Saturday, June 10, 2006

Why Did I Say Anemic?

My buddy Fergus (who really needs to get his own blog, hint, hint) emailed and asked why I called Ned Lamont's poll numbers "anemic". He said 40% sounded pretty darn healthy to him, particularly compared to the previous count of 19%. Isn't that an increase of 21%?

No, not really, and this is why. I preface all of this by saying I am keeping all appendages crossed that Lamont wins the primary in CT, either shutting Lieberman out or else forcing an Independent run. Holy Joe is a loss for the party because of his pandering to the worst of the Rethug elements. Even if he could pull out an Independent victory, it still prevents the Rethugs picking up a seat and justifies the Dems denying him support.

There are three important measures prior to the latest Quinnipiac (hereafter called the Q poll).
  1. The May 4th poll by Q
  2. The May 5th poll by Rasmussen
  3. The results of the CT Democratic party convention
The lowest numbers are from the first Q poll, where Lamont polled 19%. Then came the Rasmussen poll, conducted at almost the same time, where Lamont polled 20% behind Lieberman in a primary match up. The exact polling numbers were not shown. The convention results are known - 33% explicitly voted for him, and rumors were that more wanted to but were intimidated or something, and so did not express their true support.

In the current Q poll, there are two important numbers:
  1. In a sample of *likely* Dem voters, the polls are 55% - 40% in favor of Lieberman. This is the number that is being thrown around the blogosphere.
  2. In the sample of all registered Democrats, the the polls are 57% - 32% in favor of Lieberman.
Now, if you compare the first Q numbers to the second Q numbers, this looks for good for Lamont - he appears to be gaining. But the first Q numbers were of registered Dems, not likely voters, so the correct number to compare are the lower figures from the latest Q poll. This is comparing apples to apples, and his gain is 13% Promising, as I said in the last post, but not spectacular.

Looking more closely at those numbers (look for poll question set 19a on the 2nd Q poll) you can also see a few other things. Lamont's support came partially at the expense of Lieberman (Lieberman lost 8% of his support), but did not make that big a dent in the undecideds, which he has to capture 100% if he has a hope of winning. I suspect the actual opinions shifted something like a sliver of Lieberman supports directly flipped to Lamont (maybe 3%) while more moved to the undecided column. Lamont's gain would then be more from people previously truly undecided, rather than from Lieberman defections.

A reasonable way to look at this is that undecideds are fence sitters. They don't want to be on the losing side, but amy not feel comfortable with the leading candidate (I hear there's a lot of that going around lately...), so they say they don't know. When the leading candidate slips, some of his supporters move to the "Not Sure" column and wait to see if he recovers. When the challenger makes a gain, such as a good showing at a converntion, people who have been undecided make a commitment to the challenger. The number of undecideds doesn't change too much as the numbers leaving are balanced by the numbers entering. Negative campaigning relies on this movement in the voters, because once undecided, they may become discouraged and not vote at all, ratehr than return to their previous choice.

Now, let's look at the Rasmussen poll. Their spread between Lieberman & Lamont is 20%. They do not indicate that they were only polling likely voters, so we can make a reasonable assumption that they were polling registered Dems just as Q was doing. If you look at the spread between the candidates in the current Q poll of likely voters, you can see it is 25% - a bigger gap than between them in the Rasmussen poll. I hesitate to compare polls because the sampling logic and the questions asked make a huge difference, but here is evidence that Lamont was getting support in the low 30s from registered voters before the convention. (I really hope Rasmussen does another round so we can check out their trends.) If the Rasmussen poll is accurate, then Lamont is not significantly improving his numbers among registered voters, even after the convention.

Which brings me to the convention itself. Convention delegates are political animals, and can be considered the most hard-core of the likely voters. Their support is likely to be both stronger and more polarized than any other sub-set of potential voters. It makes sense to evaluate the likely voter numbers of the Q poll to this data sample, not to the earlier Q or Rasmussen polls.

The documented support was 33%. The scuttlebutt number being passed around was between 40% and 45%. The results of the Q poll, in my mind, validate the higher of the scuttlebutt figures, given that the delegates are more extreme than the rest. This is not inspiring news. Why? It means that Lamont has virtually no bounce from the convention. If we take 33% as a stand-in for likely voters (vs. the 45% for the core of fanatics and insiders), Lamont has only come up 8% - the exact number by which Q says Lieberman is down. That's less than double digits. If you take scuttlebutt numbers into account, it only looks worse - all bounce is gone.

So, the big picture is that Lamont is surging only if you take the worst of the early poll numbers and (wrongly) compare them to an entirely different data sample. Otherwise, Lamont is holding a solid 30-35% level of support among possible CT Dems voters and a solid 40% level of support among politically aware and commited Dem voters.

In short, Ned Lamont has not made significant gains since the convention. That is why his poll numbers are anemic. They do not say what people think they say.


No comments: