Thursday, September 14, 2006

Granholm poll, Stabenow poll

The Detroit News/WXYZ-TV EPIC/MRA poll just released new information on MI-Gov and MI-Sen. Here's my election analysis based on the numbers:

Keep in mind that there are a lot of problems with the methodology of polling in general (question wording bias, bias because of the kinds of people who decide to respond, bias in phone ownership, answers differing from actual voting behavior, question order influences). On top of that, no poll accounts for voter energy and election day GOTV. Finally, remember what Mark Grebner had to say about ballot proposals.

Polls provide a "snapshot" in time, but can miss the larger context or underlying issues that could have a big influence on races in the future. Being locally based and people who have an understanding of local politics in their area can provide much more accurate and insightful information.

Also, question order is important. This poll asked several questions before it got around to asking whether the voter would choose Granholm/DeVos/Stabenow/Bouchard. Asking about whether Michigan is on the right track, about Bush, and even each of the "issues" before asking for the final descision could have skewed the results in favor of one candidate or the other.

In this particular poll, 57% females were questioned as opposed to 43% males. The fact that females were more likely to answer the poll tells us two things. It tells us that the method used to pick respondents is not representative of the population. It also explains some of the difference in favor of Granholm and Stabenow (since women tend to favor Granholm and Stabenow by a greater margin than men).

With all of that in mind, here are the numbers:

MI-Gov: Jennifer Granholm 50, Dick DeVos 42 (last month 49-42)

MI-Sen: Debbie Stabenow 53, Mike Bouchard 34 (last month 51-38)

More detailed breakdowns from the poll are available here.

The margin of error for the study is 4%. Given the sample size, the difference between Granholm and DeVos is statistically significant (this poll's data says that Granholm's lead in the results is probably not due to experimental error). Stabenow is, of course, definately ahead of Bouchard at this point.

DeVos' unfavorables are at about that of Granholm's, which means that with equal name recognition, he won't fare any better than he does now (he only has 6% unrecognition anyway). The numbers aren't going to move much from new people finding out who DeVos is, since most people already know him.

Bouchard still has 16% unrecognition, with a 16% unfavorable rating. Stabenow has 3% unrecognition, with 27% unfavorability rating. Because all of the focus is on the more heated and controversial Governor's race, Bouchard has lower name recognition and unfavorability. Look to see both of those increase as more attention is focused on the race and Dems start going negative on Bouchard (he has an interesting history, apparently).

7% of voters are undecided in the Granholm race and 11% of voters are undecided in Stabenow's race. Undecideds lean toward the challenger at the last moment.

Of course, the issue of the day is the economy.

54% of people are concerned with the economy, 66% of people think Michigan is on the wrong track, and 82% of people are unhappy with Michigan's economy. Despite this, Granholm only has a 38% unfavorable rating. Also, these results tell us that people who really like Granholm outweigh those that really like DeVos by a small margin (5%).

Therefore, while voters consider the economy to be very important and are unhappy with it, this does not reflect negatively on Granholm herself. In other words, the dishonest Republican tactic to tie Governor Granholm to our state's failing economy has failed to resonate with voters. Furthermore, Michiganders trust Granholm more than DeVos in terms of improving the economy (48 to 39). When it comes to the economy, they blame Bush over Granholm (44 to 28).

This data says that it doesn't look like the Granholm campaign's efforts to nationalize the issues to include Bush are very effective. Over 85% of people are not basing their vote because of Bush or Bush's ties with DeVos. Despite Granholm's "Gas Prices" advertisement, less than 40% of voters trust her on that issue.

On the issues, voters inherently trust Granholm more than DeVos on affordable health care, stem cells, water quality/the environment, and even having a position on abortion. This could be due to association of these issues with Democrats, Granholm's own accomplishments in these areas, or DeVos' absolute refusal to take a position on pretty much any issue.

Michiganders are notorious ticket-splitters. The fact that 60% of them think that its bad if Republicans control every branch of Michigan Government could work in our favor. DeVos doesn't mention the word Republican on any of his ads or on his website. This means that party identification in the Governor's race is very useful, if played the right way. Mention that Republicans control everything and are responsible for the condition our country/state is, throw the Democrat label on Granholm. BAM! Party ID was about even in the survey, but more Democrats identify themselves as strong Democrats than Republicans did (27-22). Especially given how energized they are compared to Republicans, using the Democrat "brand" will help to turn out those voters.

The drawbacks of political polling aside, consistent results from multiple polls can indicate trends in the way that races are shaping up. Taking a look at the last four polls not done by EPIC/MRA (also not including Zogby Interactive):

Rasmussen (8/31): Granholm 46, DeVos 48 (Instant Voice Recording)
Selzer & Company (8/30): Granholm 46, DeVos 44 (Phone)
Strategic Vision (8/27): Granholm 48, DeVos 43 (Phone)
Survey USA (8/21): Granholm 47, DeVos 47 (Instant Voice Recording)

Aside from the Strategic Vision poll (which calls a narrower lead for Granholm than today's results), Granholm and DeVos are typically within the margin of error of each other. Most recent polls did not show a statistically significant difference between each candidate's performance.

Money wise, Stabenow has about eight times as much cash on hand as Bouchard. DeVos is expected to spend at least twice as much as Granholm, but could spend much more than that if he thinks that it will make the difference between winning and losing. For him, going on the air is a matter of oversaturation, not affordability.

Interestingly, the MCRI now is up 48% to 37%, a dramatic shift from previous studies. According to their data, the MCRI would pass if voted on today. The last MCRI poll had the opposition up by 2 points, inside the margin of error (not statistically significant). Again, ballot proposals are difficult to accurately poll.

Another interesting observation, revealing the power of subtle differences in the way questions are phrased: 70% believe that Michigan's economy is getting weaker, but 24% think Michigan's economy is getting worse.

Cross-posted on Michigan Liberal, MyDD, Daily Kos, and Kicking Ass Ann Arbor.

EDIT: Thanks to Doug (from the Lynne Haley campaign) and for the plugs.


Post a Comment

<< Home