Mommy, where do numbers come from?

One of the nice things about the internet is that, if you look hard enough, you can find people addressing important questions about where statistics (in this case, the plural of statistic) come from.

Over at MacroMania (via Thoma), David Andolfatto talks about how we measure unemployment:

So this is how it works. You are asked in a survey whether you have done any paid work in the previous four weeks. If you answer yes, then you are counted as employed. If you answer no, then you are nonemployed.

If you are nonemployed, you are asked another question: Have you been doing anything in the past four weeks to find work (and if so, what)? If your job search activity is judged to be sufficiently "active," then you are counted as unemployed; otherwise, you are counted as a nonparticipant.

This suggests an interesting point. It would appear that when we talk about job seekers (as in the ratio of job seekers to job openings), we may be excluding the severely underemployed which means things are even worse than they look.

Is this something we should worry about? Damned if I know. That's a question for an econometrician. But it is a good question to ask.

Another good question is who decides on categories and definitions get used when putting together statistics. There have been a lot of pixels spilled about this hypothetical taxpayer receipt put out by the Third Way think tank, but so far as I know, Andrew Gelman was the first to notice this:

I'm more concerned about the list itself, though. I think a lot of cognitive-perceptual effects are involved in what gets a separate line item, and what doesn't. For example, I see the FBI but not the CIA, the NSA, or weapons procurement. There's a line for "salary and benefits for members of Congress" but nothing for the courts system or the White House. And so on. So, while I agree with [Megan] McArdle that "more information is generally better," I'm not quite sure how to get there. I'd be very very suspicious of the choice of items that happens to end up included on the hypothetical itemized tax bill. Especially If it's really true that people don't notice those boxes on their W-2 form with FICA and Medicare payments, I also seem to recall seeing some glossy government documents with charts showing where the money is coming from and where it goes. Maybe there's some place other than a W-2 form to put this information where people will notice it.