After 11 weeks of interviewing, our fall tracking poll comes to an end with this, the 10th Wave Report. Note, however, that the Latino Decisions Track will continue in 2011, though on a bi-monthly basis.
On this, the eve of the 2010 Midterm Elections, we find a Latino vote more energized, more enthused, and significantly more Democratic than at the start of the general election campaign. Overall, with undecided voters pushed to make a decision, we report 70% vote intention for Democrats, and 30% for Republicans, on the generic ballot midterm question. We also find a Latino electorate still ambivalent about the Obama administration and still not convinced on its commitment to the cause of comprehensive immigration reform, which was promised, but for a variety of reasons failed to come up in the administration’s first two years.
The percentage of Latinos reporting a certainty of voting has climbed to 76.9%, its highest level in the entire track, more than five points higher than where we were eight weeks back and more than 10 points higher than the low point in Wave 5.
On a number of other dimensions, we have seen considerable back-sliding from earlier trends, or general stability over the life of the poll.
Obama approval has dropped back to about 65%. This number is over four percentage points higher than 8 weeks ago but is actually 6% weaker than on October 4. So some of the recovery in Obama’s approval in the early waves of our track has receded, outside the margin of error, in the last four weeks.
Issue-specific approval of the administration’s handling of the important issue of immigration has also backslid. In late September and early October, approval of the president on this issue hovered just beneath 50%. Today, its at 44%, a significant decline from that earlier period but stable over the last two weeks, giving us considerable confidence in that trend.
There has almost no movement in the parties’ approach to immigration, and the perceptions Latino voters have of the parties on this issue. While Democratic reputations improved slightly in the first half of our tracking period, attitudes have been stable over the last five weeks, even as the Senate took a half-hearted stab at the Dream Act as part of the Defense Appropriations bill. Latinos views of the GOP, while somewhat more volatile over all than the numbers for Democrats, are essentially unchanged over the last half of the track.
Most importantly, the self-reported vote has remained oddly stable after some earlier Democratic growth, with 60% reporting a Democratic preference, compared with 22% Republican, and a disturbingly stubborn 12% undecided, even at this late date (though we should note the generic ballot question does produce more “undecideds” than when specific names are read). Several observations are worth noting:
- At 60%, the Democratic vote share is only about five percentage points improved over the last eight weeks, and the overall pattern is stability.
- GOP vote share has almost never moved outside the margin of error. The Republican share of vote intentions has hovered around a mean of about 22%.
- The undecided vote has declined as we might expect, but certainly not to the degree we might expect. At 18% in the week before the national election, these numbers need more careful examination.
First, if we push these undecided voters to choose a side, the results are curious. As with all polls, if a voter says “undecided” we routinely ask them, which party’s candidate they are leaning towards. However, in our final week of interviewing, we asked each voter expressing indecision one additional follow-up question, whether they thought they might end up voting Democratic, end up voting Republican, or probably just not turn out. The results were interesting. Of the 18% who were undecided, 5% percent said they probably wouldn’t vote, 28.5% thought they would end up Democrat, and a surprising 47.4% thought they’d vote Republican. The remaining 19% still refused to offer an opinion on who they would vote for. Almost half of the “undecided” lean GOP if pushed twice. This reluctance to offer an initial opinion could reflect a social desirability bias—that is, an unwillingness to say they prefer the GOP until pushed—or it may reflect a bit of cognitive conflict, where GOP preferences compete with other social forces. Either way, should these results hold across our respondents, the corresponding vote percentages would shift to 63% Democrat, 27% Republican, and 10% firmly undecided. This last group could be assumed to be highly unlikely to vote, which would yield an election day distribution of 70% Democratic, 30% Republican, numbers highly consistent with the 2008 vote and the historical distribution of Latino votes.
It is worth noting that voters with stronger histories of voter turnout behave differently than registered voters who have been recent abstainers. Since our sample data allow us to examine the turnout histories of our respondents, we can group them on this basis to examine the differences or similarity in their behaviors. The news here is good for Democrats, and reinforces our speculation that social cross pressures—rather than merely a reluctance to tell the interviewer their genuine views—may be driving behavior.
If we examine turnout in 2006, the 2008 primary, and the 2008 General Election, we can decipher whether the respondent is “in the habit” of actually turning out. Consistent with a pattern we reported last week, two-party preference for the Democrats is linearly related to the frequency of turnout, with support for Democrats among those with perfect attendance fourteen percentage points higher than among those who have missed all three recent chances to vote.
Taken together, these two trends—the persistence of undecideds and their apparent GOP tilt, as well as the association of higher turnout history with Democratic vote intention—signal an important factor in Latino voting behavior, one consistent with our view that cross-pressures demobilize Latinos with GOP tendencies. If this finding is generalizable across time, it may well suggest that even carefully crafted poll estimates will underestimate Democratic vote share. Perhaps we’ll find out tomorrow.