Divergent Senate polling artifact or real movement?

Just when you think you’ve got this polling thing on Louisiana’s U.S. Senate race figured out, here comes another one that tosses aside previous conjecturing … maybe.

September closes with a couple of candidate-related polls sandwiching two independent ones. The latest comes from the pollster for Republican Rep. John Fleming and, surprise, it’s pretty flattering for him. It has Fleming just behind Republican Rep. Charles Boustany and Democrat Public Service Commissioner Foster Campbell, just ahead of former lieutenant governor candidate Democrat Caroline Fayard and Republican Treasurer John Kennedy. In other words, five candidates all poll within four percentage points.

Three major differences stand out from the last independent poll, by Southern Media Opinion and Research: Campbell’s and Fleming’s shares almost double and Kennedy’s falls by a few points. Fayard and Boustany poll about the same, with the proportion of undecided still around the one-quarter mark (although this one presses undecided respondents to answer, creating a category of leaners that leaves only one-sixth genuinely undecided).

Its sample weighing came in pretty close to the population norms of the 2014 general election headed by a Senate contest except being a little on the light side of whites and the geographical distribution seemed correct. Further, unlike the earliest September poll that the Fayard campaign alleged was independent yet by the question wording clearly was designed to favor her, this poll by JMC Analytics had neutral, unbiased questions.

So how could results be so different in a week? The SMOR one stands out for its relatively few respondents, 500, while the JMC one went over 80 percent more, which normally would imply greater confidence in the latter’s numbers. But the real difference comes in that 78 percent of JMC’s was automated while all of SMOR’s had live interviewers (both had to conduct live interviews of cell phone numbers, as the law prohibits automated calls to those). While not disastrous to achieving valid results, automated calling has problems, including not knowing whether the answerer is a registered and likely voter and of the category necessary to balance the sample, and also these calls have a larger nonresponse bias.

Thus, surveying artifact may have shaped these results. Or, perhaps these sudden moves actually have occurred. Fleming’s and Kennedy’s seem believable, as previous poll trends had the former rising and the latter sliding. Fleming’s early embrace of Republican presidential nominee Donald Trump likely soured some principled conservatives on Fleming, even though he appears as the most principled conservative in the field, but perhaps aggressive campaigning might be selling them enough to return to the fold, perhaps from Kennedy.

Campbell’s numbers seem the oddest, for he had been falling but then suddenly bounces back to where he had been a month ago. Further, despite high-profile Democrat endorsements for Fayard and her much stronger connections to the black communities in New Orleans and Baton Rouge, when including leaning undecided voters Campbell led her 38-25 percent among blacks.

Yet most interestingly of all, these numbers suggest the odds of an all-Democrat runoff scenario has gone from roughly one in a thousand to one in a hundred. While JMC suggests a 3:2 Democrat turnout advantage over Republicans, given that a number of Democrat registrants are in name only, in reality, drawing upon the 2014 race, Republican candidates probably will pick up at minimum 60 percent of the vote and Democrats at most 40 percent. If it turns out three Republicans (Boustany, Fleming, Kennedy) and two Democrats (Campbell, Fayard) are competitive, the Democrats slide in only if the Republicans nearly match each other’s vote totals and they do the same. It’s an incredible longshot, to say the least.

But if these numbers are accurate, without leaners 21 percent of blacks remain undecided. Historical norms suggest only two-thirds of them will vote, but if they split evenly that would put Campbell at 24 percent and Fayard at 19 percent of the total vote. Ideally for Democrats she would draw more from this pool than him, that could leave them split 22-21. This means one of the Republicans would have to draw a majority of the white undecided vote, 12 percent but probably only 10 percent actually turning out, to get above at least one of them. The nightmare scenario for Republicans, according to this data, is a bit more possible.

In the final analysis, Kennedy’s (for Republicans) placeholder status seems to have evaporated but only now has the campaign started to crank up the volume so he could regain some ground while the poll probably has the other placeholder (to a lesser degree for Democrats) Campbell a little ahead of himself (remarkably, it has the two north Louisiana candidates pulling down with leaners nearly a third of the statewide vote). Even if so, it leaves five candidates jockeying around 15 percent of the vote each. And it all could change with the next survey’s release.