Thursday, April 30, 2009

SUSA Poll shows Creigh best positioned to beat McDonnell, but not much else

Ah, SUSA polls.

The portion of the April 28, 2009 SUSA poll showing the head-to-head match-ups between Bob McDonnell and, respectively, Creigh Deeds, Terry McAuliffe and Brian Moran, was illuminating and, this is somewhat contradictory, both discouraging and encouraging at the same time.

The portion of the SUSA poll concerning the Democratic primary, however, utterly confused me, possibly because its methodology was flawed, but more likely because of my own limitations in understanding polling data. But that said, the poll also seems to have confused a lot of people much more knowledgeable about this stuff than I am, like Bob Holsworth.

With that in mind, just some thoughts on the general election match-ups in the poll – the part by limited intelligence could grasp a little.

For McDonnell, the good news for the general election is that he is leading every Democratic candidate by a margin that is, with the exception of Creigh, outside the MoE.

Looking at the data, however, several items jumped out at me as areas of potential encouragement for Democrats, despite McDonnell’s lead.

First, in NoVA (styled “Northeast” in the poll), the poll shows McDonnell beating Creigh 43-39%, and beating McAuliffe 44-42%, with 19% and 14%, respectively, undecided. The poll shows Moran beating McDonnell in this region, 46-39%, with 15% undecided.

One way to read this is to conclude that McDonnell is making inroads in NoVA and really is ahead, but that is counter intuitive. There is no reason to believe, as of today, that McDonnell will do any better in NoVA than Jerry Kilgore did, and in 2005, Tim Kaine won this area 60% to 40%. Creigh beat McDonnell in the AG race in this area by 58% to 42%.

The numbers in this area of the state are fairly consistent along those lines, so it is fair to assume that, in this geographical region at least, Democratic support is undercounted by some amount. Hypothetically, if you reapportion the votes in the poll in “Northeast” to reflect the 60-40 historical split, it essentially turns the race into a dead heat.

I’d speculate McDonnell’s performance in NoVA in this poll reflects the better name recognition of McDonnell versus each of the Democrats. I’d further speculate that this speculation is supported by the fact that Brian Moran, who is both from this region and has a much more well-known brother serving in Congress from this region, and who also outperforms both Creigh and McAuliffe in this region in the poll, also enjoys greater name recognition in this area.

While it is clear from the toplines that Creigh will present the strongest challenge to Bob McDonnell, consider how each Democrat did against McDonnell in attracting Republican votes:

Creigh: 10%
McAuliffe: 7%
Moran: 6%

As Creigh said the other night during the debate, he is the only candidate in the Democratic race to have won Republican votes in a statewide election.

Second, the party ID in this poll should give Republicans pause and Democrats cause for hope. This poll shows party ID, 37-33, Democrats’ way. Gallup, however, says the spread between Democratic and Republican party ID in Virginia is 9 points. I don’t have the absolute numbers, unfortunately, but this information nonetheless suggests that the ratio of Democrats to Republicans is larger than this sample.

Creigh’s ability to attract Republican and independent voters, the possible undercounting of democrats in this poll and the fact that Creigh effectively tied Bob McDonnell in 2005 after being outspent 2 to 1, all suggest that he is the strongest candidate democrats can nominate.

A quick word on the poll’s methodology.

According to the poll, it was based on the responses of 1,396 registered voters (out of 1,600 interviewed). This is fine in evaluating the general election match-ups. Also, this far out from the election, I see no problem polling registered, rather than likely, voters.

For the primary sample, however, SUSA states, “Of the registered voters, 409 were determined by SurveyUSA to be likely voters in the June 9 Democratic Primary,” but it doesn’t say how they made this determination. Regardless, however, it seems to me (and here I admit to a lack of knowledge about the art or science of polling) that this poll determined that 29% of registered voters are likely Democratic primary voters, and I don’t get that math.

In the 2006 primary, for example, turnout was 3.4% of registered voters. Even in the 2008 presidential primary, when nearly one million Virginians voted, it was only 25% of registered voters. I’m having a hard time envisioning this primary generating higher voter turnout that the Obama-Clinton primary of last year.

Given that, I am uncertain what to make of that part of the poll or whether there was anything useful to be gained from analyzing the internals of what is, for all intents and purposes, a questionable sample.

I am interested, however, to see what the next PPP poll brings – if on schedule, it should be here within a week. I don’t much care for the way they obtain responses, but the poll has the virtue of being consistent, even if it is consistently flawed, over a longer period of time. As a snapshot in time, it may be right or it may be wrong, but it will most certainly reveal which way, if any, the movement in this race is going.

No comments:

Post a Comment