Wednesday, April 03, 2013

There Is No Perfect Data, Population Version

Here's a analogy about comparing time series.

The U.S. population used to be known every 10 years, from the census. But now it's known every year, because we're a lot better at record-keeping and statistics and such.

Can one therefore not compare the US population of the last few decades to the population in 1790 or 1800 or 1810?

Do we just throw up our hands and say, we just don't know! We simply can't compare today's population time series to that of the early days of the country, because for all we know the U.S. population doubled in 1794 and then halved in 1795 and the census then wouldn't have noticed.

Of course we don't. We lay them beside each other and learn what we can, keeping the uncertainties and limitations of both in mind.

(Yes, I know this isn't a perfect analogy, because there are other records that might have recorded a sudden doubling of colonial population, but no such records for the Holocene thousands of years ago. It's just a rough analogy, OK?)

There is no perfect data.

No comments: