The first poll, done last week and released on Monday, showed Democrat Marshall trailing Republican Austin Scott 51% to 35%. Marshall spokesman Doug Moore called the poll a “public relations stunt” conducted by an agency with “very little experience in polling.”
Gabriel Sterling, Landmark’s vice president, took strong exception to Moore’s assessment.
“We’ve been conducting polling since the early 1990s,” he said by telephone Tuesday morning. “Usually we don’t publicize the results because we do polling for specific candidates. But since we did not have a client in the 8th District we decided to make these results public.”
Today Landmark’s Mark Roundtree restated their experience and added that they had taken great pains to get an accurate sampling in the 8th District but said they did another poll of more than 11 hundred voters last night just to be sure.
From the Landmark Release:
Republican challenger Austin Scott is leading over Democratic incumbent Jim Marshall by solid margins within Georgia Congressional District 8. Last night’s poll results show Marshall at 38.7 percent of the vote with Scott in the lead at 52.5 percent.
In 2006, the African American vote accounted for 25 percent of all voters in the district. For this poll, Landmark Communications raised that percentage to 26.5 in order to accommodate for demographic growth.
Currently 15 percent of African American voters say they are undecided in the election, as well as six percent of white voters. It is our projection that black vote will return to historical levels of 95% support for a Democratic candidate, and that two-thirds of white undecideds will go to Austin Scott. Therefore, it is our projection that Scott will win by a margin of 9 to 10 points and carry around 55 percent of the vote on Election Day.
Landmark Communications polled 1,133 randomly selected active voters residing in Georgia Congressional District Eight. The poll was conducted Tuesday, October 26th, 2010. Only voters who said they would “definitely vote” or “likely vote” are included in the final analysis. Voters who did not identify themselves as “definitely” or “likely” to vote were not included in the final analysis. 25.8 percent of respondents were African-American, corresponding with traditional voting participation in the district in gubernatorial election years. The margin of error of the survey is 3 percent.