If you’re waiting for Dino Rossi’s legal team to prove their allegations of fraud and ballot stuffing, don’t hold your breath, because today they finished presenting their evidence with a whimper, not a bang. Not even the GOP’s lead attorney is convinced, admitting:
“I cannot point to anybody who committed fraud. We’re just saying clearly the opportunity was there and it looks suspicious.”
Um… I’m not an attorney, but I’m guessing that doesn’t quite meet the standard of “clear and convincing.”
Much of the fourth day was spent in a “Frye hearing” in which both sides argued whether the Republican’s proposed proportional deduction methodology is accepted science. I guess I should provide a wrap-up of the days events, but I don’t really feel like it. I’m tired, I’ve got other things I’d rather write about, and besides… what the hell do I know? So instead I’m going to leave the wrap-up to DJ, an HA regular, and real life statistician.
The Frye hearing started out with Professor Katz on the stand, being grilled about the binomial and multinomial methods he proposed to “fix” the election. Katz came off as nervous (”squirrel-like” comes to mind), defensive and even evasive. Much of the testimony seemed to be Mr. Burman trying to get admissions out of Katz about limitations of his method for adjusting a statewide election. As he did in his deposition, Katz repeatedly resorted to “my analysis is conditional on the data” as a way to avoid discussing the non-representativeness of the GOP felon voter list. Burman had to press hard, but finally got Katz to admit that if the underlying assumptions are violated you can get the wrong answer. My (possibly biased) opinion is that Katz’s weaseling answers and telescopic interpretations of questions did not leave a good impression on Judge Bridges.
In contrast to Professor Katz, Professor Adolph was calm, friendly and composed. He gave an outstanding summary of the problems in Katz’s analysis. Adolph’s expertise is ecological inference, and he gave an excellent summary of the issues. He then provided two examples of a so-called “ecological fallacy.” The first concerned estimating Ichiro Suzuki’s batting average using league or team statistics. The second example was to estimate the sex of felons using Katz’s method of ecological inference, and sex proportions at the county level. The resulting 95% confidence interval was that between 44 to 50 percent of the felons were male. The true proportion is 75 percent