Latino Voting in the 2006 Election

Realignment to the GOP Remains Distant

By James G. Gimpel on March 1, 2007

Download this Backgrounder as a pdf

  • National exit polling indicates that House Republicans received about 30 percent of the Latino vote in 2006 — down eight points from the 38 percent GOP House members received in the previous midterm elections in 2002.
  • But 82 percent of the voters in the hotly contested 2006 House elections were non-Hispanic white, and all indications are that these are the politically changeable voters, not Latinos, whose political preferences exhibit stability.
  • Only in the South did Latino support for the GOP remain unchanged — a stunning irony given the pervasive Southern support for border control, immigration restrictions, and generally prevalent conservatism — viewpoints commonly thought to be inimical to Hispanic political interests.
  • The source of the Democratic bias in Latino voting is clear: For every Latino voter living in a solid Republican county, there are 2.9 Latino voters living in places that lean Democratic or are overwhelmingly Democratic.
  • Very few Latino immigrants wind up settling in heavily Republican counties – only 13 percent, compared with the 60 percent who are residing in areas where they are in contact with one-sided Democratic populations.
  • If the path to Republican Party identification is paved by upward economic mobility, there would be many more Latino Republicans if these last 30 years had not witnessed record levels of unskilled immigration.

The returns from the 2006 election appear to have quelled hopes or fears among some strategists that Latinos were in the midst of a political realignment to the Republican Party. National exit polls indicated that House Republicans received about 30 percent of the Latino vote in 2006 — down 10 points from the 40 percent President Bush received in 2004, and the 38 percent GOP House members received in 2002. Reliable estimates place the Latino support for Republican U.S. Senate candidates at 35 percent, and support for GOP gubernatorial candidates as high as 37 percent.

There was, however, substantial variability in the Latino vote across the nation, typically ranging from the high-teens to the low-40s. Incumbent U.S. Senators such as John Ensign (NV) and John Kyl (AZ), crossed the 40 percent threshold, for example. Gubernatorial candidates Charlie Crist (FL), Jodi Rell (CT), and Arnold Schwarzenegger (CA) received solid shares of the Latino vote, although given Florida’s Cuban population, Crist’s support should not be considered a surprise. Most GOP challengers did very poorly among Latinos, including Tom Kean, Jr. in New Jersey (28 percent), Mike McGavick in Washington (25 percent), and John Spencer in New York (24 percent). What accounts for this variation in GOP candidate performance among Hispanic voters?

Scrutiny of the surveys for individual races suggests that, for the most part, Republicans did well among Latinos only when their contests were very uneven from the early stages (e.g., Schwarzenegger, Ensign) or, in other words, when they had a substantial edge among all voters, usually owing to the powerful advantage of incumbency. Latinos are no different from other voters — being just as alert and attentive (or inattentive) as anyone else. They sense a surge in support for a candidate, and know full-well when a candidacy is hopeless. Where Republicans ran far behind well-financed incumbents (e.g., McGavick in Washington, Spencer in New York), the Hispanic population was stalwart in its support of Democrats.

Mismatched contests do not provide a firm indication of changing party loyalty, however, since usually they pit a well-recognized candidate against one whom few know and large majorities oppose. Going with a tidal wave that swamps a poorly financed challenger is not the same as shifting one’s basic political views as a result of a party’s outreach efforts.

Close examination of the most competitive contests indicates that Latinos fell in behind Democrats by a greater than 2:1 margin, the way they have consistently for an entire generation (de la Garza and Cortina 2007). Nowhere was this more evident than in Rhode Island, where Latinos supported successful Democratic candidate Sheldon Whitehouse by a margin of 77 percent to 22 percent over Republican Lincoln Chafee. But the winning Republican gubernatorial candidate, Don Carcieri, did almost as poorly among Latinos, garnering just 24 percent of their votes, according to exit polling.

Regional Variation in Latino Support

In balloting for the U.S. House, the national exit poll exhibits notable regional disparities, with Hispanic support for Republicans, ranging from a low of 18 percent in the Northeastern states, to a high of 41 percent in the South. In the West, home to large concentrations of Latino voters, support for GOP House members stood at only 27 percent, compared with 48 percent received from non-Latinos, a stark 21 percent difference (see Table 1). In fact, looking across all four regions, a formidable 20-point gap separates Latinos from non-Latinos in their support for the Republican Party.

Table: Voting in the US House Elections 2006 and 2002

Certainly compared to the 2002 off-year elections, Latino support for Republican candidates was a major disappointment for the believers in Hispanic realignment. Compared to the exit polls from 2002, the 2006 mid-terms show Latino support for Republicans dropped in every region, and especially in the Midwest (44 percent in 2002 to 30 percent in 2006; down 14 points) and northeast (41 to 18 percent; down 23 points). Only in the South did Latino support for the GOP remain unchanged (40 to 41 percent) — a stunning irony given the pervasive southern support for border control, immigration restrictions, and generally prevalent conservatism — viewpoints commonly thought to be inimical to Hispanic political interests.

What we do not know from the exit polls is just how turnout dropped among Democratic-leaning Latinos in 2002 or how much it jumped in 2006. Much of what passes for volatility in party support from election-to-election arises not from persuasion or attitude change, but from variations in who shows up to vote among those committed to one side or the other (Nuño 2007). Party identification is highly resistant to change (Green, Palmquist and Schickler 2002), contrary to the sales pitches of consultants who have an enormous financial stake in duping candidates into believing that voters’ choices can be easily manipulated. But even though it is exceedingly difficult to move voters through outreach and advertising, we do know that turnout varies wildly from year-to-year and across states and districts.

By now, it is well known that turnout drops in mid-term elections — this surge-and-decline of peripheral voters into and out of the electorate has been observed for decades. The off-year decline in political participation does vary in size, however. The more modest gaps between Latino and non-Latino support for the GOP in 2002 may well have been a function of lower turnout among Hispanic Democrats, and/or relatively higher turnout among Hispanic Republicans that November. This differential could have easily been reversed in 2006, producing major Democratic gains, including the loss of Henry Bonilla’s House seat (TX-23) to Ciro Rodriguez in a December special election. Scattered anecdotes from field observations of a number of 2006 campaigns suggest that the Democratic Party contacting operation was more successful than the Republican effort, and that it was vastly improved over 2002.

Evidence from the Cooperative Congressional Election Study

For further insight into the voting of Latinos in the 2006 contests, we can examine recently available data from the 2006 Cooperative Congressional Election Study (CCES), a nationwide survey of voters and non-voters interviewed both before and after the general election.1 Given its very large sample size, the CCES is well-suited for the breakdown of the political inclinations of ethnic minority groups, including Latinos and African Americans, whose views are commonly underrepresented in traditional polls.

CCES tabulations for Latino political preference in the 2006 mid-term election are broadly consistent with the national exit poll figures (see Figure 1). Republicans won an estimated 33 percent of the Latino vote in U.S. House elections, 36 percent in the U.S. Senate races, and 37 percent in the nation’s gubernatorial elections. White non-Hispanics were much more supportive of GOP candidates, though not quite by the 20-point margins indicated by the exit polling (see Figure 1). This is likely because the exit polls represented the political battleground states disproportionately, while the CCES is more representative of the nation as a whole.

Graph: Latino and NonLatino White Support for Republicans for Governor, US Senate, and US House in November 2006

In the most highly visible races of the year, however, Latino voting was commonly more one-sided in favor of Democrats than in the less-competitive elections, exhibiting the kind of split we have already noted about the Rhode Island Senate race.2 Through the vigorous mobilization of traditionally Democratic blocs, the turnout of Latino voters jumped, bringing more Latinos to the polls — which automatically means turning out more Democrats. The resultant imbalance in Latino voting verifies previous research indicating that Republicans perform better, on average, when Latino turnout remains low (Gimpel 2003).

In the non-competitive governors’ races, for example, 56 percent of Latinos supported Democrats, but this margin jumped to 64 percent in the 10 most hotly contested elections, the result of both higher levels of awareness among voters, as well as the get-out-the-vote efforts of determined Democratic campaigns. Democrats managed to turn competitive gubernatorial contests into blowouts favoring their side when it came to Latino

In the U.S. Senate races, the evidence that it was specifically strong competition that led to the most lopsided Latino voting is less convincing, because in these higher-profile races, Latinos heavily supported Democrats regardless of the heat carried by the local political winds (see Table 2).

Graph: Latino and NonLatino White Voting in Competitive and Less Competitive Governor, US Senate, and US House Elections 2006

The House tabulations in Table 2 must be judged understanding that gerrymandered districts create safe seats for many incumbents by blocking Latino and other ethnic minority voters into minority-majority or minority-plurality districts. Competitive districts occur in areas where there are few Latino voters in the first place. Only 10 percent of the nation’s voting Hispanics lived in competitive House districts in 2006, and even a smaller share, 7 percent, of the country’s voting Latino immigrants lived in these districts.3

Nevertheless, what we see in tabulations across House districts is that Latino support for Democrats was 61-62 percent regardless of the level of local political competition. By contrast, the non-Latino white vote was evenly divided in the closest races, and slightly favored Republicans in the less competitive districts (see Table 2). Latinos weren’t the swing voters in House elections — non-Hispanic white voters were! Previous research has noted that Latinos do not reside principally in presidential battleground states (Gimpel 2004). The same is true of the mid-term 2006 toss-up congressional districts. Latino voters did not bring down the Republican majority in the U.S. House given their small numbers in the contested districts. But their loyalty to the Democrats across the board certainly did Republicans no favors in contests where the outcome was in doubt.

The story these figures tell is a painful reality for Republicans expecting political realignment. Apparently, Latinos are not a swing voting bloc that can be swayed by ordinary campaign appeals. Fluctuations in the percentage of support for the two parties are a consequence of who shows up to vote. Only in the circumstances where Republicans were clearly in command of the election from the beginning could they count on a respectable share of the Latino vote, and these improved margins were likely due to low Latino turnout, not mass conversion of Latinos to the GOP.

Local Context of Latino Voting

Students of political behavior have frequently made the argument that political decision-making among voters involves a powerful contextual component (Huckfeldt and Sprague 1995; Gimpel 2003; Gimpel and Schuknecht 2003; Gimpel and Cho 2004). Voters’ political thinking is shaped by those around whom they live and work, not solely by the blunt force of national news and political advertising campaigns. No one experiences a campaign unmediated by local events, experiences, and discussion partners that shape what is read, seen, and remembered (Mutz 2006).

If there is no active Socialist Party in the area where I settle, I meet no Socialists around the places I live and work, and there are no Socialist candidates who are viable, I am unlikely to become a Socialist, even if I see Socialist advertising occasionally on an evening newscast. Some political views and identities receive social reinforcement, while others are extinguished, and the trajectory toward extinction or further articulation is influenced by our most daily routines.

Rather than think of Latinos, Asians, or other groups as uniform or monolithic voting blocs, researchers and pollsters might do well to understand the local variation in political habits and experiences that shape behavior and attitudes (Gimpel and Cho 2004). This is understandably an unpopular position among pollsters who want to study political behavior cheaply, selling their ill-informed clients on overly simplistic models of turnout and persuasion that are never critically evaluated. A contextually sensitive campaign, by contrast, would place far greater emphasis on field operations, volunteer recruitment and retention, voter mobilization, and make some effort to account for the various settings in which actual behavior takes place. To do this, far fewer polls would be commissioned, and they would include far more respondents. Such a campaign resides a long distance away from the highly centralized and overly professionalized campaign of recent times.

If Latino voting is sensitive to local context, as I have suggested, we should observe significant differences across this population in the 2006 election according to the political orientation of the places in which these voters are embedded. We might expect Latinos to tilt most lopsidedly Democratic in counties that exhibit a traditional Democratic bias, whereas we would expect a more evenly divided Latino population in locations where they find themselves mainly among Republicans. Where people settle determines, to a surprising extent, the political lives they come to live.

Another intriguing angle is the capacity to evaluate Hispanic immigrants vs. non-Hispanic immigrants with the CCES study. Given that the vast majority of Latinos, and Latino immigrants, settle in specific kinds of settings — often coethnic neighborhoods where preexisting networks exist and housing and work opportunities can be readily found — we should be in a position to suggest what kind of political leanings future immigrants may develop as a consequence of their typical residential settlement patterns.

Table 3 examines the Republican and Democratic support among Latino voters by the past political bias of the county in which these voters reside.4 The results are exactly what we would predict based on a contextually nuanced understanding of political behavior. Latino political preferences varied in 2006 according to the locations where they were found to be living. The most Democratic-leaning Latinos were found to reside in those counties that were in the highest quartile of Democratic support. In the "High-Democratic" counties, Latinos cast 77 percent of their votes for House Democratic candidates, 73 percent for Senate Democratic candidates, and 63 percent for Democrats running for governor (see Table 3).

Graph: Latino Voting Governor, US Senate, and US House Elections By Partisan Political Context, November 2006

This is to be contrasted with the "High Republican" quartile of counties, wherein Latino voting was much more evenly divided, though still showing an unmistakable Democratic bias (see Table 3). In the nation’s "reddest" counties, we find that Latinos gave 47 percent of their votes to Democratic gubernatorial candidates, 54 percent of their votes to Democratic Senate candidates, and 52 percent of their votes to Democratic House candidates. Even when living among sizable Republicans majorities, then, the Democratic support among Latinos remains quite robust.

The difficult challenge for Republicans, however, is that very few Latinos are flowing into these highly Republican areas, where they might pick-up a taste for the Grand Old Party’s politics. For every Latino voter living in a solid Republican county, there are 2.9 Latino voters living in an area that leans Democratic or is overwhelmingly Democratic. By any accounting, that’s a very unfavorable ratio for the GOP, short or long-term.

The undeniable truth is that most Latinos reside in locations that are thoroughly monopolized by Democratic Party operations — places that have been the residence of Democratic voting populations for decades, even generations — typically in dense urban counties, or lower-income, less-urbanized locales where there is ample work to be found in low-skill labor market sectors. These movement streams dictate encounters with a certain style of politics, and rarely is it the Republican brand.

Voting Patterns of the Latino Immigrant Population

While it is certainly true that new waves of immigrants are moving into areas not typically settled by previous immigration streams — for example, in Georgia and North Carolina, and in some far-flung suburbs — these rivulets remain numerically small compared to the brisk flows that have rapidly colonized larger cities and older suburbs. The only political ropes to learn in the locales with the largest Hispanic populations are those woven and tied by Democrats. This is unlikely to change anytime soon, although the trends are surely worth monitoring.

Of course what may be true of Latino settlement patterns in general, may not be true of Latino immigrants (DeSipio and Uhlaner 2007). Fortunately, the CCES can shed light on the foreign-born population as a fast-growing subset of the nation’s large Hispanic population.

Without question, the figures from the CCES on immigrant political support should hearten the heralds of a coming Democratic majority. In the most Democratic counties, immigrant Latinos vote as much in favor of the Democratic Party’s candidates as Latinos more generally — by a better than 2:1 margin (see Table 3). Latino immigrants in the most Democratic counties gave Democratic gubernatorial candidates 67 percent of their votes, Democratic U.S. Senate candidates 70 percent, and House Democrats 73 percent. The voting of non-Hispanic immigrants also exhibits a Democratic bias, although it is not quite as imbalanced (see Table 3).

In the most Republican locations there is some attenuation of the Democratic bias as the GOP was able to win about half of the immigrant votes. But, once again, there are very few Latino immigrants who wind up settling in the most heavily Republican counties — only 13 percent, compared with the 60 percent who are residing in areas where they are primarily in contact with Democratic populations. That Latino immigrants are evenly dividing their votes in locations that commonly cast 70 percent of their vote for GOP candidates is a questionable sign of success.

What is suggested by this short research note is far from new. Where people settle is of enormous consequence for the politics that they encounter, learn, and adopt. This is because people do not apprehend political information from mass media without filtering those messages through a set of meanings they have acquired from the social psychological forces at work in their environment.

More concretely, local candidates and party apparatus are sensitive to local populations, and care greatly about steering them to the "right" side of the aisle. One does not ride in from a Washington consultancy and change entrenched patterns of thought and belief with a foreign-language advertising campaign or a few pieces of direct mail accenting a commitment to a generous immigration policy. This is a hopelessly naïve view of the way people come by their political convictions, though it is a view that continues to make people in the campaign business a great deal of money.


What is the surest way for the Republican Party to make inroads with the growing Latino population? Accelerating the flow of mass immigration is not only the wrong course of action, but it may stand directly opposed to what is necessary to make gains. The evidence suggests that this population is the same as any other — its political interests are differentiated mainly along socioeconomic lines, although there is also some emerging evidence of religious differentiation (Lee and Pachon 2007).

Practically speaking this means that to attract a steady stream of Latinos toward the GOP, continued economic prosperity and upward economic mobility will be important issues of concern. Republicans will make steady gains among Latinos through policies that facilitate Latino economic prosperity, business ownership, and secure employment.

There is no evidence that a more open immigration policy is one of those policies, as there is ample evidence from economics that unskilled immigrants compete in the same labor market niches as unskilled natives, lowering wages and living standards among all unskilled workers (Borjas 2001; 2003). The best course toward the long-term political realignment of the Latino vote may be a less open immigration policy. The share of Latinos voting Republican has remained largely unchanged across three decades, with fluctuations barely exceeding the error margin in most surveys. If the path to Republican Party identification is paved by upward economic mobility, there would be many more Latino Republicans if these last 30 years had not witnessed record levels of unskilled immigration.

Finally, there is some evidence that Latinos themselves understand this, as they commonly cite health care, education, and economic concerns as ranking well above immigration policy in importance to their lives, while expressing surprising ambivalence about increasing the number of immigrants flowing into the United States (Gimpel 2004). The CCES survey indicated that only 7.5 percent of Latino voters ranked immigration as the nation’s most important problem, well behind the war in Iraq (30 percent), terrorism (16 percent), government corruption (11 percent), and health care costs (8 percent). In summary, the Hispanic population brought the same set of concerns to the polls in 2006 that everyone else did.

The 110th Congress appears on the threshold for a vigorous debate about guest-workers, amnesty, earned legalization, employer sanctions, and border security. Human capital policies of this magnitude have sweeping implications for the future political balance of localities, states, and entire regions of the country. Members of Congress on both sides would do well to study these implications carefully and honestly as they ponder the momentous committee and floor votes that lie ahead.

End Notes

1 The 2006 Cooperative Congressional Election Study (CCES), is a survey conducted by Polimetrix supported by a consortium of research universities and leading U.K. polling organization YouGov. The study was conducted between October 27 and November 5, 2006. The CCES is led by Professor Stephen Ansolabehere of the M.I.T. Department of Political Science and includes participants from 34 universities. See for further details.

2 Hotly contested governor’s races occurred in the following states: AK, ID, IA, MD, MI, MN, NV, OR, RI, WI. Hotly contested U.S. Senate races occurred in the following states: MD, MO, MT, NJ, RI, TN and VA.

3 Competitive House districts included one Democratic held seat and 38 Republican held seats: GA-12, AZ-1, AZ-5, CA-11, CO-4, CT-2, CT-4, CT-5, FL-13, FL-16, FL-22, ID-1, IL-6, IN-2, IN-8, IN-9, KS-2, KY-3, KY-4, MN-1, MN-6, NV-3, NH-2, NM-1, NY-20, NY-26, NC-11, OH-1, OH-2, OH-15, PA-4, PA-6, PA-7, PA-8, PA-10, TX-22, VA-2, WI-8, WY-At large.

4 This kind of analysis is not possible with the national exit poll since that poll does not make information available on the specific residences of voters other than the state in which they reside.


Borjas, George J. 2001. Heaven’s Door: Immigration Policy and the American Economy. Princeton, N.J.: Princeton University Press.

Borjas, George J. 2003. "The Labor Supply Demand Curve Is Downward Sloping: Reexamining the Impact of Immigration on the Labor Market." Quarterly Journal of Economics 118: 4: 1335-1374.

de la Garza, Rodolfo O. and Jeronimo Cortina. 2007. "Are Latinos Republicans but Just Don’t Know It? The Latino Vote in the 2000 and 2004 Elections." American Politics Research 35: 2: 202-223.

DeSipio, Louis and Carole Jean Uhlaner. 2007. "Immigrant and Native: Mexican American Presidential Vote Choice across Immigrant Generations." American Politics Research 35: 2: 176-201.

Gimpel, James G. 2003. "Latinos and the 2002 Election: Republicans do Well When Latinos Stay Home," CIS Backgrounder. Washington, DC: Center for Immigration Studies.

Gimpel, James G. 2004. "Losing Ground or Staying Even? Republicans and the Politics of the Latino Vote." CIS Backgrounder. Washington, DC: Center for Immigration Studies.

Gimpel, James G. and Wendy Tam Cho. 2004. "Getting out the Asian-Pacific-American Vote." Campaigns and Elections 25: 6: 44-45.

Gimpel, James G. and Jason E. Schuknecht. 2003. Patchwork Nation: Sectionalism and Political Change in American Politics. Ann Arbor: University of Michigan Press.

Green, Donald, Bradley Palmquist and Eric Schickler. 2002. Partisan Hearts and Minds: Political Parties and the Social Identities of Voters. New Haven: Yale University Press.

Huckfeldt, Robert and John Sprague. 1995. Citizens, Politics and Social Communication: Information and Influence in an Election Campaign. New York: Cambridge University Press.

Lee, Jongho and Harry P. Pachon. 2007. "Leading the Way: An Analysis of the Effect of Religion on the Latino Vote." American Politics Research 35: 2: 252-272.

Mutz, Diana C. 2006. Hearing the Other Side: Deliberative versus Participatory Democracy. New York: Cambridge University Press.

Nuño, Stephen A. 2007. "Latino Mobilization and Vote Choice in the 2000 Presidential Election." American Politics Research 35: 2: 273-293.

James G. Gimpel is a Professor of Government at the University of Maryland, College Park. he can be reached at [email protected]

Topics: Politics