
Edited on Wed Nov0205 05:59 PM by Time for change
undervotes There were 2382 (8.82 per precinct) undervotes in the V precincts, compared to 8316 (7.14 per precinct) in the N precincts. If we assume that the difference was due to Kerry losses to "disqualified" in the V precincts, and if we assume that 7.14 votes per precinct is "expected", we get the loss of 454 Kerry votes
Peroutka votes Note: My dataset doesn't distinguish Peroutka from Badnarek, but I think I know which is which.
Assuming I'm correct, there were 1.23 votes for Peroutka per precinct in the V precincts, 0.61 votes per precinct in the N precincts (As you know, Kerry would likely lose votes to Peroutka in the N precincts, not the V precincts). Using the same calculations as above, we estimate that Kerry probably lost 722 votes to Peroutka.
Badnarek votes Badnarek is more difficult to calculate because Bush may have lost votes to him in the V precincts. But I think that it's safe to estimate, for now, that overall, Kerry probably lost as many votes to Badnarek as he did to Peroutka.
Bush votes lost from Kerry's column This of course would be much more difficult to estimate, but perhaps it's reasonable to assume that as many votes were lost from Kerry to Bush as Kerry to the other candidates. The effect would be doubled of course, so perhaps we come up with 1444 votes lost in this manner.
complicating factors There are two main complicating factors, each working in the opposite direction: One is that Bush would have lost some votes also, but of course nowhere near as many as Kerry. And the other issue is that Kerry would also have lost votes to "disqualified" in the N precincts. Perhaps these two factors are a wash.
preliminary estimate So the preliminary estimate of lost net Kerry votes in Cuyahoga County due to ballot order rotation problems is about 3500. I'll need to refine that, of course, by entering the more specific data and recalculating.
Question about Richard Hayes Phillips' analysis As I'm sure you are aware, Richard Hayes Phillips noted lost Kerry votes due to ballot order rotation problems. But these were of a very different nature. They were concentrated heavily in about 16 precincts, rather than distributed throughout in the manner that I've calculated above. How did that happen? Once I understand that, maybe I'll have to subtract those precincts and repeat my analyses.
unusual distribution of ballot orders I also need to note that the V precincts were heavily concentrated in Cleveland, where Kerry won by a 83% to 16% margin, compared to the rest of Cuyahoga County, where his advantage was only about 6%. How did that happen?  I thought the ballot orders were supposed to be random? In Cleveland, 27.3% of precincts were V, compared to only 15.3% in the rest of the county. The chances against that happening by chance exceed a million to one. So, how did that happen?
I'll enter the more specific data and get back with you in the next couple of days. In the meantime, any observations on your part would be appreciated.
