TDS Co-Editor Stan Greenberg of Greenberg Quinlan Rosner and Democracy Corps has a very interesting response to a memo by McCain’s lead pollster, Bill McInturff, which made a case that McCain is on track for an upset win on Tuesday. Greenberg’s letter follows:
I very much enjoyed your spirited note on the state of the race and Barack Obama’s “ballot position.” It reminds me how much I miss our times working together on the bipartisan polls for NPR and for many of our corporate clients. I miss in particular the banter before those meetings when your Republican colleagues fretted over their teenage children going off to Obama rallies.
Using your last four days of tracking polls in your battleground states and your underlying analysis you conclude that “this is a hard election to predict” because of the unprecedented interest and likely turnout, and that “all signs say we are headed to an election that may easily be too close to call by next Tuesday.” You get there by underscoring a number of emerging patterns – “what we know for sure.” First, “The McCain campaign has made impressive strides over the last week of tracking.” Second, that the campaign is “functionally tied across the battleground states.” Third, “the key number in our mind is Senator Obama’s level of support,” which is dropping below 50 percent, with good reason to believe that Obama “gets what he gets in the tracking” – with few prospects of more votes from the undecided, African Americans or turnout patterns. Fourth, that Obama’s margin over McCain is “beginning to approach margin of error with a week left.” Fifth, McCain is achieving breakthroughs with “soft Democrats,” “Wal-Mart women” and an “impressive pop” with independents. And finally, that “Joe the Plumber” has astounding recognition and changed the debate on taxes and the economy.
All of us Greenberg Quinlan Rosner were in awe of the boldness of these assertions, as they, we are sure, honestly reflect the data, show a nuanced use of language and topics, and weave a story that almost gets to the conclusions that this race is opening up, unpredictable, and closing, without formally saying those things. What is interesting is that some of the findings match ours, but with a fuller picture, take us to a different close.
We wanted to take this opportunity to discuss the “things we know for sure”
– this time using the battleground polls that we completed yesterday. These are complete sample surveys of 1,000 and 1,200 likely voters, not rolling tracking polls; the first is in the fifteen presidential battleground states, and the second in the thirty-five Republican-held congressional districts, from the 21st to 55th most difficult districts for a Democrat – more than half of which are in presidential battleground states. These voters gave George Bush a 15-point margin in 2004. In addition, to look at the “undecided” and “refused” respondents, we combined 3,000 interviews from our presidential battleground states over the last 2 ½ weeks.
Let’s start with the very important conclusion that Obama’s tracking number, now 50 or 51 percent, is “what he gets.” First, let’s not disparage 50 percent, as no Democrat has received that level of support since Jimmy Carter in 1976. The 3rd party candidates are polling about 3 or 4 percent – and showing about 2 points in our surveys of early vote tallies. That suggests that Obama would win even if you gave McCain all other votes. But our data does not support the “Obama gets what he gets” thesis.
* The memo reports that Obama is already getting virtually 100 percent of the African American vote in McCain’s polls. That is not true in our combined database of the presidential battleground states where Obama is polling 89 to 6 percent. On that basis alone, one would expect Obama’s overall vote to rise a point.
* Note that the same is true of Latino voters. In special surveys of Hispanics, using special lists, Obama is polling close to 70 percent, but in the combined battleground polls where Hispanic respondents are more acculturated and English-speaking, we have Obama’s vote at 56 percent to 36 percent for McCain. That too can produce another point of Obama support.
* The memo says that the “undecided” and “refused” voters “will break decisively in our direction, adding a “net three plus points to our margin.” That is pretty amazing. Using the combined database, we looked at the “undecided,” “refused” and the undecided “leaning” to a candidate – 7 percent of the electorate. Using their stated leanings to the candidates and feelings toward the parties, this undecided vote broke near evenly between Obama and McCain. In our latest presidential battleground poll, they broke near evenly as well. To get a 3-point net gain, the undecided would have to break 5 to 2 for McCain. There is no evidence to indicate such an impending break against Obama. Instead, the undecided could push Obama’s vote up at least another point.
You are right to marvel at the remarkable intensity of interest – the reason to believe the outcome is unpredictable. We agree that surging turnout brings unpredictability, but likely with greater risks for Senator McCain than Obama. We are not looking at simply random unknowns. Let’s focus on turnout models and assumptions.
* Virtually every model that allows for expanded turnout shows an increased lead for Obama. Yesterday, Gallup’s “traditional” model showed the race at 3 points for Obama, but its “expanded” model put it at 7 – and has consistently shown a wider lead with larger turnout. In our combined database for the presidential battleground, Obama’s lead in the full likely sample is twice his lead in our narrower, core sample.
* What do we make of the early voting – those most motivated or most organized to vote? In our latest presidential battleground poll, nearly one-fifth of the electorate had already voted and they broke 64 to 29 percent for Obama. In North Carolina, 1.8 million people have already voted, half the vote of 2004, with the turnout 5 points more Democratic.
* The McCain polls assume that young voters and other new voters will not play a disproportionate part in an expanded electorate. And yet, in the primaries, though turnout more than doubled, young voters and Latinos significantly increased their proportion of the overall expanded electorate. We would not assume that the same cannot happen next week.
* While we have allowed a more modest increase in our proportion of young voters and Latinos in our weighting, we still cap the African American proportion at 10 percent, Latino at 8 percent and the under 30 at 14 percent. These caps could all be low, but not likely high.
While we agree that Obama’s vote may have dropped a point, that does not take you to the memo’s observation that Obama’s lead is “beginning to approach margin of error.” We are not entirely sure of the meaning of “beginning to approach” and “functionally tied.” Obama’s vote dropped a point to 51 percent in our presidential battleground poll this week and McCain narrowed the lead to 8 points – in states carried by Bush by 4 points in 2004. In short, nowhere near the margin of error, but maybe moving slightly in that direction, and hardly “functionally tied.” In our survey this week of Republican congressional battleground seats, Obama’s vote went up a point to 48 percent – 2 points ahead of McCain in districts Bush carried by 15 points four years ago.
The McCain campaign is like a football team that has just gained 5 yards but still faces 4th and long, deep in its own territory.
The only way to get to “functionally tied” is to narrow the competitive battleground to those few states where Obama’s lead is less than 3 points and where Obama’s vote is at 48 percent or less, like Indiana, Florida, Georgia, Missouri and North Carolina. Since the memo assumes Obama will only get his tracking number, the two candidates are “functionally tied” in this battleground, which we agree could be carried by McCain. Unfortunately, McCain could win them all and not come close to winning the Electoral College. The memo is silent on Pennsylvania where Obama has a double- digit lead and where Obama is polling 52 percent, enough to win the state and the presidency under McInturff rules.
What is hopeful for McCain in the memo are the groups that are “swinging back” in the last two nights of the campaign’s tracking. We conducted our two surveys at the same time, and many of these changes were not evident in our polls.
* The memo says McCain is “beginning” to get over 20 percent of “soft Democrats,” but we show absolutely no erosion of Democrats in either of polls this week. Remarkably, McCain is getting only 9 percent of moderate and conservative Democrats.
* The McCain campaign claims to be retaking rural territory, but both our polls show Obama splitting evenly the white rural vote.
* The memo offers hope among the “Wal-Mart women,” yet in both our polls Obama and McCain are running evenly with white non-college women – well ahead of Kerry and Gore among these voters.
We do show that McCain has made some gains with independents and that is not trivial. Still, across our two polls this week, Obama has an average lead of 7 points with independents. And Obama’s outside lead on the economy and financial crisis has fallen back in these battleground states to his traditional advantage of 10 points on the economy and 8 points on the financial crisis. But Obama is maintaining a 6-point lead on who would do a better job on taxes – despite the sustained effort by McCain to move the number.
True, McCain and Palin have made “Joe the Plumber” famous, but not in ways that have been helpful to McCain. In our tracking of people’s viewing of the news and communications over the previous few days, what they have heard about Obama has made them more favorable by 10 points (46 to 36 percent), while what they have heard about McCain has made them less favorable by 8 points (36 to 44 percent). Week after week, Obama gets favorable comments from voters, but McCain, mostly negative.
Last week, we tested the big, unfolding tax debate – including the “Joe the Plumber” storyline of wealth redistribution, raising taxes on the wealthy and cutting taxes for the middle class. Obama was winning that argument by 14 points (see Democracy Corps poll of 1,000 likely voters nationwide conducted October 21-23, 2008). In this battleground poll, Obama has taken virtually no water on being “too liberal” or “will raise my taxes” – both essentially unchanged over the past month at 51 percent.
In our regression modeling for the presidential battleground states, the two biggest drivers of the vote are who “brings the right kind of change” and who is “on your side.” Obama enjoys double-digit advantages over McCain on both – with about 53 percent of likely voters choosing him, creating an upward pressure on the vote. The structure of this race just has not changed as we have moved to the finish line.
None of that takes away even a bit from our respect for you, as a pollster, building on a series of plausible assumptions and theories, elevated uncertainties and carefully chosen words to get us to a race that is not yet over. This is your job and indeed, the voters are only now beginning to have the final say.