First off, most OH counties give county-wide summary results. Many of these summaries can be accessed via an index here. Four of the counties give precinct-level data. These are Frankin (Columbus),Cuyahoga (Cleveland), Butler(north of Cincinnati), and Lucas (Toledo). It looks like Franklin has already taken it's report off the web (though many of us already have the 414pg precinct-level data). Cuyahoga appears to be rerunning their totals with Friday's run here. Cuyahoga also gives the precinct-level data from the Gore/Bush contest in 2000 here.  The Cuyahoga precinct maps are also there.

The data irregularies show up in the east side of Cleveland (that is, the part of Cleveland east of the Cuyahoga river - not "East Cleveland").  This area is represented by wards 1 through 10, with letters denoting the precincts within each ward (about 200 total precincts, each with roughly 500 registered voters).

This region is very democratic.  Gore beat Bush in these precincts 61,444 to 2,256 in 2000, about a 27-to-1 ratio (check the links above to verify... I'm not including absentee ballots).  Kerry beat Bush in these precincts by 74,934 to 3,816 in 2004, about a 22-to-1 ratio. It is no surprise that in 2004, Kerry took the median precinct with 96% of the vote here, nor is it surprising that he won 3/4 of the precincts by at least 94%.

The precincts that stand out are the following 7:

precinct     BAD BC K/E PER Bal Sp %Kerry 

CLEVELAND 4F   0 21 290 215 554 28 55.1

CLEVELAND 4N 163 11 318   7 507  8 63.7

CLEVELAND 5B   0 12  74  16 102  0 72.5

CLEVELAND 8G  51 19 225   1 305  9 76.0

CLEVELAND 3I   1 13 464  70 567 19 84.7

CLEVELAND 8I   0  8 245  27 318 38 87.5

CLEVELAND 3B  41 17 456   1 533 18 88.5

Bad is Badnarik, BC is Bush/Cheney, K/E Kerry Edwards, Per is Peroutka. Bal is the total number of ballots cast, and Sp are the number of spoiled ballots for the presidential race.

In each of these precincts, either Badnarik or Peroutka received a disproportionately high number of votes (as if one or more voting booths at these locations were labelled incorrectly, or some levers punched the wrong column). It's not like people in this urban area will be drawn to the Texan Badnarik or the Constitution Party's Peroutka. If you take these 580+ votes in these 7 precints away, these two guys received 260 votes in all other 200 precincts in these 10 wards in the east side of Cleveland (an average of 1.3 third-party votes per precinct). Again, this is in a region where it is surprising if Kerry receives less than 90% of the vote (in each of these precincts, Bush is still receiving 2-6% of the vote, except for 5B where he gets 11%). It's not possible to argue that these third-party votes were meant for Bush.. he rarely received more than 5% in any of the 200 east-side Cleveland precincts.

Because this is such an extreme case, it's an easy argument that these 580+ voters were disenfranchised by either bad equipment or by being deceived... and that these votes are meant for Kerry.

Lastly, this shows how useful it is to have precinct-level data publically available.  I suspect the data irregularities are much more likely to show up in the rural areas..and in counties that are not confident enough to publically publish the data.