Politics & Policy

Bush’s “Real” Hispanic Numbers

Debunking the debunkers.

On November 3, the National Election Pool (NEP) reported that George W. Bush received 44 percent of Hispanic votes cast on November 2: a 9-percentage-point gain from his 35-percent performance in 2000. Since then, these figures have been widely challenged, and internally revised.

Ruy Teixeira of the Century Foundation estimates that 39 percent of Hispanics voted Republican. NBC, a part of the NEP pool, now calculates Bush’s Latino vote at 40 percent.

Immigration critic Steve Sailer of VDARE acknowledges a 3-4 percent Bush increase, but doubts its significance. Bush’s Latino gains, he writes, “confirm the general pattern that the Hispanic vote for Republicans rises and falls in the same cycles as the white vote–just consistently more Democratic.”

So did Republicans make progress among Latinos in 2004? A closer examination of the NEP numbers shows that Bush’s gains among Hispanics, although lower than initially estimated, were both real and significant. Specifically, the GOP’s increased Latino vote share offset the potential Democrat advantage from a hefty increase in Hispanic registration and voting.

But talk of a GOP “average gain” masks how that gain was distributed. In states where conservative 527 groups, such as Council for Better Government and Hispanics Together (both of which I served as a consultant), ran intensive campaigns on Spanish-language media, the president’s Hispanic vote share increased sharply. In states where no such effort occurred, his Latino vote share improved hardly at all.

What is a reasonable estimate of Bush’s performance among Hispanics?

The most significant internal correction by NEP pollster Edison Media Research and Mitofsky International involved Bush’s Hispanic vote share in Texas. The NEP recalculated the percentage of Latinos in its Texas sample from 23 to 20 percent, and the pro-Bush Latino percentage from 59 to 49. Texas is home to 19 percent of the nation’s Hispanics. This revision alone lowered Bush’s national numbers 2-percentage points.

The NEP survey was by far the largest post-election poll, dwarfing those of the Los Angeles Times, the New York Times, and the Velasquez Institute. Over 13,600 respondents, including 1,100 Hispanics, completed the “long form” on which the initial Hispanic numbers were based. An additional 62,600 respondents, including 3,600 Hispanics, completed the “short form,” on which the NEP based its state exit polls.

The NEP state exit polls, as published on the CNN and MSNBC websites, broke out Hispanic numbers in 20 states: That is, they gave the total number of subjects interviewed, the percent of those who were Hispanic, and the percent of Hispanics who voted for Bush and for Kerry. From this, one can calculate the number of Hispanics surveyed in each of the 20 states, and the numbers who voted for Bush and for Kerry in each.

These 20 states account for 91 percent of the nation’s Hispanic population. Respondents, including 3,586 Hispanics, completed 35,891 short-form surveys. Within this 20-state set, George W. Bush won the votes of 41.28 percent of Hispanic respondents polled, compared to 57.47 for John Kerry. But the NEP sampled battleground states more heavily than non-battleground states. Rebalancing the Hispanic totals for Bush and Kerry state-by-state to reflect the Hispanic population in each relative to the total Hispanic population for the 20-state set, Bush won 38.07 percent of the Hispanic vote, compared to 59.67 for Kerry.

How significant was this gain? To political professionals, a gross three-point shift in vote share will translate into a 5 or 6 net vote shift per hundred cast. It is by the sum of such shifts, demographic group by demographic group, that elections are lost or won.

In 2000, the Voter News Service (VNS), predecessor to the NEP, reported a Gore victory among Latinos of 62-to-35 percent, or 27 votes per 100. In 2004, the NEP 20-state set, as balanced above, finds Kerry winning 59.67 percent of Hispanic votes, compared to 38.07 percent for Bush–a 21.6-vote advantage per 100.

In 2000, roughly 6 million Hispanic votes were cast. At 62-to-35 percent, this gave Gore a 1,620,000-vote victory over Bush among Latinos nationwide.

In 2004, Hispanic turnout rose by roughly 25 percent, to 7.5 million voters. Had the Democrats held their 2000 margin, their national advantage among Hispanics would have grown by 405,000, to 2,025,000. The improved Bush percentage in 2004 nullified this gain completely, holding the Democrat advantage in the Hispanic community to the same 1,620,000 as in 2000. The impact of a massive and successful Democratic voter registration drive was nullified.


But the 20-state set hides, rather than reveals, the factors that determined Republican progress (or lack thereof) among Hispanic voters. In Arizona, Nevada, Colorado, Florida, New Mexico, and Wisconsin–six battleground states, containing an NEP sample size of 1,768 Hispanic voters–conservative 527 groups ran a vigorous 12,000-spot broadcast campaign on Spanish-language media in support of the G.O.P. ticket. In the remaining 14 states, containing an NEP sample size of 1,818 Hispanic voters–no such campaign aired.

The difference in results was night-to-day. In the states where the conservative Spanish-language 527s were active, Bush carried 47.17 percent of the population-weighted Latino vote, compared to 52.25 percent for Kerry. In the 14 states where the 527s were inactive, Bush won 35.96 percent of the Hispanic vote, Kerry 61.65 percent–a result almost unchanged from Gore’s 62-to-35 percent advantage of 2000.

Now, the 6 states with ads include Florida, traditionally the most Republican of the Hispanic populations. This “prior disposition” of Florida Hispanics obviously skews the 6-state set. Is there some other way to compare Bush’s performance in program and non-program states?

Yes. We can examine Republican vote-share changes on the margin.

Exit-poll pool member CNN lists changes in Bush’s vote in 4 of the 6 “program states” on its website. Comparing the 2004 NEP state exit polls to their 2000 VNS counterparts, Bush’s Hispanic share grew 5 points in Colorado, 7 in Florida, 9 in Arizona, and 12 in New Mexico. The sample size in these four states is still a healthy 1,503. Weighting these states by Hispanic population, so that the GOP’s 12-point increase in New Mexico is not treated equally with, say, Florida’s 7-point increase, the weighted GOP increase in the 4-state sample is 7.60 percent–a shift of better than 15 votes per hundred for President Bush.

This performance was broadly consistent with the 2002 results of similar pro-GOP Spanish-language broadcast campaigns, which saw top-line Republican performance in Senate and gubernatorial races increase 6.14-percentage points.

In other words, when Republicans have aggressively courted Hispanic votes, they have won them. And when we haven’t, GOP Latino vote-share numbers have barely budged.

Almost all conservative progress on minority media has been generated in the 527 world, by party irregulars. The official organs of the Republican party have been missing in action, producing either weak product or no product for mass-minority audiences.

What would have happened had the Republican party and its fellow travelers campaigned aggressively nation-wide on Univision, Telemundo, and Spanish-language radio?

Extrapolating from results in the 527 program states, Kerry’s projected majority of 2,025,000 votes could have been reduced by nearly half, to 1,035,000: a one million vote swing in a voting population of roughly 7.5 million.


Steve Sailer and his friends in the anti-immigration lobby regularly criticize efforts to woo Hispanic voters two ways. First, they contend that the Hispanic vote “tracks” the white vote anyhow, making such efforts superfluous. Second, they contend such efforts constitute a sell-out to advocates of open borders and mass immigration.

The Hispanic vote does not track the white vote. Above, we’ve seen that the Hispanic vote “tracks” what’s happening on the ground in particular places, not some macro-trend. But even in its generalized (i.e., irrelevant) form, this assertion is untrue. Look at the trends in white and Hispanic voting for Republican presidential candidates over the last 28 years in the New York Times exit polls:

1976-1980: White vote +4, Hispanic vote +9

1980-1984: White vote + 8, Hispanic vote +4

1984-1988: White vote -5, Hispanic vote -7

1988-1992: White vote -19, Hispanic vote -9

1992-1996: White vote +6, Hispanic vote -4

1996-2000: White vote +8, Hispanic vote +10

Over the entire period, the average variation between changes in white and Hispanic voting was 5.5 points per election. To an election strategist, these figures signal wild variation. Any professional looking at those numbers would seek independent variables to explain them. And the first thing he’d examine is differences in the communications that the two groups are receiving.

The second assertion–that Spanish-language campaigns are a sell-out to advocates of open borders–is a non sequitur. The ads of Republican 527s could have been so designed. In fact, they were not. In 2002, the Council for Better Government campaigns featured 15 scripts, not one of which dealt with immigration. In 2004, two scripts of 32 aired by the council and Hispanics Together broached the subject, defending the president’s policy. But the overwhelming preponderance of the spots dealt with traditional conservative issues. Scripts advocated traditional marriage, tax breaks for families and small business, school choice, military preparedness, the right to life, personal savings accounts, and faith-based social-service delivery. Given the steeply rising percentage of Hispanics who supported George Bush in the “527 program” states, one may conclude that conservative issues gained considerable traction among them.

The complaint conservatives lodge against open immigration is that it fosters the balkanization of our nation, creating enclaves of “hyphenated Americans” ideologically isolated from our values, but parasitically attached to our pocketbooks. But whether our borders are thrown wide open or slammed tightly shut, it is hard to see how conservatives, by ignoring 7.5 million Hispanic voters, will make them less balkanized, or less liberal.

Richard Nadler is president of America’s Majority, a not-for-profit dedicated to building the demographic base of the conservative movement.


The Latest