The GOP's Latino base: Real or imagined?

By Edward Schumacher-Matos
Friday, December 3, 2010;

I don't know Lamar Smith, but I feel like I do. The Texas Republican, who is likely to chair the House Judiciary Committee in the next Congress, writes often to disagree with my columns.

I respect Smith for his consistency, especially on immigration. If all congressmen voted their conscience, I suspect that two-thirds of current House members would legalize most unauthorized immigrants in the country. Not Smith.

He seems convinced that we should deport even the youths who came illegally with their parents but later prove their worth to the nation by going to college or joining the military. President Obama and Senate Majority Leader Harry Reid appear intent on forcing a vote this month on the Dream Act that would give these youths a path to citizenship.

But if Smith isn't cynical, he does engage in political wishful thinking. In a column in last Saturday's Post ["The GOP's other Election Day victory"], he wrote that hard-line immigration views are winning over so many Hispanics that it paints "a very bright picture" for the Republican Party. If Smith believes this, he is whistling in the dark, and the tune isn't "The Eyes of Texas." It is "Over the Rainbow."

Smith cites the national exit polling following last month's midterm election that - despite the Arizona law and immigrant-bashing by many Republicans - gave the GOP 38 percent of the Latino vote. This is indeed a big improvement on the 31 percent the exit poll gave John McCain against Obama in 2008.

But Smith is making two politically fatal mistakes. One is that the midterm result is far below the 44 percent of the Hispanic vote that George W. Bush got in 2004, and within the range of mid-30s that Republicans regularly receive. Significantly, it is nowhere near the 45 percent that party strategists know they need to compensate in the future for the declining Anglo share of the vote.

The second concerns the exit poll itself. It tends to overcount the Latino and African American vote as Republican. Only the poll's trends are valuable, as the same poor measures are used each election. Even Bush's 44 percent, reported widely as fact for six years, is suspect.

Warren Mitofsky, former head of the national exit poll, recognized as much after the 2004 election. A Latino-only exit poll by the William C. Velasquez Institute put the Bush number at 31 percent, a huge difference.

Last month, Matt Barreto of the University of Washington and Gary Segura of Stanford carried out a survey similar to an exit poll in eight heavily Hispanic states. The night before polls closed, they sampled early voters and highly likely voters who had cast ballots in the past. Sharron Angle, who may have run the nation's most reprehensible campaign against Latino immigrants, was said by the national poll to have won an incredible 30 percent of the Latino vote against Reid. She won 8 percent in the poll by the two academics.

The spread is similar in almost every state. To confirm who might be right, Barreto and Segura are studying just-released official vote records and applying a widely accepted statistical technique called "ecological inference" that courts use in voting-rights lawsuits.

So far, they have found that in the two counties that make up 95 percent of the Latino vote in Nevada, 94 percent of Latinos voted for Reid. In the five counties that make up nearly 90 percent of the Latino vote in Arizona, they estimate that Gov. Jan Brewer won 12 percent for reelection, and not the totally unbelievable 28 percent that the national poll gave her for the state.

Barreto and Segura's results coincide with what was being universally reported on the ground.

The national poll is good at projecting how states vote but was never meant to measure vote by race or ethnicity. It samples precincts, not people, and even then not randomly. It grossly misses Latino voters, especially Spanish speakers, who are heavily concentrated in only some precincts, mostly urban ones. It overcounts the few, acculturated high-income Latinos who live in mostly white suburbs.

Still, I wish Lamar Smith had been right, though for other reasons. Latinos need a Republican Party that reaches out to them, but on the old grounds of work ethic and family values. That party, however, seems to be history.

Edward Schumacher-Matos is syndicated by the Washington Post Writers Group. His e-mail address is

© 2010 The Washington Post Company