Friday, November 14, 2008

I'm really not wrong

Blake emailed me some responses to my evaluation of his election data crunching in the Mike Hamby/Elton Dodson non-race, though most of it - with all due respect - appears to be tantamount to him saying 'no way dude!'

Let's examine this nugget ...

For Dodson to have polled in the low 30s – and I can’t emphasize this enough – when signs were plastered all over every polling place saying he’d dropped out tells me he would have done much better had he stayed in the race. I’d argue that not only was a vote for Dodson a vote for Dodson, but a blank ballot was a vote for Dodson because those are folks who would’ve voted for him if he’d stayed in the race.

Not only is there absolutely no scientific way to verify this (because it merely relies on the presumption that Blake 'knows' what those voters were thinking in leaving those ballots blank), it, of course, goes against the stated and defined trend of 25 percent (or more) of the ballots in most non-contested and contested down ballot races being left blank.

As noted earlier, in three non-contested Athens-Clarke County Commission races with well-known incumbents all garnered 75 percent or less of the vote, meaning that 25 percent or more were left blank in those cases. In other down ballot races, this trend was duplicated.

Furthermore, in the contested District Six race between Ed Robinson and Red Petrovs there were 7,240 ballots cast overall, but only 5,289 in that race. Which means - wait for it - that 27 percent of the ballots in District Six were left blank in a contested race.

In two districts that lie within the boundaries of District 10, two races for commissioner - one contested and one not contested - featured identical percentages of 27 percent of ballots being left blank. This is a statistical trend. It suggests that a quarter of the voters who turned out simply did not vote in certain down ballot races and, given the fact this trend spans across numerous races of equal attention and status, as well as covers contested and non-contested races, I would posit they did so without any regard to allegiances, feelings or connections to a particular candidate.

I'm not saying that this doesn't mean that folks didn't cast a ballot against Hamby in a form of protest because they either didn't like him or absolutely loved Dodson. What I'm arguing for is that 25 percent or more of the total number of folks who cast a ballot in District 10 opted to not vote in that race for the same reasons they didn't vote for Court of Appeals or for Superior Court or Public Service Commission or District Eight.

Again, 41,000 votes cast in the Powell-McDonald race. 42,000 cast in the state Senate races. 41,000 cast in the sheriff’s race. Those are your points of comparison, not uncontested judicial races or PSC races where the choice was a Republican or Libertarian. You’re cherry-picking to try to prove a point.

Based on those vote totals, I calculate that more than 22,000 people would’ve voted in the Commission District 10 race had it been truly contested.

The 'cherry-picking' argument works both ways, no? On one hand, those races were the ones that garnered the most attention and had the most interest, thus their participation numbers would be drastically higher. This is true with every election.

In the general election for 2006, 9,279 voters cast a ballot in District Nine, but only 7,579 voted in a four-man race for that commission seat (81 percent). Granted, that's higher than the 25-30 percent non-participation rate that existed this year for down ballot races, but it still proves the point that the further down on the ticket you move, the lower the participation level falls among participating voters.

Excluding early voting, 7,340 folks in that district voted on Election Day in the governor's race in 2006 ... meaning the final tally for District Eight was almost equaled by day of voting for the governor's race (5,203 votes overall were cast early, meaning that probably 45 percent of that total would be tacked onto the governor's race).

The point being ... races up higher on the ticket garner more attention and thus lead to more informed and more involved voters. This is why the sheriff's race and State Senate races drew participation rates of 90 percent or more this cycle compared to the District Six race which was 73 percent.

Furthermore, by including the uncontested races for commission - and the contested District Six race - I don't see how I'm cherrypicking at all. I conceded from the beginning that these down ballot races were low information ones due to less media scrutiny, but it's impossible to argue - given the existing statistical trends - that 22,000 people would vote in a down ballot race. It would be more along the lines of 17,000 or 18,000 given that the data clearly shows that 25 to 27 percent of voters in that district simply weren't voting in those races.

Did Hamby 'underperform' in some areas, particularly District Eight where Andy Herod received much higher vote totals? Perhaps, and I'm not arguing against that notion. I would imagine that some folks didn't want to vote for Hamby for one reason or another, that some folks wanted to vote for Dodson because they really liked him and that some folks, despite the postings, voted for the incumbent because they had little information on the race.

What I don't agree with is some notion that Dodson 'won' because there's absolutely no statistical evidence to suggest that.


Post a Comment

<< Home