Walter R. Mebane, Jr.
Associate Professor
Date: November 16, 2000
Contact Information:
Associate Professor
Department of
Government,
Cornell University
wrm1@cornell.edu
(phone: 607-255-3868)
HTTP://macht.arts.cornell.edu/wrm1/
This conclusion is based on several pieces of evidence. No single piece of evidence on its own is decisively convincing, but all together the body of evidence proves to a reasonable degree of scientific certainty that the election-day ballot frustrated the will of thousands of voters, so that their votes did not express their intent.
My conclusions about the features of the ballot are based on my extensive experience analyzing survey data and working with survey samples and survey interview instrumentation. At the University of Michigan during the 1980s I directed a pilot project to investigate whether it was possible to use sample survey instrumentation to trace the paths of individual Americans' political reasoning. I am knowledgeable about the design of instrumentation to be used to query people about their beliefs. Drawing on such expertise I can say with a high degree of scientific certainty that the format of and instructions on the two-page ``butterfly ballot'' used by voters at polling places on election day is likely to have led many voters to make mistakes in casting their votes, and the mistakes are likely to have followed one of a few characteristic patterns, including patterns of overvoting. The ``Report on Voting and Ballot Form in Palm Beach County'' by Dr. Henry E. Brady documents and explains numerous important defects in the design of the butterfly ballot that almost certainly confused many voters, in particular the formatting of the presidential candidates' names into two, vertically skewed columns and the positioning of the punch-holes. An additional feature of the instructions printed on the ballot that I judge to have contributed to overvoting is the instruction to ``Vote for a Group'' (Figure 1). Plainly the ``Group'' being referred to is the pair of candidates for President and Vice President for each party. The principal ambiguity in the instruction to vote for a group is that it is not clear from that wording whether one is to vote for the group as a whole or for each member of the group severally.
``Overvoting'' in the election in Palm Beach County refers to any situation in which for a single office more than one hole is pushed out on the ballot card. The patterns of overvoting mistakes most likely to have occurred because of the defects in the design of the ballot are ones that imply spoilage of a higher proportion of the ballots of those who favored Al Gore and Joe Lieberman (henceforth ``Gore-Lieberman'') than of those who favored George W. Bush and Dick Cheney (henceforth ``Bush-Cheney''). In particular, the most likely overvoting mistake is that a would-be Gore voter would produce a double punch in trying to correct a vote mistakenly cast for Pat Buchanan and Ezola Foster (henceforth ``Buchanan-Foster''). Doing that would produce punches on the holes for Buchanan-Foster and Gore-Lieberman. Next most likely is that a voter would punch two holes because of a mistaken idea that the instruction to ``Vote for a Group'' meant that it was necessary to punch one hole for the presidential candidate and one hole for the candidate for Vice President. Such an intention on the part of a would-be Bush-Cheney voter would produce punches for Bush-Cheney and Buchanan-Foster. Such an intention on the part of a would-be Gore-Lieberman voter would produce one of two pairs of punches: Buchanan-Foster and Gore-Lieberman; or Gore-Lieberman and David McReynolds and Mary Cal Hollis (henceforth ``McReynolds-Hollis'').
Absentee ballots used in the county lack the problematic combination of features that occur in the butterfly ballot (Figure 2). Like the butterfly ballot, the absentee ballot also includes the instruction to ``Vote for a Group.'' On this ballot, however, there is no confusing alignment of punch-holes in one column. The presence of a single number and arrow for choice, always placed to the right in both columns of names, disambiguates the reference to ``group'' and indicates clearly to the voter that the two members of each pair of candidates are to be treated as a unit.