Lesson Zero: Lecturing, Churchill, and End of the World

Teaching as Performance

The class is two weeks of solid statistics. Eight hours a day weeks; save the weekends. I don’t know about other instructors, but my technique is to try to will the information into the students. It is exhausting.

There is little to none of classical, frequentist theory that I like or agree with, but I make sure that everybody understands it. Not the math or the formulae, but the interpretation, especially for confidence intervals and p-values, which I promise that nobody ever remembers. For example, I have not yet run into a non-statistician user of classical methods who can recollect the screwy meaning of confidence intervals. The troubles with over reliance on p-values is well known.

Oh, sure, many can remember to use classical methods—everybody can tell the difference between estimation of metaphysical, non-observable (conditional on model correctness) parameters and hypothesis testing—but they can’t remember what these things really mean. And as such, they are too confident in their inferences.

Part of the difficulty is our reliance on canned examples, which seem to make sense to the student in the classroom, but which perplex him out of it. And part of the problem is with our insistence on treating our subject as a branch of mathematics. We’re happy to see our students to hand calculating tables of sums of squares etc., but the only thing they bring away from this is knowledge of how their calculators work. Why they are doing it is never quite assimilated.

I try and keep the canned examples to the minimum and make students create their own problems. Particularly, I require each of them design a project to answer a question that interests them. If data is required (it almost always is; but our field is not just about data), then they have to collect it. They have to make their own mini-databases (spreadsheets), they must figure out what each variable is and how it relates to the question at hand.

They’ll then naturally develop a feeling for what the data is and does, they’ll learn the instinct that the compromises in the data should make them less certain about their results. Best of all, they’ll figure out how to ask proper questions.

Like I said, we do this over two weeks. So it’s certainly possible to implement this scheme into a semester-length course.

Winston Smiths Churchills No More

The health police from the Ministry of Truth in England have airbrushed the cigar from Winston Churchill’s hand.

Presumably, some raving bureaucrat worked himself into a frenzy over the idea that a prominent figure could be seen smoking, and she Photoshopped a cigar from between Winston’s fingers. He then plastered an enormous poster of the bowdlerized photo over a museum devoted to World War II.

I’d be willing to bet a substantial amount of money that whoever did this—everybody is now denying it—is a good little socialist who has “freedom” on his lips. You know who I mean: the kind of person who uses phrases like “victimless crimes” and has nightmares about armies of Christians led by George Bush installing a authoritarian state above them.

From that, he works out the theory that smoking should be banned because smokers aren’t smart enough to know that they should not be smoking. But he—he himself—has risen about the rest of us and have reached Intellectual Enlightenment. He knows what is best.

World to End; Reporter Indifferent

Frank Fenner, “emeritus professor of microbiology at the Australian National University and the man who helped eradicate smallpox” says “Homo sapiens will become extinct, perhaps within 100 years.”

Why? Well, climate change of course. And overpopulation. Also, “unbridled consumption.”

Somebody surely has written a book with a title along the lines of “Why Smart People Say Stupid Things.” It’s not the end-of-the-world predictions don’t have their place—they are fun to contemplate—but the mechanism of the apocalypse should at least make sense; it should be internally coherent and logically consistent with itself.

Any theory which predicts our demise because of “unbridled compromise” doesn’t meet these criteria. Take food. If there is no food, then there can be no breeding. You cannot breed until you run out of food and then everybody dies.

It’s true that in particular times and places, crops go bad and people starve and die. But as they are doing so, they are not producing babies, which could in turn eat more and then kill everybody. It just cannot happen.

It may be true—it almost certainly is—that there can only be so much food, such that more than a certain number of humans could exist. We may even be at that point, though this is doubtful. And even if we are, it means that we have enough food for the number of mouths we have. If that quantify of food should decrease—through war, helpful bureaucracy, disease, socialism, etc.—then the maximum number of humans will decrease. But if it should increase, then our theoretical maximum increases too.