Friday, November 07, 2008

Franklin Co., where vote totals are more like guidelines or suggestions...

In 2004, I got the blogging bug. I posted a diary on DailyKos that showed one Gahanna precinct with more than 4000 votes for W, several times the number of registered voters. This got picked up and started a buzz that ultimately helped fuel the stolen elections meme (I'm sorry about that, but it most certainly would have happened without me).

(I pored over numbers from several BOE's, and I'm pretty sure that Bush's margin was in 2004 was far less than the 118,000 that went on the books officially. I'm also pretty certain that Bush's margin was closer to 118,000 than to 0, and that Bush carried Ohio more substantially than Kerry carried Wisconsin, for example.)

I bring up ancient history, because Franklin County is home to one of the closest congressional races in the country (Kilroy/Stivers), and has a state house race (Marian Harris's race in the 19th), where the margin is currently listed as 40 votes out of more than 65000 cast. It is imperative that the count be accurate, yet what do we get?

1) Franklin County is alone in reporting "overlap" numbers, meaning that for congressional races in which the majority of voters reside in Franklin County, Franklin County puts the votes from all counties in the district into its "county totals." This practice fooled everyone from CNN to ONN into reporting totals that double-counted Union and Madison in the 15th and double-counted Delaware and Licking in the 12th. This confusion was so pervasive that the SOS results website, which was the only major vote count site to have the totals accurately reflect the unofficial results, was 'corrected' yesterday to show the inaccurate results consistent with everyone else. Way to go, Franklin BOE.

2) After having an embarrassing glitch in which the unofficial count included precincts with more votes than voters in 2004, that would never happen again, right? Wrong. Three precincts in the unofficial canvass had more votes than voters this year. This casts doubt on every single one of the precinct-level totals reported. I am not exaggerating or engaging in hyperbole. Let me explain:

The official explanation is that votes "were counted twice." As an example, Worthington 3-D originally had 633 votes counted for 534 registered voters. As you can tell by the odd number of votes cast, we're not looking at a simple doubling of all the precinct's votes. We're looking most likely at a doubling of the machine-votes from election day. The corrected total listed for that precinct this morning is 411 votes, meaning that 222 votes were double-counted.

Now imagine a precinct with 1000 registered voters, 600 of whom voted, 400 of whom voted early/absentee/paper ballot. Now imagine that the election day machine votes were double-counted. The report would show 800 votes (400 + 2*200), for a turnout of 80%. When turnout is listed at 119%, you look for a problem and fix it. When turnout is listed at 80%, you applaud civic engagement.

If these get missed, it will have the effect of slightly amplifying e-day machine voting, and diluting early/absentee/paper voting. If a race has a 40,000 vote margin, this won't matter much. With a 400 vote margin, or a 40 vote margin, it could make the difference.

Out of curiosity, you might ask, are there any precincts that show unexpectedly high turnout that could reflect such a situation? I'll throw out a candidate:

UA Ward 1 2006 turnout 2008 turnout
UPPER ARLINGTON 1-A 52.7% 91.1%
UPPER ARLINGTON 1-B 57.8% 74.1%
UPPER ARLINGTON 1-C 42.4% 75.2%
UPPER ARLINGTON 1-D 50.4% 75.6%
UPPER ARLINGTON 1-E 61.2% 84.2%
UPPER ARLINGTON 1-F 54.6% 83.4%

UA Precinct 1-A had average-to-below-average turnout compared to other UA ward 1 precincts in 2006, yet had far-and-away the highest turnout in 2008. In fact, after having good (but not remarkable) turnout in 2006, they had the highest turnout of any of the 854 precincts in Franklin County in 2008. Perhaps they had a couple of really good precinct captains, but perhaps the BOE might want to double-check...

And in case you're wondering, Obama is listed as winning UA1-A 449-390, but Stivers is listed as leading Kilroy 471-321.

Of course, all of this probably pales in comparison to what's going on in the provisional balloting, which is apparently concentrated in certain precincts (inferred by the opposite pattern, where turnout is unexpectedly low given past turnout and turnout in nearby precincts, e.g. Cols 12-B, Cols 42-B, Cols 51-D,Cols 84-G, Dublin 1-F, Prairie-J, Marble Cliff, etc.).

Some of this is due to random variation and the perfectly mundane influence of unknown events, but...

I guess we'll know more next week.

1 comment:

britney said...

nice post..

The best place for the best ENTERTAINMENT