An interesting thing came up in the CA recall. There were certain also-ran candidates who did well primarily because they were next to the leaders on most, if not all, ballots. Schwartzman, Burton and McClain. All 3 finished well ahead of the pack. See Mickey Kaus and Election Law and Volokh
Now, assume that almost all votes for Schwartzman were intended for Arnold, but mismarked in a Florida-butterfly-ballot manner. Since the vote is reported by county, and voting machines are also chosen by county, one could possibly infer a machine-induced voter error rate from this data.
Garry Young did this and found the following Schwartzman vote totals:
Punchcard counties: 5336 (0.166%)
Other counties: 5621 (0.125%)
The difference here between puchcard and other voting methods (if statistically meaningful, and I'm not convinced) amounts to 410 additional voter errors per million votes cast, caused by an increased tendency to mispunch the adjacent line.
Is this a problem? You be the judge.
Could it be that people make mistakes, too?!?
"I want to vote for that Schwarz-man"
"Look for the Mick and vote for him"
"On the indian mailer, John Burton told me to look for the letter 'B' then vote for someone..."
All plausible scenarios...
Posted by: BoiFromTroy at October 10, 2003 01:35 PMIt's a problem only in a close race.
Posted by: John Jorsett at October 11, 2003 07:33 AMI would have thought that Schwartzman and Burton's strongest points were their misleading names. This could also explain why the other Davis (Scott, the murder suspect) ranked near the bottom of the list, and why the fake Bob Dole did even worse in this state than the real Bob Dole did in 1996.
Posted by: Xrlq at October 15, 2003 06:11 PM