Last month came the second iteration of a Market Research Insights poll whose initial effort varied considerably from almost all other polling. It forecast an incumbent Democrat Gov. John Bel Edwards reelection without having to go to a runoff, although with less certainty than the previous incarnation.
The first poll’s divergent results came in large part due to a sampling frame that likely disproportionately drew from Edwards supporters. It’s possible that the pollster’s guess in this regard could turn out correct on Oct. 12, but runs against the field of play. The second such poll didn’t have all of its statistics made public, but it’s almost certain it used the same sampling frame and therefore has the same validity issue.
Yesterday, another came out that to some degree agreed with this other. Southern Media and Opinion Research found Edwards drawing 47 percent of the vote, GOP Rep. Ralph Abraham taking 24 percent, and Republican businessman Eddie Rispone getting 16 percent, with the remainder undecided.
Understand that, even as the poll asked only about “chronic” voter choices, a good chunk of the undecided vote at this late date won’t vote. Still, it’s conceivable that Edwards could pick up enough of that to avoid a runoff.
Except that the effort had a major error – it didn’t solicit vote intentions for Democrat Omar Dantzler or independent “Go Gary” Landrieu. Together, they won’t draw much of the vote, but they will draw up to five percent from liberals disaffected with Edwards on some select social issues or because of the Landrieu name. And that vote will come almost entirely from Edwards.
Recognize as well that the undecided vote came disproportionately from Republicans and particularly other party respondents (about a seventh and quarter of their totals, respectively). In the former case, these votes will redistribute to Abraham or Rispone, if cast. In the latter case, few will vote at all. (Possibly some of the six percent of Democrats with no choice wanted to state an intention for Dantzler of Landrieu.) This means Edwards can count on few from this pool.
Unfortunately, the poll released no detail on what constituted a “chronic” voter or the distribution of landline and cell phone contacts. If “chronic” largely is based upon recent statewide elections, that distorts the actual electorate which will show up in October that will be more prone to falling back on partisan self-identification, which favors the Republican candidates. If the landline/cell phone balance falls too far in favor of the latter – as it did with the MRI poll – this reinforces that bias against people voting their self-identification (which differs from actual registration).
We don’t know this because these details aren’t public. But we do know the absence of Dantzler or Landrieu does create a pro-Edwards bias in these numbers.
We also know that in the last gubernatorial election pollsters didn’t do well, with a quasi-incumbent bias in their efforts. Throughout most of 2015, GOP former Sen. David Vitter consistently led the field with totals significantly higher than he received on election day and the relatively unknown Edwards the opposite. Only towards the end did Vitter’s numbers come down to near his actual totals, while those for Edwards – even from a very Democrat-friendly pollster – remained well below his actual result where almost all undecided voters appeared to break for him. Past performance doesn’t guarantee future results, but, if the pattern holds, the now well-known Edwards will poll higher than his eventual Oct. 12 total and will get few of the currently undecided voters.
Assuming that and given the results and shortcomings of the latest polls, the conclusion remains the same: Edwards remains at best an even-money proposition to win reelection.