The RNC Report, the Hispanic Vote, and the 2004 Election

Recents posts have discussed the RNC report on the state of the Republican Party.  At one point, the report makes a claim about the 2004 election:

President George W. Bush used to say, "Family values don't stop at the Rio Grande and a hungry mother is going to try to feed her child." When Hispanics heard that, they knew he cared and were willing to listen to his policies on education, jobs, spending, etc. Because his first sentence struck a chord, Hispanic Americans were willing to listen to his second sentence. We heard this from other demographic groups as well. President Bush got 44 percent of the Hispanic vote, a modern-day record for a Republican presidential candidate.

The 44 percent figures comes from a national exit poll. But there is reason to think that this poll overstated Bush's Hispanic support.  In 2005, the Pew Hispanic Center reported:

The controversy grew more complex when Ana Maria Arumi, a polling specialist then of NBC, which was a member of the NEP consortium, offered fresh insight on the exit poll at an event hosted by the National Association of Hispanic Journalists (NAHJ) on December 2, 2004, and in a brief news item about the event on MSNBC.COM. Arumi said that the selection of sample precincts in the NEP produced an overrepresentation of Cuban respondents in Miami-Dade County, a population that is typically the most pro-Republican segment of the Hispanic electorate. A better assessment of the Hispanic vote, she said, could be developed by aggregating exit polls conducted individually in the 50 states and the District of Columbia. That analysis showed that Bush had drawn 40 percent of the Hispanic votes, she said. Arumi’s comments appear to be the most extensive analysis of the state exit poll findings regarding the Hispanic vote that have been made public by members of the NEP consortium. Full data from both the national and the 51 state polls has become publicly available, and in conjunction with data from the CPS it is now possible to assess these findings.

The national exit poll was based on a sample of 250 precincts designed to be representative of the nation as a whole, and 1,037 respondents at those precincts identified themselves as Hispanics. At the same time as the national poll was being conducted on election day, the NEP was also conducting 51 individual polls designed to produce results representative in each of the 50 states and the District of Columbia. These polls were taken at 1,469 precincts at which 4,469 Hispanics were interviewed. The Pew Hispanic Center has aggregated data from the 51 state polls and weighted the results to produce results for the nation as a whole. As Arumi noted, the 51 state polls show that Bush drew 40 percent of the Hispanic vote rather than the 44 percent in the national poll.

David Leal and colleagues wrote in "The Latino Vote in the 2004 Election" (PS:  Political Science and Politics 38 (January 2005): 41-49):

We conclude that the pre-election data provide little evidence that President Bush received the 44% level of support from Latinos estimated by the 2004 exit polls. We examined 10 such surveys and found Latino support averaging 60% for Kerry and 32 percent for Bush—which is the traditional two-to-one ratio of support enjoyed by the Democratic Party. Support for Kerry and Bush was large and found within almost every standard Latino demographic. Equally problematic is the lack of movement over time for Bush in these surveys. In addition, two surveys by the Washington Post/Univision/TRPI in both July and October found a 60–30 split, and Latinos were generally more likely to identify as Democrats (66%) than Republicans (24%).
...
Evidence from Texas counties and urban precincts also calls into question the exit poll claim that Bush achieved 59% of the Latino vote in his home state, which in turn suggests there may be problems with the national Latino exit poll data. Given these consistent patterns, it seems more logical to conclude that the exit polls mistakenly depicted the Latino vote than to accept that Latino preferences could have changed so substantially in such a short period.

Moreover, National Annenberg Election Survey put the Bush figure at 41 percent.

Clearly, Bush's 2004 showing among Hispanics surely was much higher than Romney's 27 percent. So why hype it by citing an inaccurate number?