Prior to the first statewide televised debate last week, a consortium of stations airing it conducted a poll. Early this week, the same pollster released one on behalf of an interest group representing health insurers that might be expected to favor Edwards. Not long afterwards, another put out a survey on behalf of contender Republican Rep. Ralph Abraham.
In a nutshell, in these Edwards received 41 to 47 percent of the intended vote; Abraham got 20 to 24 percent, Republican Eddie Rispone bagged 16 to 22 percent; and minor candidates Democrat Omar Danztler and independent Gary Landrieu corralled 2 to 3 percent (their names were missing as choices in the television station poll).
One tactic employed by campaign professionals aggregates polls done relatively close together temporally. While tempting to do so here, as previously noted differences in turnout models, landline/cell phone ratios in sampling, and usage (and definition) of likely voters vs. registered voters used (and no minor candidates offered in one) might make these three too incompatible to make that a useful piece information. Nevertheless, that construction puts Edwards 44 percent, Abraham at 22, and Rispone at 19.
At this point in the contest close to election day, three other considerations need understanding. First, the undecided vote equates closely to nonvoting. One of the first analytical discoveries in voting behavior using survey research was that only a very small portion of the public made up their minds in voting for high-profile races within the final month or so of the campaign, absent some earth-shattering campaign revelation. Of the small portion of the genuine undecided vote out there, most will not vote for the incumbent. For Edwards, what you see here is basically what you’ll get, while together the major GOP candidates will see some small padding.
Second, truly minor candidates (not major-minor candidates that hang in the 5-10 percent region, but below that) will poll under their actual election day totals. Thus, if you see Dantzler and Landrieu together at 3 percent now, they and Republican Patrick Landry will end up closer to 4 or 5 percent. Given the dynamics of this contest, that small bump will almost entirely come at Edwards’ expense.
Finally, a good rule of thumb for this contest nominates 45 as the crucial number for Edwards. Below that percentage in the general election and he is an underdog in the runoff (and below 40 means he has no chance to repeat); above that and he becomes the favorite. If we take 44 as his current number that will manifest on Oct. 12 (without adjusting for the minor candidate bump), he’s an underdog.
All this said, a ray of hope for eventual victory does appear for him in the latest of polls released, that for Abraham. Remington Research performed this one, with a sample size twice as large as typically seen in this election cycle using a defined set of likely voters. It was the one putting Edwards at 47 percent.
Unfortunately, the publicly-available information doesn’t give the landline/cell phone breakdown (JMC Analytics, which did the other two polls, has been using a very reasonable 63/37 ratio, but others have gone much higher on the cell side, which introduces a bias in favor of Edwards). Nor is information about the turnout model, although the model Remington used last year – and they haven’t likely changed it – appears a bit on the high side for Democrat turnout, which introduces a bias in favor of Edwards.
Without that additional information, taking the Remington poll as it is produces the most optimistic result for Edwards at this late date. Still, overall, best we can tell now is (1) there will be a runoff (2) we don’t know which Republican will face off against Edwards in it, and (3) Edwards is at the inflection point between underdog and favorite.