Oh, yes. I'm a Bayesian statistician. Let me remind you what that means. It means I begin with a set of "prior" probabilities, and I update them as new information becomes available. New information is available. Christine Ford, the formerly anonymous woman who wrote the letter to Diane Feinstein, has now come forward as Brett Kavanaugh's accuser. We also have new details about the history of the accusation. This leads to two types of Bayesian updating.
First, can I update my Bayesian assessment of Kavanaugh's guilt or innocence? Yesterday, I began with the baseline probability of Kavanaugh's guilt based entirely on the low probability of an accusation being false. I had nothing else. Today, we have more. Not only has Ford revealed her identity, there is history to the accusation. I don't put much stock in the polygraph. Those things are pretty much bullshit. However, the accusation predates the Kavanaugh confirmation by years, so this clearly isn't just an attempt to derail his nomination. The accusation was made in private, where Ford had nothing to gain, and there is a therapist's record. The therapist's records give some cover for anyone looking for discrepancies, including the question of four versus two attackers, although that could be explained. This isn't a perfect recording of events, but all things considered, update the probability of Kavanaugh's guilt upwards. Criminal conviction on this kind of thing? Statute of limitations aside, I don't know, but from a Bayesian point of view, I side with Ford.
Next, does this affect Kavanaugh's chances of confirmation? Yesterday, I was skeptical that anything could. I didn't consider the possibility of the anonymous woman coming forward. This does affect his chances. As of this morning, over at PredictIt, Collins's chances of voting for Kavanaugh are down to 64 cents on the dollar, with Murkowski at 53. Why the difference? Murkowski called for postponing the Judiciary Committee vote. She actually sounds like she might take this seriously. That's easier for her since she never dug that Kavanaugh hole for herself. Murkowski has always been the smarter of the two.
Yes, Kavanaugh really might lose. The odds still favor him, though. You may be tempted to look at those probabilities and start multiplying them together, but they are not, in statistical terminology, independent events. They will turn on the same factors. One of those factors is Donald Trump.
Very soon, Trump will begin making phone calls and scheduling visits with any wavering Republicans. The pitch will combine appeals to party loyalty with the need to defeat the Democrats on all things with threats. The threats will get ever more aggressive if anyone keeps wavering. If I had to bet right now, I wouldn't bet because I don't bet. I invest in a diversified portfolio, but I'd still say the slightly smarter money is on Kavanaugh getting confirmed. When in doubt, bet on Republican cowardice in the face of Donald Trump. Yes, a few Republican Senators are calling for delay. So far, none have announced outright opposition to Kavanaugh.
Christine, you are a brave woman. Challenge anyone in power, and try to force them to face the consequences of their actions, and you will face retaliation. That is an absolute, 100% guarantee. I am a statistician, and I don't give 100% guarantees. Except on this. Of course, Kamala Harris wasn't there during the Anita Hill hearings, so this could be much more interesting. Still, when in doubt, bet on Republican moral cowardice in the face of Donald Trump and other sexual predators.
Yeah, I was wrong on the Jones-Moore race in Alabama. I could be wrong on this too. Notice how uncertain my assessment is. So, who's gonna be the first Republican Senator to oppose Kavanaugh and face the wrath of a Trump tweet? Gee... oppose a rapist on the Supreme Court, or dodge a Trump tweet. Yeah, I can see how that would be a tough choice.