starting with individual polling station numbers and comparing them with officially reported votes. it's the canvassing report that looks like this:
Vote for 1
01 = MICHAEL BADNARIK/RICHARD V. CAMPAGNA 1,828 .28
02 = GEORGE W. BUSH/DICK CHENEY (REP) 215,624 33.05 04 = CANDIDATE DISQUALIFIED 0
03 = JOHN F. KERRY/JOHN EDWARDS (DEM) 433,262 66.41 05 = MICHAEL A. PEROUTKA/CHUCK BALDWIN1,667 .26
----------------------------------------
01 02 03 04 05
----------------------------------------
0111 BAY VILLAGE 1-A 0 237 191 0 0
0112 BAY VILLAGE 1-B 0 221 224 0 0
0113 BAY VILLAGE 1-C 0 203 210 0 1
0114 BAY VILLAGE 1-D 1 244 201 0 1
0115 BAY VILLAGE 1-E 1 235 238 0 0
BAY VILLAGE WARD 1 2 1140 1064 0 2
Although the presidential vote is over 2300 lines, most of the data can be imported into Excel.
What the County is saying is correct: if you add up the total of the individual polling locations (eg, "0111 BAY VILLAGE 1-A"), the total matches the reported votes. Using the subtotals (eg, "WARD 1") gives an inflated number. I assume the reason they give is correct. It doesn't really matter since those numbers weren't used.
Just thought I'd share that to defuse any remaining suspicions. If anyone wants the spreadsheet, just PM me and I can send it. There are plenty of irregularities to focus on -- this isn't one of them.