Sunday, September 17, 2006

Libnews' Editorializing is Very Much Appreciated Right Now

"Finally, we apologize for the lengthy and editorializing post, but the misuse and abuse of statistics is near the top of our list of pet peeves, right up there with cyclists who don’t stop at red lights, people who toss their dog droppings in our garbage and the new Transport Canada regulations that consider bottled water to be dangerous."

That quote came from, and while they might have apologize a little for it, I couldn't be happier with their gigantic and brilliant post. I concur wholeheartedly with Libnews: I hate the misinterpretation of statistics. To be more technical and more specific, I hate it when individuals analyze point estimates without any understanding of the properties of the estimators. I hate it when individuals analyze only the first moment of a set of data, not the higher moments.

But at the same time, I have to admit, it's kind of fun to make conclusions from data. I do it all the time, and it's wrong, but darn it, it's just so much fun.

Anyways, I think libnews for their comprehensive list of arguments for and against every candidate. It saves the Liberal blogosphere from having realistically 20-odd--and potentially several dozen--comparisons, as the debate over the polls' findings is replayed between supporters of every potential pair of candidates candidates (much as Lobster Thermidor and I have already done in one instance.)

In any event, while it's usually the case that libnews reports on our blogs, I wanted to reverse that for a moment and throw my kudos out to libnews for their brilliant post.

I do, however, apologize in advance for the inevitable future when I continue to apply Coase's maxim. As is evident from my blog, I have an unhealthy obsession with torturing data, as I try to draw out whatever analysis I can, using whatever assumptions are necessary along the way.

I can only hope that, while I might get the information I want from the torturing, I make it painfully clear that that information is of limited use and quality and that not terribly much ought to be inferred from it.

And that's the problem with torture. You might get the information you need out of it, and you might be dead on, but you'll never know; you've had to torture so much that you don't know whether it's actually giving up the information or whether it's giving it up because it's had enough of the torture.

My solution is not to avoid torture. My solution is to act cautiously on the information that torture provides.


Post a Comment

<< Home