau- 90% of their support, while 72% of those un- and education. Responses also included “refuses cuses are the first test of presidential candidates decided in the general election remained un- to say” and “does not know.” in 2008. I’d like to ask you some questions decided. McCain got about half of those 15. “Do you believe in a literal interpreta- about the caucuses. First, did you attend a presi- previously undecided who chose a caucus tion of the Bible?” Respondents could answer dential caucus in either 2000 or 2004?” with preference. yes or no.

References Adkins, Randall, and Andrew Dowdle. 2001. for Multiparty Politics in the United States.” _. 1986. “Correlates of Vote Choice in the “How Important are Iowa and New Hamp- Social Science Quarterly 86: 147–59. 1980 Presidential Primaries.” Journal of Pol- shire to Winning Post-Reform Presidential Hull, Christopher C. 2007. Grassroots Rules: itics 48 ~1!: 156–66. Nominations?” Political Research Quarterly How the Iowa Caucus Helps Elect Ameri- Squire, Peverill, ed. 1989. The Iowa Caucuses 54 ~2!: 431–44. can Presidents. Stanford, CA: Stanford Uni- and the Presidential Nominating Process. Democracy in Action. 2007. Iowa Caucuses Web versity Press. Iowa City: University of Iowa Press. Site. www.gwu.edu0;action020080 Fiorina, Morris P., Samuel J. Abrams, and Stone, Walter J., Ronald B. Rapoport, and Alan chrniowa08.html ~June 29, 2007!. Jeremy C. Pope. 2004. Culture War? The I. Abramowitz. 1992. “Candidate Support in Donovan, Todd, and Shaun Bowler. 2004. Re- Myth of a Polarized America. New York: Presidential Nomination Campaigns: The forming the Republic: Democratic Institu- Longman. Case of Iowa in 1984.” The Journal of Poli- tions for the New America. Upper Saddle Norrander, Barbara. 1989. “Ideological Repre- tics 54 ~4!: 1,074–97. River, N.J: Prentice Hall. sentativeness of Presidential Primary Voters.” Donovan, Todd, Janine Parry, and Shaun Bowler. American Journal of Political Science 33 2005. “O Other, Where Art Thou? Support ~3!: 570–87.

138 PS January 2008