Preliminary Report, President's Committee on the Statistical Issues in Presidential Elections

• Activity Number: 64
• Type: Invited
• Date/Time: Sunday, August 3, 2003 : 6:00 PM to 8:00 PM
• Sponsor: Section on Statistical Computing
• Author(s): Lewis Henry LaRue*+
• Companies: Washington and Lee University

Abstract

The election process is a repetitive stochastic process that is best analyzed by breaking it down into three parts: registration, voting, and counting. In each of these steps there can be "misses" and "false-positives." The registration process should produce a list of voters at the polling place that lists only and all of those who are entitled to vote. If someone who is qualified to vote is omitted from the list that is delivered to the polling place, then we have a "miss." Conversely, if someone is on the final list but is not entitled to be, then we have a "false-positive." In voting, the voter can "miss" by failing to record properly the candidate for which he or she intends to vote. And conversely, the voter can record a "false-positive" by marking the apparatus for the wrong person. To make matters even more complicated, both of these errors can occur simultaneously or merely singly. In counting and recounting, the same sort of errors that the voters make can be made again (and independently so) by the machines and the people who tabulate the votes. A detailed empirical study would determine the likelihood that these types of errors will occur and would establish ways to reduce them.