A couple of people asked for more detail concerning the Mississippi
reporting errors, so I thought the list as a whole might be interested.
The Lowndes County error that cost Bush some 10,000 votes can be viewed in a
pdf available on the Mississippi SoS website. Essentially, the county
dropped a digit in the county totals, which can easily be verified by adding
up the precinct results.
http://www.sos.state.ms.us/elections/2004/General/Lowndes.pdf
The errors cut both ways. You can view an error reducing Kerry's vote by
some 300 votes in DeSoto County, Bridgetown precinct. Here, Kerry's votes
seem to have been transposed with Harris's vote in that precinct.
http://www.sos.state.ms.us/elections/2004/General/DeSoto.pdf
Mississippi seems to be plagued by errors since they report by paper and I
understand they are moving to an electronic filing system. Reporting by
spreadsheet would reduce some of the errors. The SoS office reported that
the counties sought court orders to change their certified results, but I do
not know the final outcome. I posted the corrected numbers on my website,
both for the state of Mississippi and the nation as a whole.
There are other cases, such as California where I could not reconcile the
total number of ballots cast among the official documents. The California
Statement of Vote reports total voters of 12,589,683 while the vote for
president, if you sum all the votes including the category "votes not cast",
totals 12,591,495. I could perhaps believe if the numbers were the other
way around, if the total voters included tallies of spoiled ballots. But
having more people vote for president than total voters doesn't make sense.
My call to the California SoS office on this issue was unreturned and I
never followed up.
http://www.ss.ca.gov/elections/sov/2004_general/contents.htm
I want to give proper credit, though I can't remember who first discovered
these errors. Joe Lenski, Kim Brace, and I were all sorting through the
numbers as they were released by the states, though I am almost certain that
Joe found the DeSoto County error. Even as late as February I recall states
were still amending their certified results, primarily as they internally
found errors when they responded to the Election Assistance Commission's
Election Day Survey. I didn't keep a real-time track of the amended results
as they were posted since there were simply too many.
And foreshadowing being a key to good literature....undoubtedly, there are
many other instances where the numbers don't add up.
==================================
Dr. Michael P. McDonald
Brookings Institution, Visiting Fellow
George Mason University, Assistant Professor
Dept of Public and International Affairs
4400 University Drive - 3F4
Fairfax, VA 22030-4444
Office: 703-993-4191
Fax: 703-993-1399
mmcdon@gmu.edu
http://elections.gmu.edu/