Reply #4: but he is _cherry-picking_ the "published figures"

[OnTheOtherHand]

Here, let's go CSV together:
http://www.meckboe.org/ENR_MediaDownload/full.txt

"DateTime",,,,,,,,,,,,,,,,,
11/20/2006 21:50:00,,,,,,,,,,,,,,,,,
"Office","Office Number","Precincts Complete","Total Precincts","Voter Turnout","# Registered Voters",,,,,,,,,,,,
,,195,195,157252,522149,,,,,,,,,,,,
,,,,,,,,,,,,,,,,,
"US Congress, District 8",104,29,29,19287,62925,,,,,,,,,,,,
,"Larry Kissell","Robert C. (Robin) Hayes",,,,,,,,,,,,,,,
"Total",10931,5157,,,,,,,,,,,,,,,
"Precinct #",,,"Voter Turnout","# Registered Voters",,,,,,,,,,,,,
002 ,491,169,678,2575,,,,,,,,,,,,,
004 ,441,217,696,2112,,,,,,,,,,,,,
005 ,223,75,320,1363,,,,,,,,,,,,,

OK, so that's a bit hard to read in one's sleep, but it says that in precinct 002, there were 491 votes for Kissell, 169 votes for Hayes, 678 voter turnout, and 2575 registered voters. 491 + 169 = 660, so that's 18 undervotes out of 678. Hmm.

And in precinct 004, 441 + 217 = 658, so, 38 undervotes out of 696.

And in precinct 005, 223 + 75 = 298, so, 22 undervotes out of 320.

Etc., etc.

Finally we get to this:

AB1 ,74,73,1095,0,,,,,,,,,,,,,
AB2 ,5,20,443,0,,,,,,,,,,,,,
CS1 ,17,7,187,0,,,,,,,,,,,,,
CS2 ,0,0,57,0,,,,,,,,,,,,,
PR1 ,76,16,902,0,,,,,,,,,,,,,
PR2 ,7,1,123,0,,,,,,,,,,,,,

Whoa, lots of (apparent) undervotes among the absentees, curbsides, and provisionals. That could bear some investigation. In fact, it's been investigated. The results of the investigation have been described in several posts now. There's no way to interpret the results as pointing to double-digit touch screen undervotes. ("And, of course, these are touch screens." No, they're not: absentee, curbside, and provisional votes are not cast on touch screens. Not that the undervotes are nearly as high as they appear here anyway.)