Allocating Undecideds

Incumbent Rule Legacy blog posts Likely Voters The 2004 Race

I started to do something this evening I haven’t had a chance to do in several days, which is to simply browse other blogs and ponder where things stand. Naturally, I found myself quickly falling asleep. The curse of new fatherhood. Perhaps that is a sign I should quit and wait for the results tomorrow.

But since I’m awake again, a few quick thoughts. First, the closeness of the final polls is really remarkable. The final national polls from 2000 in my spreadsheet gave an average of 46.6% to Bush, 44.7% to Gore, and 4.0% to Nader. Tonight, according to RealClearPolitics, the average of the final polls is closer: 48.4% for Bush, 46.9% for Kerry, 0.9% for Nader.

As I scan the final polls among the battleground states, Bush’s final polling percentage is averaging 48% to 49% and his margin 2% or less, in Florida, Ohio, Wisconsin, Iowa and New Mexico. Kerry seems to have equally slim leads in Pennsylvania and New Hampshire. If this race behaves as they usually do with an incumbent involved, the popular vote and Electoral College may well be just as close as they were four years ago. I am not sure there is an electoral vote counter anywhere that can divine the final result, and I am not going to try. This outcome of his race certainly looks to be well within the margins of what modern polling can predict.

One interesting late development is the way several polling organizations chose to explain their allocation of undecided voters. This practice has always been something of a personal pet peeve. Media pollsters will spend the entire election arguing that their methods are objective and scientific and beyond reproach, and then they do a subjective allocation at the last moment with little or no explanation. Worse, some pollsters report only the projection, leaving us scratching our heads about what the actual survey showed before they performed their magic allocation.

The Pew and Gallup surveys made different decisions about how they allocated undecided voters, but they at least took a half step towards explaining what they did. The Pew Research Center divided up their 6% undecided evenly, giving three percentage points each to Kerry and Bush, putting their final projection at 51% Bush, 48% Kerry. They write:

Pew’s final survey suggests that the remaining undecided vote may break only slightly in Kerry’s favor. When both turnout and the probable decisions of undecided voters are taken into account in Pew’s final estimate, Bush holds a slight 51%-48% margin.

While the Pew analysts did not elaborate, their report and questionnaire provide a few clues. A few pages later they note that “Bush registers a higher percentage of strong supporters in the final weekend of the campaign than any candidate since former President Ronald Reagan in 1984.” The questionnaire also includes a question that asked those who had been undecided the following question: “Even though you may not have made up your mind, could you make a guess so I will have something to put down?” One percent each “guessed” Kerry or Bush and the remaining four percent still undecided.

Gallup included the following passage in their final release:

The allocation of undecided voters is part of the tradition started in 1936 by Dr. George Gallup, who wanted to provide the public with the pollster’s best estimate of what the data indicate. This year, the allocation of the undecided vote is based on Gallup’s experience in previous presidential elections, showing that in election contests with an incumbent, virtually all of the undecided vote among likely voters will break for the challenger(s). Thus, in this case, with 3% undecided, 2% is allocated to Kerry and 1% to the Nader/other group, resulting in the estimated tie.

I may be wrong, but this is the first time I can remember Gallup issuing any sort of public explanation of their allocation decision.

I know, from the emails I get that some would like to spend the remaining wee hours debating this issue. I’ll pass. Obviously even Pew and Gallup do not agree. I would prefer to get some sleep. Besides, we’ll know the answer in 24 hours.

One more fascinating finding and explanation in the Gallup survey: They seem to be conceding, wisely, a bit of uncertainty about the ability of the traditional model to predict actual turnout:

Before allocation of the undecided vote, Gallup’s likely voter model shows Bush ahead by two points, 49% to 47%, while the results among all registered voters show Kerry with a two-point lead, 48% to 46%.

The poll also shows that among “high interest” voters — all Americans who express a high verbal commitment to voting, regardless of whether they have actually voted in previous elections — Kerry enjoys a two-point lead, 49% to 47%.

In its traditional likely voter model, Gallup screens out older people if their past voting performance does not reinforce their stated intentions to vote. (Younger people who could not have voted in 2000 are included in the likely voter model, based solely on the intensity of their expressed commitment.) Some observers have suggested that in this election year, the intensity of the public’s interest will stimulate a large number of older people to vote for the first time. If that is the case, then the “high interest” voter model could come closer to the final outcome than the traditional likely voter model.

Among all registered voters, 13% say they will be voting in a presidential election for the first time. Among Gallup’s “likely voters,” just 7% say they will be first-time voters.

Prediction? My hunch has been for weeks that the traditional models are underrepresenting those who will turn out out tomorrow, some for the first time, some who have not voted in many years. I think that Gallup’s “high interest” voters may be a better indicator of the result than the traditional likely voter model. But it is a hunch, I am a Democrat, and I may be seeing what I want to see.

The only intellectually honest answer to these questions, including the debate about how undecideds will “break” is to admit that we do not know for sure. Good for Gallup for putting their own uncertainty out in the public domain. I wish they had done so sooner. Fortunately, we should have some answers in about 24 hours.

And now it’s time for everyone to get some sleep, and then go vote.

UPDATE: I see that at least one blog that tracked back to me is actually predicting about what I’ll predict. I have now seen it all. OK: I’ll say the popular vote will be Kerry 49% Bush 48.5%, Nader 0.8% Electoral college? I’m just too sleepy….

—————-
A Special Request: Have you found this site useful? Please help me keep it alive by taking a few minutes to complete this brief (3-5 minute) about Mystery Pollster. Respondent identities will not be tracked and, as such, your participation will remain completely anonymous and confidential If you have any problems with the survey please email me. Click here if you’re asking, “why a reader survey now?” Thank you!!

Mark Blumenthal

Mark Blumenthal is the principal at MysteryPollster, LLC. With decades of experience in polling using traditional and innovative online methods, he is uniquely positioned to advise survey researchers, progressive organizations and candidates and the public at-large on how to adapt to polling’s ongoing reinvention. He was previously head of election polling at SurveyMonkey, senior polling editor for The Huffington Post, co-founder of Pollster.com and a long-time campaign consultant who conducted and analyzed political polls and focus groups for Democratic party candidates.