November 12, 2004
Why Colorado?

Colorado Luis takes a look at the numbers which show an improvement by President Bush in Hispanic voting preference, and asks why it is that Hispanics in Colorado went so much more heavily for John Kerry. He also links to this article, which says that Bush may have gotten up to 59% of the Latino vote in Texas.

But among the dozens of numbers produced by the national exit poll, perhaps none were more surprising than the Latino totals for Texas. Bush, the poll concluded, earned 59 percent of their votes, a 16 percent jump from the same poll's 2000 number.

Latino-voting experts agree Bush did better in their communities than he did in 2000. In the lower Rio Grande Valley, for instance, Kerry won Hidalgo County 55 percent to 45 percent and lost narrowly in Cameron County. Both have Hispanic populations exceeding 80 percent.

But if Bush actually did claim almost 60 percent of the Latino vote statewide, his overall margin over Kerry in Texas should have been closer to 70 percent, not the final 61 percent to 38 percent, [Antonio Gonzalez, president of the Willie C. Velazquez Institute, which researches Latino voting patterns] said.

I've taken a look at the election returns in the most heavily Hispanic counties in Texas, and I think we can safely dismiss the notion that Bush got 59% of their vote. I've put together a spreadsheet which show the result. There's no question that Bush did much better in these counties than he did in 2000 (more on that in a second), but the best showing he got in any of them was 59%, which he achieved in Val Verde County where fewer than 12,000 votes were cast. Throw in the fact that he topped out at 54% in Bexar, Harris, and Dallas Counties, and it's hard to see how Bush could have come close to that 59% mark without assuming a big dropoff in his support among white voters. Since there's no evidence of that, I conclude his support in Texas among Hispanics was less than 59%, and probably less than 50% though we can't say that for sure.

As I said, Bush's performance in these counties was decidedly better than in 2000. He increased his share of the two-party vote from 39.99% to 44.54% while improving his percentage in every single county, added over 46,000 votes to his total to 15,000 for Kerry/Gore (almost all of which came from El Paso and Webb Counties), and carried three counties that he didn't in 2000 (Val Verde went his way both times). If Texas is at all representative, then he had to have improved on these numbers nationwide. The NDN blog looks at some of the nationwide numbers, and they ought to provide a wake-up call for Democrats. Especially read point #4, since people were sounding the alarm about this well before the election.

I should note that there is one bit of good news for Texas Democrats in the 2004 numbers: Bush's improvement among Hispanics did not appear to have any coattails. I spot-checked the three statewide races in each of these counties, and Bush outperformed the other Republicans everywhere. That includes in the Railroad Commissioner race, where the Republican incumbent is Victor Carillo (the Libertarian candidate, Anthony Garcia, did pretty well in these counties, and in some cases his vote plus Carillo's slightly exceeded Bush's, but even in those cases, Democrat Bob Scarborough did better than Kerry did). You can check for yourself here, and recall that I've already shown that Bush generally outperformed the statewide Republicans in the big urban counties as well.

Luis gives his thoughts as to what happened in his state, and there's a lesson that we in Texas need to pay heed to:

I think the lesson for Democrats is that there has to be a visible presence of respected Latinos in the upper leadership in the party in order to achieve the overwhelming majorities Dems used to get back when, frankly, only the most politically active Latinos went to the polls. There has to be real visible integration, not just "reaching out" by non-Latinos or simplistic finger-pointing at Tancredo-style race baiting, which is on the decline on the GOP side anyway. With Ken Salazar as the party leader, we've achieved that in Colorado, but I'm not seeing it elsewhere in the southwest.

There's been a lot of discussion here and elsewhere about who we want to see run for what in 2006, with the various unelected Congressmen taking leading roles, but much of what I've seen (and written myself) is conspicuously short on Hispanic candidates. Let's not forget that one former Congressman as of January will be Ciro Rodriguez, and he ought to be given as much consideration for various spots on the ticket as anyone else, assuming he doesn't go for a rematch with Henry Cuellar in CD28. I'm not advocating for a repeat of 2002's "Dream Team" concept, I'm just saying that we can't afford to overlook qualified people. It's more important than ever now.

UPDATE: Big Media Matt thinks the 59% number may be plausible. I think the problem with his math is that he is overstating Hispanic turnout. Twenty-three percent would be an all-time record in Texas - I believe Hispanics comprised something like 12% of the electorate in the 2002 statewide elections. If you assume 13% are Hispanic, and 73% are Anglo, then Bush getting 45% of the Hispanic vote, which is entirely consistent with what I found in the heavily Hispanic counties, brings him to about 60.33% of the vote total, which is pretty close to the actual result. Getting 50% of the Hispanic vote, which would mean Bush did better in the big urban counties than elsewhere (not a very big stretch) would put him right at 60.98%, which is even closer. I stand by my assertion that 59% is too high a figure.

Posted by Charles Kuffner on November 12, 2004 to Election 2004 | TrackBack

I don't know that racebaiting is on its way out.

Prop. 200 has spawned a host of copycat movements in other states.

Posted by: julia on November 12, 2004 3:28 PM

I'm suspicious of using exit polls to calculate things like Hispanic voting preferences. They simply aren't designed for that purpose. I've seen this discussed elsewhere but I don't have the link. The problem is this.

Exit poll locations are chosen for their predictive value in determining the winner in the general election. As such, they are looking for "bellweather" types of precincts that have good predictive value over the years. These turn out to be a lot of suburban swing type of precincts.

Now a lot of Hispanics have been moving to the suburbs and show up in these sorts of exit polls. However the real question is whether upwardly mobile Hispanics that have moved to the suburbs are really representative of the Hispanic vote overall. The fact of the matter is that upwardly mobile Hispanic voters in the suburbs are more likely to lean Republican than Hispanic voters in traditionally Hispanic neighborhoods in the inner cities and border areas.

So the question I would have is this. Were the exit poll numbers part of a study that was designed to accurately sample the Hispanic population as it is distributed throughout the US? Or did they simply take whatever Hispanic results they got from sampling in the suburbs and then extraplolate to the entire Hispanic vote? I suspect it was the latter given that the primary objective of the exit polls is to predict the winner of the election.

Posted by: Kent on November 13, 2004 12:16 PM

The national exit-poll is supposed to be useful for exactly this sort of analysis; yet it is clearly off in the Latino voting percentages. They originally oversampled the Hispanic vote in Texas, by more than two-fold. At the same time they had more than five points of Kerry support to eliminate; if you take out that much of Kerry's share of the Latino respondents, the result would be that Bush's share of the Hispanic vote is falsely made to rise 12 or more points. To approximate the truth, they would have to now also deduct, from the Bush column, respondents sufficient to bring the Hispanic share down to its correct percentage.

Posted by: John S Bolton on November 13, 2004 8:44 PM