Sherman Dorn

On the proper state of being bothered

Are you bothered?

Seasonal bother: It's summer in Florida, and if you park a car anywhere outside a meat locker, touching a steering wheel earns you a second-degree burn.

Caffeinated bother: I started the day a little after 7 am at a local coffee shop, grading student papers. My brain fried about 210 minutes later, after a few cups of coffee and my getting to the point where two-thirds of the papers are now read (no, not two-thirds read this morning).

Unreasonable bother 1: I'm at a public library, where a children's program started in one of the library's rooms an hour ago, and one of my fellow (adult) patrons was bothered that there might possibly be a crying child anywhere in the building whom he could hear. (The child was taken out into the hallway reasonably quickly.)

Political/policy bother: Ezra Klein (along with Matthew Yglesias) seems to understand the long-term game of the Obama administration on health care (among other issues). Unfortunately, most reporters still don't get it, about health-care politics or, to pick another random topic except that it's my interest, education politics. It's too much fun to report the latest (wording-dependent) poll results or the latest pronouncements by the diva du jour. 

Unreasonable bother 2: TMI in the library. You really don't want to know (and neither did I). But in my head and heart, I know that I'd rather be bothered in the public library than not have a public library.

Intellectual bother: The popular philosopher's text by Howson and Urbach on Bayesian reasoning troubles me, less because of its style (which is fine, if dense for us nonphilosophers) or omissions (which I will trust statisticians can correct) than because of the disturbing but sensible point early in the book and that Steven Goodman has described as the p-value fallacy: statistical tests of significance say nothing about the probability of ruling in or out various hypotheses. If I understand Howson and Urbach's analogy between the standard discussion of medical tests and inferential statistics, the conditional probability of any hypothesis (after gathering data) depends not just on the inferential equivalent of false-positive rates (tied to statistical significance and p-values) or the equivalent of false-negative rates (power) but also on the underlying probability of the hypothesis being true. I pondered this last night while cleaning the kitchen, and the small point got under my skin. On what basis would a non-Bayesian (frequentist) respond? If I remember correctly, the easy response is to say, "Ah, a frequentist perspective is close to a Bayesian one with a non-informative prior." Except that the prior for categoricals, even with a non-informative assumption, depends on the number of bins, or hypotheses being tested. I think that the only way out for a frequentist is to either artificially restrict the number of hypotheses or to not care about the number of hypotheses being compared. To answer a question Gene Glass asked me a few years ago, it's just about at this point that my brain begins to dribble out my ears: historians are generally not theoretically minded. 

Unreasonable bother 3: I need to concentrate on an article that's already late, but rewinding to 7 am and having the whole day over again to work on the article as well as grading? Not going to happen.

Why bother: decaf nonfat latte with sugar-free flavoring, no whip.