SOSS Bulletin Preliminary Draft 1.1 Presidential Race Nip and Tuck in Michigan Darren W. Davis Professor of Political Science Brian D. Silver Director of the State of the State Survey (SOSS) and Professor of Political Science Douglas B. Roberts Director of the Institute for Public Policy & Social Research (IPPSR) Institute for Public Policy & Social Research Michigan State University East Lansing, Michigan 48824 June 28, 2004 Acknowledgments: The data upon which this report is based are from the State of the State Survey (SOSS), which is conducted by the Office for Survey Research of the Institute for Public Policy& Social Research (IPPSR) at Michigan State University. The results and interpretations reported here are the responsibility of the authors. Neither IPPSR nor MSU is responsible for any errors of fact or interpretation in this report. Further information about SOSS is available at http://www.ippsr.msu.edu/soss. Contact Brian Silver at: 517-420-5575 or bsilver@msu.edu.
1. The Presidential Vote Michigan is one of the battleground states that are considered potentially decisive to the 2004 reelection fortunes of President George W. Bush in his race against the expected Democratic nominee John F. Kerry. As part of its quarterly survey of opinion in Michigan, the State of the State Survey (SOSS) has examined both the public images of these candidates and the expected vote choices on November 2 nd. With the nominating conventions still pending July 26-29 for the Democrats, and August 30-September 2 for the Republicans Michigan citizens have already been bombarded with television ads as well as several visits by each candidate. Ralph Nader, an independent, is also competing for votes in Michigan. Based on the results of the 34 th State of the State Survey (SOSS) completed by the Institute for Public Policy & Social Research on June 15th, the presidential race in Michigan is too close to call. By a simple count of the expected votes by 736 likely voters (see box on likely voters in Technical Notes), Kerry receives 42% of the votes, Bush 40%, and Nader 6%, with 12% of the voters still undecided. These estimates have a margin of sampling error of "3.6%. The SOSS figures are consistent with other polls conducted in early June, though the percentage undecided in SOSS is somewhat larger, probably mainly as a result of the fact that the SOSS survey began in late April when fewer voters had made up their minds. 1 Expected 2004 Vote, by Region. Examining the vote preferences by region (Table 1 and map) reveals that Bush is supported by a majority of likely voters in two regions: West Central and East Central. Kerry commands a majority in the UP, Southwest, and Detroit. 1 As reported by PollingReport.com, in an EPIC/MRA poll of 600 active voters in Michigan on June 2-6, 45% favored Kerry, 43% favored Bush, 3% favored Nader, and 9% were unsure. Many polls identify leaners among the undecided voters and add them to the expected percentages committed to the named candidates. SOSS did not count leaners in this survey, but will do so in later rounds this year campaign. 1
The region with the largest undecided vote is Detroit, and the next in order is Southeastern Michigan (excluding Detroit). Whereas in the rest of the state no more than 10% are undecided, and no more than 5% support Nader or another third party candidate, in these two southeastern regions one out of 6 voters have not made up their minds, and another 7 or 8% favor neither Bush nor Kerry. Table 1. Vote Preference by Region of State Bush Kerry Nader/Oth Undecided UP 35% 56% 4% 4% 100% Northern LP 43% 48% 2% 7% 100% W. Central 50% 38% 3% 9% 100% E. Central 57% 30% 5% 8% 100% SW 37% 50% 3% 10% 100% SE 39% 38% 8% 16% 100% Detroit 16% 59% 7% 18% 100% Statewide 40% 42% 6% 13% 100% Race Differences. Vote preferences differ substantially by race. Whereas among whites, 44% declare their intention to vote for Bush (and 38% for Kerry, and 5% for Nader or another candidate), among African Americans 69% favor Kerry (only 12% favor Bush, and 2% Nader/Other). We would surmise that if the undecided African American voters do go to the polls in November they are likely to vote for Kerry; perhaps their indecision at this stage in the campaign is due to their not having focused as much on the electoral campaign as have residents in other regions of the state. Since nearly 90% of African American voters supported Democratic candidate Albert Gore in the 2000 election, few of the undecideds among African Americans are likely to support Bush in 2004, especially because of the controversy over the final counting of ballots in Florida in 2000. The undecided whites come disproportionately from those who are psychologically crosspressured. For example, among those who identify themselves as both conservative and Democrat, 38% report that they are undecided about for whom to vote in November. The other combination of party ID and ideology with a disproportionately large number of undecideds are the middle-middle citizens who say they are both moderate in ideology and independent in partisanship. Among this group of voters, who may be disinclined in general to take sides in the political fray, 21% report that they are undecided about their vote choice at this time. Gender Differences. Women prefer Kerry to Bush: 45% to 38%, with just 3% preferring Nader, but 15% undecided. Men prefer Bush: 42% to 39%, with 10% preferring Nader, and 10% undecided. Age Differences. Kerry finds his greatest level of support among younger voters, where he earns three votes for every one that Bush earns. He finds his next greatest support among persons age 65 and over (Table 2). 2
Table 2. Age and Candidate Preference among Likely Voters 18-24 25-29 30-39 40-49 50-59 60-64 65+ All Bush 22% 32% 43% 48% 43% 42% 32% 40% Kerry 62% 29% 38% 30% 46% 49% 55% 42% Nader/Oth. 10% 31% 1% 5% 5% 0% 1% 6% Undecided 6% 8% 18% 17% 6% 9% 12% 12% Base N 50 65 148 161 118 45 137 724 Total Pct. 100% 100% 100% 100% 100% 100% 100% 100% The Partisan and Ideological Divides. Using a standard set of questions about party self-identification, 45% of Michiganians identify themselves as Democrats, 36% as Republicans, and 19% as independent or other (Table 2). 2 SOSS has been tracking party ID statewide in Michigan for 11 years. The distribution of party identifications had changed very little in the past several years, including during the last year. Party ID is a strong factor affecting people s voting decisions on election day. However, a majority of Michiganians 55% identify themselves as conservative or as moderate but leaning in a conservative direction; in contrast, just 31% identify themselves as liberal or those leaning toward the liberal side, while 19% call themselves moderate (Table 3). The ideological distribution of Michiganians as a whole has also changed little in recent years, including during the last year. Table 2. Party ID of Michigan Residents Freq. Percent StrongRep 135 14 WeakRep 121 13 IndepRep 89 9 Independent 177 19 IndepDem 113 12 WeakDem 143 15 StrongDem 168 18 Total 946 100 2 Our way of measuring party ID follows that of the American National Election Study, and begins with the question, Generally speaking, do you consider yourself to be a Republican, a Democrat, and Independent, or something else? Those who answer Republican or Democrat, are asked Would you consider yourself a strong [Republican/Democrat] or not a very strong [Republican/Democrat]? Those who answer Independent are asked Do you generally think of yourself as closer to the Republican Party or the Democratic Party? These answers are used to derive a 7-point scale ranging from Strong Republican to Strong Democrat. Our measurement of political ideology is analogous and begins with the question Generally speaking, do you consider yourself to be a conservative, a moderate, or a liberal? Analogous follow-ups are used to create a 7-point scale. 3
Table 3. Ideology of Michigan Residents Freq. Percent VeryCons 92 10 SomewhCon 268 29 LeanCon 151 16 Moderate 139 15 LeanLiberal 152 16 SomewhLib 100 11 VeryLib 35 4 Total 937 101 Expected 2004 Vote Choice, by Respondent s Party ID and Ideology. The current candidate preferences among likely voters in the November elections show clear effects of both partisanship and ideology. Both Republican and Democratic party identifiers are strongly committed to the candidate of their own party. In contrast, independents spread their support broadly: although more prefer Bush than Kerry (31% vs. 22%), 20% support third-party candidates such as Nader, and 26% remain undecided. Thus, the electoral contest on the Michigan battleground mainly involves how independents will vote. The picture with respect to ideology is different. While overwhelming majorities of liberals are committed to voting for Kerry, 56% of moderates also prefer Kerry as opposed to 26% who prefer Bush. Kerry has more support among conservatives than Bush does among liberals. Of course a significant part of this conservative crossover vote consists of selfidentified Democrats. However, it is also true that Kerry nationally is perceived as being about as close to the average voter s values as is Bush. 3 Thus, one should not assume that the average voter perceives Kerry as a liberal. Table 4. Party ID and Candidate Preference among Likely Voters Bush Kerry Nader/Other Undecided Total StrongRep 91% 3% 1% 5% 100% WeakRep 81% 4% 2% 13% 100% IndepRep 79% 6% 6% 9% 100% Independ. 31% 22% 20% 26% 99% IndepDem 8% 74% 12% 6% 100% WeakDem 6% 84% 0% 10% 100% StrongDem 0% 87% 1% 12% 100% All 40% 42% 6% 12% 100% 3 See Washington Post, June 21, 2004, in which 46 percent of respondents say that Bush shares your values, compared to 48 percent who say that Kerry shares your values. 4
Table 5. Ideology and Candidate Preference among Likely Voters Bush Kerry Nader/Oth Undecided Total VeryCons 84% 11% 1% 4% 100% SomwhatCon 57% 17% 12% 15% 101% LeanCon 52% 32% 3% 12% 99% Moderate 26% 56% 3% 15% 100% Lean Lib 10% 78% 2% 10% 100% SomwhatLib 3% 74% 7% 16% 100% VeryLib 4% 79% 0% 17% 100% All 40% 42% 6% 12% 100% These results reflect not just the influence of partisanship and political ideology as fixed characteristics of people but also the influence of events and the issues that develop over the course of a presidency and the campaign itself. Moreover, vote preferences are also influenced by people s sense of economic well-being and optimism about the future. The personal characteristics and the appeal of the candidates also matter independently of partisan and ideological considerations. Bush s Approval Ratings. Unlike his challengers, an incumbent President s performance as president can be an important factor in itself that affects whether voters support his reelection. When asked How good a job has President George Bush been doing as President? Michiganians approval of this performance had changed markedly in the past two years. In Winter 2002 (SOSS24), 77% of Michiganians judged President Bush s performance to be excellent or good (Table 6). This was at a time when Americans rallied behind the President and the government in the immediate aftermath of the horrific terrorist attacks on the World Trade Center and the Pentagon. By May-June 2004 (SOSS34), the rally effect had completely dissipated and Bush s popularity was low: only 40% of Michiganians judged the President s job performance to be excellent or good. This has set the stage for Bush s highly competitive race against his Democratic challenger. Table 6. Bush s Overall Performance Rating, Winter 2002 through Spring 2004 SOSS 24 25 26 27 28 29 30 31 32 33 34 Win02 Spr02 Sum02 Fall02-1 Fall02-2 Win03 Spr03 Sum03 Fall03 Win04 Spr04 39% 27% 25% 21% 20% 14% 20% 14% 10% 17% 12% EXCELLENT GOOD 38% 45% 44% 38% 37% 29% 39% 38% 26% 27% 28% FAIR 18% 22% 20% 25% 29% 29% 25% 26% 31% 24% 26% POOR 4% 5% 10% 16% 14% 28% 16% 23% 33% 32% 34% Base N 998 939 933 915 974 983 962 944 968 929 946 Total Pct. 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 5
Bush vs. Kerry: Who s Is Perceived to Be Better at What? President Bush is perceived as better able to protect America from terrorism at home, while Senator Kerry is perceived as better able to lead America s foreign policy (Table 7). The latter result is consistent with a recent national ABCNews/Washington Post poll that showed for the first time that Kerry is perceived as better able to lead our international policy as well as better able to lead the war against terrorism, if we interpret that question as referring to the fight against terrorism abroad. 4 However, a USAToday/CNN/Gallup Poll national which was published on June 25, 2004, shows that Bush is perceived as better able than Kerry to handle terrorism, by a margin of 54% to 40% -- a finding that parallels our own result in Michigan. 5 John Kerry is perceived as better able than George Bush to manage the economy, to reduce unemployment and create jobs, to protect civil liberties, and to protect the environment. 2. Changing of the Guard in Michigan? In addition to voting for President, on November 2 nd Michigan voters will have the opportunity to vote for the state legislature and U.S. Congressional seats. Furthermore there is a possibility not only that partisan control of the White House will change from the Republicans to the Democrats but that the two houses of Congress, as well as the two houses of the Michigan State Legislature will change from Republican to Democratic control. In sharp contrast to the virtual tie among Michigan voters in the race between Bush and Kerry, and despite the plurality of self-identified Democrats in the Michigan electorate, 45% of likely voters in Michigan expect to vote for the Republican candidate in their district for the Michigan House of Representatives, while 36% expect to vote for the Democratic candidate, and 18% volunteer that it depends on the candidate. Judging by these responses, it would appear at that the Republican majority in the Michigan legislature is likely to be preserved. And yet, when we asked prospective Michigan voters whether they would like to see the Michigan State Legislature remain in control of the Republicans, or would you like to see the Michigan Legislature change to Democratic control, 52% say they would like to see a change to Democratic control, compared to 41% who would like Republicans to remain in control, and 7% who volunteer that it depends on who the candidates are. Thus, based on their own stated vote intentions, Republicans are likely to maintain control of the Michigan legislature, yet most people prefer a change to Democratic control. We would speculate that this reflects the public s desire for increased cooperation or harmony between the governor and the legislature. Perhaps, however, the 18% who say that their vote for the legislature will depend on the candidates will tend toward the Democratic side if the candidates are of high enough quality. In fact, 42% of those who say that it depends on the 4 See Bush Loses Advantage in War on Terrorism, Washington Post, June 21, 2004, and supporting data at Hwww.washingtonpost.com/wp-srv/politics/polls/trend_062104.htmlH. 5 Susan Page, Poll: Sending Troops to Iraq a Mistake, USA Today, May 25, 2004. 6
candidates say they would prefer a Democratic majority in the state legislature to be elected in November, while 30% say they would prefer a Republican state legislature, and 27% say again that their preference depends on the candidates. The seemingly contradictory preferences in answer to these two questions remind us of the cliché that the only poll that really counts is the one that takes place on election day. We also have to bear in mind that because of partisan gerrymandering of district boundaries, the relationship between votes cast and seats won in the legislature can be distorted on a statewide basis -- in the 2004 election in a way that is likely to favor the Republicans. 3. Conclusions The presidential election contest in Michigan remains close. However, based on our surveys as well as those reported by major news organizations, public opinion in some key issues appears to be changing very fast on a national scale. This includes people s assessment of the situation in Iraq and the conduct of foreign policy generally, about which John Kerry has recently come to be perceived as better able to manage the job than George Bush. In the context of low presidential approval, the 9/11 commission hearings and findings about failures and weaknesses in America s homeland security, and continuing difficulties and cumulating costs to Americans lives, resources, and prestige because of the Iraq events, increasing attention has focused on the fundamental ability of the President to manage either domestic or international affairs. However, Senator Kerry is far from carrying the day in Michigan, much less the nation. Moreover, by traditional reckoning the campaign has scarcely begun. Neither major party has even nominated its presidential candidate. More so than the campaigns themselves, however, events beyond the campaigns continue to shape and reshape perceptions of the prospects of the contending parties and candidates. 6 6 We will rely on later rounds of SOSS to monitor developments through the Summer and Fall. Our work on these surveys is related to two national surveys that we have conducted concerning the impact of the terrorist threat on Americans political beliefs, in particular their willingness to trade civil liberties for greater security. We will have a third national survey in the field during the Summer and early Fall 2004. For further information about this Civil Liberties Project, see Hwww.msu.edu/~bsilver/CivLibPage.htmH. 7
Table 7. Which Candidate Can Do a Better Job of...? Bush or Kerry? (likely voters)* Bush Kerry Both Equally, Neither Protecting America from 61% 32 7 Terrorism in the U.S.? Conducting America s 44 49 7 Foreign Policy? Managing the 44 51 5 Economy? Protecting Your 40 52 8 Civil Liberties? Protecting the 33 61 6 Environment? Being Strong Leader 59 36 5 In Time of Crisis? Reducing Unemployment, 36 57 7 Creating Jobs? * Volunteered responses Someone else, Don t know, and Refuse to answer are excluded from the calculations. 8
Technical Notes Survey Design The State of the State Survey (SOSS) is a telephone survey using Random Digit Dialing (RDD) technology and is conducted by the Office for Survey Research of the Institute for Public Policy & Social Research at Michigan State University. The sample is based on a stratified random sample of adults age 18 and older living in Michigan. The sample strata are the regions used by the MSU Extension Office. The data analysis weights the results so that they are representative of the adult population of Michigan. This report is based on the 34 th wave of SOSS, which was conducted between April 19 th and June 15 th, 2004. The median interview date was May 14 th ; the first quartile ended on April 28 th, the third quartile on May 23 rd. The total number of completed interviews is 962. The sampling error is ±3.2%. The sampling error for likely voters (N=736) is ±3.6%. The questions about the presidential election were embedded in a larger survey that took an average of 17 minutes. Likely Voters Surveys generally overstate the voting rates of the population, for two basic reasons. First, respondents tend to be self-selecting: people who respond to surveys are more likely to participate in politics than the population as a whole. In that sense, participation in a survey is akin to participation in politics. Second, respondents often try to appear to be good citizens by saying that they are likely to vote (or that they did vote) even if they are actually nonvoters. This so-called social-desirability bias is especially likely among those who truly care about politics and are committed to the norm of voting. Most surveys cannot correct for these two types of error when estimating vote outcomes for a population. However, in order to reduce the latter effect, respondents are often screened by survey organizations to exclude those who are highly unlikely to vote. In this survey, we exclude as unlikely to vote those who either (a) are not currently registered to vote, or (b) do not say they expect to vote in the November presidential election, or (c) say they did not vote in the 2000 Presidential election even though, by age at least, they were eligible. The remaining respondents, the likely voters, tend to be older, more economically secure, and more likely already to have decided already for whom they would vote if the election were held today. They are also somewhat more conservative. In this survey, the number of likely voters is 736, out of a total of 962 respondents. 9