Theme: This paper describes the Latino vote in the 2004 presidential election. It pays particular attention to conflicting claims regarding the extent to which Latinos voted for President Bush. It also evaluates the overall impact Latinos had on the results of the election.
Analysis: The 2004 election lived down to expectations. It closely followed the pattern set in 2000. That is, President Bush’s victory slightly expanded his electoral base by adding Iowa and New Mexico to the states he carried previously while only losing New Hampshire from the states he won in 2000. Additionally while he lost the popular vote in 2000 by 539,947 votes, in 2004 he won it by 3,311,608, increasing his percentage of the vote from 48% to 51%. President Bush, thus, won in 2004 with a small minority that provides a legitimate basis for claiming a clear victory but not a mandate.
Preliminary analysis cites three unexpected patterns to explain these results. The least controversial of these is that, consistent with opinion polls entering the final weeks of the campaign, exit polls found that instead of strongly supporting the Democratic candidate as they historically have done, women –who constitute 54% of the electorate– increased the support they gave to Bush by 5% and narrowed the Democratic candidate’s normally robust majority among women to 51%-48%. This swing alone accounts for much of Bush’s victory margin.
A second but much more dubious claim is that Bush’s victory was heavily influenced by voters’ concern with ‘moral values’. The validity of this assertion is muddled because of how the exit polls operationalised this concept. Rather than specific items regarding issues such as gay marriage and abortion, the exit polls relied on a general category of ‘moral issues’, the saliency of which is unknown. Post-election surveys found that the importance of moral values varies with how the question is framed. When comparing the significance of moral values with Iraq and terrorism, 27% indicate greater concern for the former, but when asked to identify the issue that most influenced their vote, significantly fewer (14%) mentioned moral values. Also, while 44% of those who chose moral values as the most important factor in their vote say the term relates to specific concerns over social issues, such as abortion and gay marriage, the majority referred to more general issues such as the candidates’ personal qualities or to religion and values in general. This ambiguity ensures that the meaning and impact of moral values will be the focus of extensive research in the future.
The third and perhaps most unexpected claim regarding the outcome of the election concerns the alleged dramatic increase in Hispanic support for Bush. Initially, the National Election Pool conducted by Edison Mitofsky reported that Bush received 44% of the Latino vote, up from 35% in 2000. Numerous Latino and non-Latino commentators began using this figure as proof that Latinos could no longer be seen as part of the core Democratic vote but were instead swing voters who would respond to whichever party and candidate made them the best offer. The most outlandish such claim was by Dick Morris, the well known political consultant whose former clients include the President, who announced that ‘the biggest reason for Bush’s victory was that he finally cracked the Democratic stranglehold on the Hispanic vote’.[1]
There are several reasons why the accuracy of the alleged 44% total cannot withstand scrutiny, however. First, that total is based on results from state-level exit polls. The extent to which these are flawed is especially evident from the Texas sample which like California includes 25% of all Latino voters. Because of its size, Texas’ results heavily influence national Hispanic tallies. The Texas exit polls were based on interviews in twenty sites, far fewer than the 35 each in Arizona and New Mexico which together include less than 8% of Latino voters. Clearly, the number of interview sites in Texas is too few to include a representative Hispanic sample. Thus, the exit survey results showing that Bush received 59% of the state’s Hispanic vote are not based on reliable data. Reinforcing this conclusion is that Kerry’s support in highly Hispanic Texas counties ranged from 55%-70% or more. As one well known analyst points out, there are not enough probable Hispanic Republican votes in the rest of the state to get to the proclaimed 59% level.[2]
This pattern is replicated across the country. In California, the most egregious case outside of Texas, NEP reports Latino support for Bush was 32%, an increase of 4% over 2000. Given this minimal increase, for Latino support to have reached the total support reported in the NEP for the western states as a whole, Hispanics in the remaining western states ‘must have supported Bush at the rate of 167%’.[3] In other words, either there was massive vote fraud or the exit polls were systematically flawed.
Other pollsters conclude that Hispanic support for Bush remained virtually unchanged from 2000. Zogby International argues that Bush received 33%-38% of the vote. Using a national sample of 5,154 Hispanics interviewed at 136 polling places, the Willie Velasquez Institute found only 33% support for Bush. Ruy Teixera, interpreting all the available data, concludes that no more than 39% voted for Bush.
Supporting this pattern are the results of public opinion polls conducted from July up to the election. The Washington Post/Univision/Tomás Rivera Policy Institute national polls in July and late October national surveys showed Latino Democrats identifiers totalled 66%, compared with 24% Republicans and 60% of likely voters supporting Kerry. Moreover, as Table 1 illustrates, no poll showed Bush receiving more than 34% of the Latino vote. Unlike the case of women voters there were no indications that Hispanics were switching to President Bush.Table 1. Latino Political Preferences in Pre-Election Polls
| Survey Sponsor | Date | Number | Bush (%) | Kerry (%) |
| Bendixen & Associates | 6/3/2003 | 800 | 34 | 48^ |
| Democracy Corps | 2/16/2004 | 1,564 | 34 | 56 |
Miami Herald/Zogby | 4/01/2004 | 1,000 | 33 | 58 |
| Gallup | 6/30/2004 | 500 | 38 | 57 |
| Washington Post/TRPI | 7/16/2004 | 1,605 | 30 | 60 |
| Pew Hispanic Center | 7/20/2004 | 751 | 32 | 62 |
| Democracy Corps | 7/22/2004 | 1,000 | 30 | 61 |
| Bendixen & Associates* | 9/20/2004 | 900 | 29 | 64 |
| Washington Post/TRPI | 10/15/2004 | 1,603 | 30 | 60 |
| Miami Herald/Zogby | 10/24/2004 | 1,000 | 33 | 61 |
| 2004 Survey Average | 2/04 – 10/04 | 9,923# | 32 | 60 |
Moreover, the Washington Post/Univision/TRPI polls show fewer than 35% of Latinos in all standard demographic categories supported Bush (see Table 2). The only exceptions are non-Catholics and born-again Christians, which are small non-standard categories.Table 2. Latino Vote Preference by Demographic Characteristics
| Kerry (%) | Bush (%) | Undecided (%) | N | |
| Total | 59.4 | 30.0 | 9.4 | 1,603 |
| Likely voters | 60.6 | 29.8 | 8.6 | 1,390 |
| Less HS | 63.5 | 23.5 | 10.3 | 489 |
| HS Grad | 57.9 | 32.8 | 8.2 | 369 |
| College | 58.6 | 33.9 | 6.4 | 664 |
| Less 25K | 59.5 | 29.6 | 9.3 | 442 |
| 25 – 49K | 63.5 | 28.3 | 7.5 | 402 |
| More 50K | 62.0 | 33.7 | 2.7 | 375 |
| Immigrant | 57.3 | 31.2 | 10.7 | 845 |
| Native – 2nd | 65.8 | 25.9 | 6.7 | 374 |
| Native – 3rd | 60.0 | 31.0 | 7.6 | 361 |
| 18 – 39 | 58.9 | 30.9 | 8.1 | 420 |
| 40 – 59 | 61.9 | 29.7 | 7.4 | 621 |
| 60 and over | 59.4 | 29.9 | 10.3 | 493 |
| Catholic | 66.4 | 23.9 | 8.7 | 1,146 |
| Non-Catholic | 38.3 | 51.1 | 8.8 | 329 |
| Born again | 47.0 | 41.7 | 9.5 | 436 |
Source: Washington Post/Univision/TRPI National Survey of Latino Voters, October 15, 2004.
Given these consistent patterns, it is easier to believe that the exit polls mistakenly depict the Latino vote than to accept that Latino preferences could have changed so substantially in such a short period. As Fernando Guerra, a widely respected Hispanic analyst states, ‘there is nothing special that Bush did to get a higher turnout. What would explain this (alleged) tremendous amount of Latino support for Bush?’
There is also no evidence in TRPI’s July and October surveys that Latino fundamentalists and highly religious Catholics could have produced the increase in Bush’s support as described in exit polls. While Latino evangelicals favoured Bush over Kerry by 58% to 33%, and 49% of other Christians also supported him, 68% of Latino Catholics and 62% of non-Christians or seculars supported Kerry. Latino non-Catholic Christians thus made up part of the core of religious voters that contributed to Bush’s victory. However, non-Catholics are only 18% of the Latino electorate. Thus, even with their high levels of support the Latino Christian vote could not have pushed Bush’s overall support far along the road to 44%.
What does all this suggest about the Latino vote in 2004? First, despite claims to the contrary, there is no clear evidence that, overall, Republicans significantly increased their share of the Latino vote. Indeed, the NEP admitted as much when it announced in early December that it was lowering its estimate of Bush’s Latino vote to 40%. This estimate suffered from some of the same flaws as the original estimate, however, and there is no persuasive reason for accepting it. Nonetheless, Republicans seem to have increased their vote in some states but not in others. For example, Republicans apparently increased their support in New Mexico where it is reasonable to conclude that their support helped President Bush carry the state, but their share of Latino votes in Florida seems to have declined.
Second, despite a record high of approximately 7 million voters, the Latino vote did not significantly affect the results of the 2004 election. This was primarily because most Latino voters reside in states won by Democrats. In Texas, Arizona, Nevada and Colorado, Latino majorities voted for Kerry but President Bush carried these states. Only in Florida does it appear that a majority of Hispanics voted for the winning candidate, President Bush, but there his margin of victory was so substantial that he would have carried the state even if he had received no Hispanic support.
The results in New Mexico are unclear. While it seems that Latino voters were crucial to President Bush’s victory, it is unclear what proportion of the Hispanic vote he received. With or without a majority, however, there is little doubt that Hispanic voters played their most important role in this state. While significant, it must be recognised that helping to carry this state added to Bush’s overall triumph but was not crucial to it. That is, he would have won the election even if he had lost New Mexico.
Fourth, it is clear that electoral strategies designed around the structure of the Electoral College dampen Latino turnout. Candidates have no incentive to campaign in states that they are certain to win or lose. All too often, this includes those states with the largest concentrations of Hispanics. Such strategies not only reduce their role in presidential elections, but by marginalising Latinos from presidential campaigns they also deprive them from gaining the experience needed to develop national political operatives and a mobilised electorate.
In 2004, the result of such campaigns contributed greatly to the confusion regarding increased Latino Republicanism. Given that neither party had an incentive to mobilise Hispanics in California, Texas, New York or Illinois, where 60% of Latino voters reside, there is at present no data on whether those Latinos who turned out were representative of all Hispanic registered voters or whether they were disproportionately Republican. Until that issue is clarified, any claims regarding Latino electoral preferences in 2004 are incomplete at best. In the absence of these data, one conclusion that seems clear is that the Latino vote in 2004 maintained its well-established Democratic commitment.
An equally clear second result is that the Latino vote is a major component of the Democratic core. This means that if the Democratic Party is to regain electoral supremacy, it will have to build on its Latino foundation. This will be especially significant in states with well established but rapidly increasing Hispanic electorates like Arizona, Nevada and Colorado where mobilised Latino voters can take advantage of a divided Anglo electorate to carry the state for the Democrats. It will also be important in key states such as Michigan and Pennsylvania, to name just two, where a small Hispanic population may have great electoral influence because the Anglo population is so evenly divided between the two major parties. To reach these voters will require more than symbolic outreach, however. It will necessitate long term investments in immigrant naturalisation programmes, voter registration drives and get-out-the-vote campaigns as well as the development of a policy agenda that is attentive to Latino concerns.
Rodolfo de la Garza
Eaton Professor of Administrative Law and Municipal Science, Department of Political Science. Columbia University
[1] Dick Morris, ‘The Hispanic Vote Elects Bush’, NewsMax.com, November 5, 2004.
[2] Ruy Teixera, ‘44 Percent of Hispanics Voted for Bush?’, Alternet: Election 2004, December 15, 2004.
[3] Teixera, loc. cit.
