By Mark Mellman - 02/23/05 12:00 AM EST
Imagine the surprise!
Before November, public polls showed Sen. John Kerry with 65 percent of the Latino vote. “Kerry is winning the Latino vote by a margin greater than Al Gore’s,” one analyst proclaimed.
Then, on election night, the exit polls showed Kerry getting nine points less than Gore had among Latinos. (The Kerry campaign’s internal polling had revealed consistent sub par performance among Latinos.)
Shortly after the election, I wrote here that for too long Democrats had considered Latinos part of the base, failing to acknowledge changes and contradictions in their political views. For example, we found in 2002 that while Latinos identified as Democrats, unlike other partisans, they bore relatively little ill will toward Republicans — a dangerous situation for us.
A couple of weeks later, the Congressional Hispanic Caucus took an even stronger position, urging that the community no longer be treated as part of the base at all.
At the time, controversy swirled around the numbers. The NEP exit poll said just 53 percent of Latinos voted for Kerry, while the Velasquez Institute poll said it was 68 percent.
Now, with the full exit-poll results in hand, the data speak more clearly. Democrats did lose meaningful support among Latino voters. But our problem is concentrated with the relatively assimilated, English-dominant and bilingual segment of the community.
It was quickly apparent that the NEP poll had fallen prey to “clustering bias”. Most of the Hispanic interviews in the national poll were from just a few precincts.
Combining the 51 separate state exit polls provides a more accurate portrait of the Latino vote. Applying NEP’s weighting system to these data yields a national vote of 58 percent Kerry, 40 percent Bush. Our own analysis of their data suggests it was 57 percent to 41 percent. Thus, the decline in Latino support was a less precipitous, but still significant, four to five points.
While polls still differ on the precise level of Kerry’s Latino support, they all converge on a decline of four to seven points compared to 2000. Over a longer period, the declines are even steeper. Dukakis garnered 65-69 percent of the Latino vote, a far cry from Kerry’s 57 percent.
Both the problems facing the pre election public polls and the challenges confronting Democrats can be understood clearly by examining the underlying data.
Consultants who specialize in Latino politics have long been directing Democrats’ attention to Spanish-dominant recent immigrants. Those are the easiest voters to poll, but, important as those voters are, they are not where the problem seems to be.
Latinos who voted for Kerry in very large numbers tended to be poorer, Spanish-speaking and living in Latino neighborhoods. Those less likely to have voted for Kerry include better-off, English-dominant and bilingual folks who live mainly in more diverse neighborhoods.
According to the Annenberg polling, the decline in Democratic support among Latinos came almost entirely in English-language interviews. Kerry did just one point worse than Gore had among Spanish-dominant voters.
Our analysis of the NEP data demonstrates that 70 percent of Latinos in heavily Latino neighborhoods voted for Kerry, compared to just 52 percent among the quarter of the Latino electorate living in precincts that are less than 5 percent Latino.
Religiosity plays a role in this community, as in nearly all others. Kerry’s support was 26 points higher among Hispanics who go to church more than once a week, than among those who rarely attend.
Recapturing the enthusiasm of the Hispanic community is a central task for Democrats. To be successful, we must first admit we have a problem and locate it with precision. Only then will we be able to develop the strategies and tactics to stanch the losses.
Mellman is president of The Mellman Group and has worked for Democratic candidates and causes since 1982, including Sen. John Kerry (D-Mass.) last year.