FOX5 Atlanta/Opinion Savvy GA Poll: HRC 43/Trump 43

[Adam Griffin]

^^^ I made this back in 2013 to show what percentage of the white and (combined non-white) shares of the vote would be needed to generate a specific result in GA. There are 99 scenarios here. The 2008 scenario is highlighted in green; the 2012 scenario in orange.

The 2016 GA electorate will likely be 58-59% white.

As you can see, a performance by Clinton equivalent to 2008 (23% of whites & 89% of non-whites) would give her between 50-51% of the vote; a fairly comfortable victory.

A performance equivalent to 2012 (20% of whites & 86% of non-whites) would give her between 47-48% of the vote.

I did it as a combined non-white number just because it was too complicated to do on a spreadsheet with three or four variables as opposed to two. Basically, in 2008 and 2012, those "other" non-white voters combined were around 55-60% Obama.

So, Clinton would need to do better with Latino voters (who in all likelihood are more of that "Other" than 2% or whatever; they disproportionately have registered in the past few years and SoS no longer mandates that race be specified in voter registration applications) by several points. She would also need to do at least as well as Obama did with black voters in 2012 and she'd need 2008 numbers among whites to win.

You must be logged in to read this quote.

[NOVA Green]

^^^ I made this back in 2013 to show what percentage of the white and (combined non-white) shares of the vote would be needed to generate a specific result in GA. There are 99 scenarios here. The 2008 scenario is highlighted in green; the 2012 scenario in orange.

The 2016 GA electorate will likely be 58-59% white.

As you can see, a performance by Clinton equivalent to 2008 (23% of whites & 89% of non-whites) would give her between 50-51% of the vote; a fairly comfortable victory.

A performance equivalent to 2012 (20% of whites & 86% of non-whites) would give her between 47-48% of the vote.

I did it as a combined non-white number just because it was too complicated to do on a spreadsheet with three or four variables as opposed to two. Basically, in 2008 and 2012, those "other" non-white voters combined were around 55-60% Obama.

So, Clinton would need to do better with Latino voters (who in all likelihood are more of that "Other" than 2% or whatever; they disproportionately have registered in the past few years and SoS no longer mandates that race be specified in voter registration applications) by several points. She would also need to do at least as well as Obama did with black voters in 2012 and she'd need 2008 numbers among whites to win.

You must be logged in to read this quote.

Yes!!!! President Griffin is in the House!

Actually when we first started to see these neck and neck Georgia polls, I tried to pull up some of your previous posts, remembering your detailed analysis of Georgia from back in '08, but you have so many posts I couldn't find what I was looking for.

So basically if Clinton can garner 23% of the White vote in a 4 person race, or even further 25% GA will likely flip?

[Adam Griffin]

So basically if Clinton can garner 23% of the White vote in a 4 person race, or even further 25% GA will likely flip?

Assuming the vote for Johnson is large enough (Stein will not be on the ballot in GA), then yes: that could work.

What may be more likely (assuming Johnson doesn't get more than 1% or so) is that she'll need 23% of whites (2008 Obama) + 86% among non-whites (2012 Obama) to be in contention...and even then it's no guarantee.

Unfortunately, I did not put more combinations in the spreadsheet; the combinations that exist among white/non-white levels of support tend to increase/decrease at roughly the same amounts up and down the sheet. So, for instance, there isn't a scenario on there were Clinton gets 23% of whites (2008 Obama) and 86% of non-whites (2012 Obama).

However, that parallel increase/decrease between support scenarios does make it possible to roughly, basically, physically 'average' the distance between two support scenarios up there (remembering that whites are roughly twice as numerous as blacks) to get an idea of where her share of the vote would be. So, for instance, if we take Obama 08 white support and Obama 12 non-white support, and we know that whites are roughly twice as numerous as blacks, then we'd plot that point (twice as close to  the relevant white scenario as the non-white scenario) and draw a line over to the white/non-white electorate we're expecting, and:



Somewhere between 49-50% of the vote in that case.

[Adam Griffin]

^^^ Actually, I forgot to include the non-white, non-black voters in my proportioning there. The electorate is closer to 60/40 than 67/33, so that line should be a tad bit further up. That means the scenario would be more like 48.5 to 49.5% of the vote instead of 49 to 50%.

EDIT: It's also worth noting that all of those scenarios on the sheet above assumed 1% of the vote goes to third parties. I mean, I made it in 2013...