The Bill Marcy camp released their latest round of internal polling. The JMC Enterprises poll shows Bennie Thompson leading Bill Marcy by a slim 42-41 margin (certainly within the polls’ 4.5 percent margin of error). This mirrors a poll from a month ago, but with less undecided’s this time around.
Before moving further, I’ll point out the same thing I pointed out last time. The racial split is effectively 50/50 for this poll. Here is what pollster John Couvillon said about the breakdown last month:
“One point I do wish to make which wasn’t explicitly mentioned in your blog: everyone is focusing on the 60+ black % of the CENSUS population. I have not seen anything mentioned about the black % of the REGISTERED VOTER population. The best available information I have says that number is 54%. And when you look at those races that, in my judgment, allows me to determine a “likely voter”, the 54% number drops to 50%. Thus, I believed it was appropriate to use 50% as the basis for properly weighting the top line numbers in the survey.”
I still respectfully disagree with that thought. And here is where I’m coning from beyond the census numbers from the district. As far as I can tell, Thompson has never had crossover appeal to white voters. I would venture to guess he wins around 10 percent of the white vote (similar to national Democrats), yet still wins the district with 60 percent of the vote or so thanks to the black vote. Further, black voters in Mississippi tend to vote their share of the population (at least). A slight decrease of a couple percent for a midterm without Obama on the top of the ballot won’t be enough to bring the breakdown to 50/50 in my opinion.
As for the vote by race, the poll finds Thompson leading 75-3 among blacks and Marcy up 77-6 among whites. Findings from the poll indicate that they believe Marcy can win at least six percent of the black vote. I believe all of those numbers are accurate. It simply comes down to the racial split next Tuesday.
Here is a direct link to the full results from the poll.