New Polling Results on Presidential Race and Charter Schools

October 26, 2012 18:31 pm

by Mark Rountree · 20 comments

Our firm released a joint poll today of the Presidential race and Charter Schools Amendment. We (Landmark Communications) conducted the poll jointly with Rosetta Stone Communications last evening for Channel 2 News in Atlanta.

I really find the results of the Charter Schools Amendment particularly interesting.

First the disclaimers and background: 500 likely voters were interviewed last evening, Thursday, October 25, and participated in the entire poll. Black voters made up 31% of the survey, which is about what the turnout percentage will be when all the votes are counted. The margin of error is 4.3%. WSB owns the poll, but we are allowed to release it after they do.

The bottom lines: Romney substantially ahead with a week and a half to go, and while the Amendment leads 47-37%, neither side of the Charter Schools Amendment battle currently carries a majority. The Charter Schools vote will apparently go down to the wire.

     PRESIDENTIAL RACE RESULTS AND KEY FINDINGS

1.     Former Gov. Romney leads President Obama in Georgia 53-42%.

2.     Inside the 15-County Metro Atlanta area: Obama leads Romney 50-46%. Outside of the Metro Atlanta area: Romney leads Obama 60-33%.

3.     Romney leads with all age groups and both genders. Men give Romney a 16-point margin, while women give Romney an eight-point margin.

4.     Younger voters aged 18-35 say they support Romney by a 49-42% margin. Voters aged 65 and older give Romney 56% of the vote vs. 37% for President Obama.

5.     Romney is running stronger among black voters than Republicans traditionally perform. 22% of black voters said they support Romney as of Thursday evening, which is more than twice the level of support given to Republican nominee John McCain in 2008.

6.     Self-described “Independent” voters give 58% of their votes to Romney vs. 25% going to President Obama.

     CHARTER SCHOOL AMENDMENT RESULTS AND KEY FINDINGS

1.     The Charter Schools Amendment leads in Georgia but has not yet earned a majority level of support among likely voters. 47% of Georgians say they will vote for the Charter Schools Amendment while 37% say they will vote no on the Amendment.

2.     There is a stark difference in levels of support based on the age of the voter. Younger voters are strongly supportive of the Amendment (57-32% among those aged 18-35), while older voters slightly oppose the Amendment (40-41% in opposition among those over age 64).

3.     There is very little difference in support or opposition based on gender, race or political party. In other words, there is no significant difference based on whether a voter is a Republican or a Democrat, a male or a female, or based on race.

4.     Support is much greater in the 15-County Metro Atlanta area than outside of it. 54% of voters in the Metro Atlanta area support the Amendment while 32% oppose. Outside of Metro Atlanta 42% oppose it while 41% support it.

5.     To keep perspective about the findings of this poll: With a week and a half until the election, neither side has a majority of the vote. Supporters of the Charter Schools Amendment currently have the lead but with 16% of the vote remaining undecided, this vote remains open for either side to win.

Chris Huttman October 26, 2012 at 6:46 pm

I will bet you $100 that Mitt Romney doesn’t get 10% of the black vote here, but other than that this poll looks pretty legit.

Harry October 26, 2012 at 6:56 pm

Looking at the nationwide polls, there could be a reverse Bradley effect with black voters. They’re not willing to admit to an unknown person that they’re not voting for Obama. In Georgia, they’re not so paranoid about recriminations and Mark’s poll picks up on the actual black support for Romney. Many in the minority communities are quite upset with Obama’s pushing the abortion and homosexual agenda. I’m just saying…cause who knows for sure?

Chris Huttman October 26, 2012 at 7:58 pm

I have done my own extensive polling here and elsewhere and have never seen Romney poll above 11% or 12% with black voters.

You’re probably right about a Bradley effect – they are telling a white voice that they’re voting for Romney but it isn’t truthful.

Luckily, there are plenty of precincts in Georgia that are 100% white and 100% black and in 10 days we can settle this argument.

Harry October 27, 2012 at 12:03 am

I don’t know if I could trust the Fulton elections Board to get it right.

Kenneth Stepp October 26, 2012 at 8:21 pm

I’m liking the numbers so far. Looks like Georgia will remain a red state for a little longer.

Calypso October 26, 2012 at 9:53 pm

Kenneth, I just went to your blog. Interesting. I’ll be back.

SallyForth October 26, 2012 at 10:07 pm

What is Kenneth’s blog?

Calypso October 26, 2012 at 10:28 pm

Click on his name. Have a good weekend, I’m outta here.

SallyForth October 26, 2012 at 10:31 pm

Thanks. And back at you.

KD_fiscal conservative October 27, 2012 at 11:58 am

Romney will not get even close to 22% of black votes in GA. A “professional pollster” should know better than that.

Mark Rountree October 27, 2012 at 2:44 pm

KD, as a professional complainer, perhaps you should look at our firm’s accuracy via Google prior to writing.

Now, back to the post: we are not making any projections at all, actually. We are simply reporting to you that 22% of black voters said on 10.25 that they were for Mitt Romney. I agree with both you and Chris that, ultimately, the ultimate black vote that Mitt Romney wins with black voters will be much lower, and future polls will probably reflect that. I would expect it to be lower when the full persuasion effort occurs this week.

Remember that this year, in 2012, there is nowhere near the statewide effort or money flowing from the Obama campaign nor the national Democratic Party. There is no substantial TV, calling or mail from them. Georgia is not a targeted state.

So without the usual harsh and scary persuasion tactics that we’ve seen in 2008 and 2010, I think you’ll see a lot more free agency among voters who might have been persuaded to vote Democrat in previous years.

That said, there was a spike in opposition to President Obama among blacks beginning earlier this year in Georgia. This occurred immediately following the President’s statement of support for same-sex marriage.

I would expect, ultimately, two things: that 1. the self-stated support for Romney among black voters shrinks back over the course of the coming week, and Romney ultimately does better among black voters in Georgia than McCain did.

xdog October 27, 2012 at 3:15 pm

“harsh and scary persuasion tactics”

Another professional term? Not that I disagree with your conclusion that Mitt will perform better among black voters in Georgia than McCain did. Probably among white voters too.

Chris Huttman October 27, 2012 at 4:44 pm

It is almost impossible for Mitt Romney to do worse among blacks than McCain did.

Cedar Grove Middle
Obama 458
McCain 2

Covington Highway Library
Obama 317
McCain 3

Flat Shoals Elem
Obama 591
McCain 6

That said, I still think Obama will receive 95%+, instead of 98 or 99. I also think it’s almost impossible (in Georgia, but not in other states) for Romney to do that much better among whites. Maybe a point or two, but Democrats here are already scraping the bottom of the barrel with white voters as of 2004 or so.

Mark Rountree October 27, 2012 at 9:00 pm

Yes, xdog. Harsh and scary. If you’ve not seen them, you aren’t in the target market. ;)

John Konop October 27, 2012 at 5:44 pm

Mark,

KD is a rational and smart person that uses his brain. If you got a 22 percent number, you must have a problem with the methodology you used.

Chris Huttman October 27, 2012 at 6:41 pm

90-10 seems like a number to expect to get with a poll, even though I think it will be 95+.

I agree 22% for Romney – problem with the poll?

Chris Huttman October 27, 2012 at 7:04 pm

Though I will say – the overall numbers seem reasonable, if maybe a tad too Republican for the margin on President. Here’s what likely happened – weighting to get the African American % to 31 from a low actual respondent base.

So 500 raw respondents, at 31% would be 155 respondents. Let’s say in actuality the # of black respondents was like 45 – by the way, typically in Georgia because of the nature of who is most likely to answer the phone and the relative difficulty (compared to non-blacks) of finding good landline numbers, you could have a poll expecting 30% of respondent to be black and the unweighted would only be 10% or so. So even though among all black voters, the odds of supporting Romney are low – let’s say they are less than 12%, with a small universe the odds of accidentally interviewing too many Romney supporters are high. And then when you weight the poll to give more weight to the black respondents, you’re representing a universe that should be 155 – margin of error of 8% or so but you’re using data that has a much higher margin of error – 14% or so.

Some math to illustrate. Say that you have a bucket of hundreds of crayons, most of them are blue, but 3% are red. You pick 10 out of the bowl – odds are that because 3% of 10 is 0.3 that no red crayons will be selected. Right? Wrong! Even if only 3% of the crayons are red, the odds are nearly 27% that at least one red crayon will be selected if you choose 10 randomly.

So let’s say that of all black voters in Georgia, 12% will support Romney. Again I think this is high but this is likely to be about what a poll will find. You would need 10 of these 45 respondents to be Romney supporters to get to 22% – and what are the odds that if you randomly interview 45 black voters 10 or more of them will support Romney even if the real % in the population is 12%? About 4%. While that is low, that is right about on the edge of the typical 95% confidence interval for polls. So it’s definitely possible that you could find that – especially if the poll only has 500 unweighted respondents.

Even with a black respondent error like this though, it’s dangerous to just reweight a poll. If Romney is getting 22% of black votes, and black voters make up 31% of the poll, that means he’s only getting about 67% of non-black respondents – who are primarily (but not completely) white. Since Romney is actually likely to get near 77% of white voters and probably break even with other voters, the 67% seems too low for Romney.

So – even though Romney’s black support seems too high, it does mostly seem balanced out by a non-black support that seems too low. It’s hard to explain exactly, but you’ll just have to trust me that the nature of statistics and polling is that you could poll Georgia to see how many voters support Romney – and 53% could be the exact answer, but that same poll could include too many black Romney supporters and too many white Obama supporters but still overall be accurate.

KD_fiscal conservative October 27, 2012 at 9:39 pm

Thank you John…:)

KD_fiscal conservative October 27, 2012 at 9:37 pm

Never said anything about your “firms accuracy”…was just commenting on this one poll…

The fact that “22% of black voters said on 10.25 that they were for Mitt Romney” in your poll could be the result of several factors…maybe sampling error(result of smaller sample of blacks vs. total voters giving you a larger margin of error for blacks) or maybe even a “reverse” Bradley effect. Either way, a little political intuition should raise red flags that those numbers don’t accurately represent the actual support of Romney for that specific population(black GA voters).

Considering the whole purpose of any given poll is to describe a given population at a given time by using a random sample of the population, I would have to say your poll doesn’t do that, at-least for black support of Romney….

jakey October 27, 2012 at 6:44 pm

Release the cross-tabs

Comments on this entry are closed.