The trend matters as much as the average

I would characterize this Politifact analysis as basically accurate but not particularly meaningful.

Republican consultant Karl Rove thinks Georgia Republicans need to be more like their Texas counterparts.

In a May 18 speech at Georgia’s GOP state convention, Rove said Republicans have “got to get outside of our comfort zone and go places Republicans are not comfortable going,” according to a transcript provided to us by the Atlanta Journal-Constitution. “And we’ve got to get candidates who represent the diversity of our country,” Rove said.

“Look, in Texas we get 40 percent of the Latino vote on average,” Rove said. “And that’s because every Republican is comfortable campaigning everywhere in Texas and because we go out of our way to recruit qualified Latino candidates and run them for office.”

Nationally in 2012, Barack Obama defeated Republican nominee Mitt Romney while enjoying substantial Latino support. Some 71 percent of Hispanic voters favored Obama, compared with 27 percent for Romney, according to voter exit polls undertaken for a consortium of news organizations.

We wondered about Rove’s 40-percent-in-Texas claim.

[…]

Mike Baselice, an Austin pollster who has counseled Rove, Perry and numerous Republican candidates, said in an October 2012 memo based on his firm’s Oct. 10-14 survey of 851 likely Texas voters that at that time, Obama had the support of 49 percent of the state’s Hispanic voters, with Romney at 40 percent. According to Baselice’s memo, Republican U.S. Senate nominee Ted Cruz was supported by 36 percent of Hispanic voters, while Democrat Paul Sadler had 40 percent.

The Politico story also mentioned a Texas poll taken on the eve of the November 2012 elections indicating Cruz had 35 percent of the state’s Hispanic vote, outpacing Romney, shown at 29 percent. The poll by Latino Decisions, a Seattle-based firm that specializes in Latino political opinion research, was based on 400 telephone interviews with Texas Latinos who had voted or were certain to vote. Its margin of error of plus or minus 4.9 percentage points suggests that Cruz, but not Romney, was on the verge of drawing 40 percent of the Texas Latino vote.

James Henson, director of the Politics Project in the Department of Government at the University of Texas, struck a cautionary note as we explored Rove’s claim. Any look at how Latino voters divide relies on extrapolation, Henson said, “since there is no direct measure for Latino voting.”

[…]

Upshot: The best a Republican fared with Texas Hispanics in the elections was Kay Bailey Hutchison when she drew half the Hispanic vote in 2000, by one analysis. The same year, Bush got 49 percent in his first run for president, according to that year’s exit polls taken for news organizations, or 33 percent, according to a poll by the William C. Velásquez Institute. Bush also drew 49 percent in 2004, according to the national exit poll.

The worst any Republican fared among Texas Hispanics was Romney’s election eve 29 percent, according to the Latino Decisions poll.

Considering every result except the one for Perry in 2006 (when he faced multiple challengers) delivers an average of 39 percent of the Hispanic vote for Republicans at or near the top of the tickets. We also averaged the poll showings for each election year, reaching an across-the-years average of 40 percent. Trying another tack, we counted only the polled results for nonpresidential candidates, also landing at 40 percent.

There are two basic issues here. One is that whatever polling can tell us, it’s not the only data we have available to us. We also have election returns, and while that doesn’t tell us how many Latinos there were voting and how specifically they did, we can get a pretty decent estimate. As it happens, I did look at Presidential voting in the heavily Latino State Rep districts recently, and the totals for Mitt Romney ranged from 21.8% to 34.1% – actually, Romney went all the way up to 37.3%, as I just realized I missed HD31 when I compiled that list – which needless to say suggests he fell well short of 40%, as we’ve basically known all along. In fact, it’s likely the case that he did even worse in these districts than the numbers given suggest, since some of the voters there were Anglo, and I think it’s safe to say he got more support those voters. As for Baselice, as far as I know he never released the data of his poll, which claimed that “Romney did 12 to 15 points better among Latinos in Texas than in California”, not specifically that Latinos voted 40% for Romney. I’m always extra skeptical of polls whose data I can’t see, especially when they come from the same guy who claimed just before the GOP Senate primary runoff last year that David Dewhurst was going to beat Ted Cruz.

I should note that there were other polls in Texas besides the two mentioned by PolitiFact. The Wilson Perkins poll had Romney at 32% among Latinos; the Lyceum poll had him at 32.5% among Latinos; and the last YouGov poll had Romney at 40% among Latinos. So that’s three out of four polls that publicly released their data showing Romney no higher than 33%, while one poll that did release its data and one poll that didn’t had him at 40.

Getting back to my point about actual election returns, sure there are plenty of Latino voters in places other than those specific districts, but these are the districts where SSVR is over 70%, which gives some assurance that the actual vote totals and the Latino vote totals will be similar. It’s an estimate, like polls are estimates, and in this case it gives some idea of what the upper bound of Romney support from Latinos in Texas likely was. Again, that would put it significantly below 40%.

OK, but Rove was talking about Latino support on the average. That’s all well and good, and for all I know his statement may be perfectly accurate, but how much does the data from the 2000 election really tell you? Texas is a very different place now than it was back then. It would be equally accurate to say that over the 2000-2012 time frame, Texas Democrats averaged two members of Congress from predominantly Republican rural districts. Of course, nearly all of those members of Congress were elected in 2000 and 2002, and the last one was elected in 2008, but the math still works. The point here is that while averages are useful, so are trend lines. Latino support for Republicans is lower now than it was in 2000, or 2004, and it’s likely to stay at those lower levels, at least for the time being. Surely, the high profile opposition to immigration reform among the entire Texas Republican Congressional caucus isn’t going to help their cause here. If the next couple of elections go like the last few have been, it will be about as accurate to talk about Republicans winning 40% of the Latino vote as it is now to talk about Democrats winning 40% of the East Texas vote.

Related Posts:

This entry was posted in Show Business for Ugly People and tagged , , , , , , , , . Bookmark the permalink.