ASA Connect

A colleague posted this on the INFORMS Open Forum, and I thought it would be an interesting, timely, and relevant topic for the ASA Community.  Dr. Wikum's post follows.  We'd appreciate your thoughts.

I recently authored an article titled "A Decidedly Insufficient Statistical Analysis" that appeared in the January 2023 issue of OR/MS Today. Here's a link to the article:   https://pubsonline.informs.org/do/10.1287/orms.2023.01.05/full/

The article describes legal proceedings that occurred after the 2022 Arizona gubernatorial election, won by Katie Hobbs, but contested by Kari Lake. Among the witnesses at a legal hearing was Richard Baris, director of Big Data Poll, who argued based on statistical analysis that but for problems with voting machines on election day, Kari Lake would have won. The title of my article was drawn from the legal opinion of the presiding judge, who did not find in Baris's or others' testimony clear and convincing evidence of misdeeds that affected the result of the election. That ruling was subsequently affirmed by the Arizona Court of Appeals.

How might statistical or other analysis provide clear and convincing evidence of anomalies or misconduct in an election sufficient to change the election results?

Have you ever testified before a court of law, governmental agency, etc. concerning results of modeling or analysis? If so, please share a brief description of the setting and your testimony. In testifying, what issues did you encounter regarding professional ethics?

I welcome your response, either in this forum or via email to me, but I request that in doing so, you refrain from political commentary.