We found out yesterday that the nation's unemployment rate slipped downward firom 8.3% to 8.2% last month. You probably think you know what that means, but do you really?
The unemployment rate is calculated each month from a survey of 60,000 households. Don't believe me? Read this. The net jobs addition is also calculated from a survey, but in that case of employers.
Just like any opinion survey, mathematics dictates that there must be sampling error -- the margin of error reported with every other opinion survey you read about -- but the Bureau of Labor Statistics never bothers to mention it.
It's relatively easy to figure out the margin of error for the household employment survey from numbers that you know. Since a survey with a sample size of 1000 has a margin of error of roughly 3%, and since the margin of error varies inversely with the square root of the sample size, the margin of error in the unemployment rate is about 1/sqrt(60) x 3% ~ 3%/8 ~ 0.4%.
That raises a really big question: why doesn't the headline-generating unemployment rate rattle around more from month to month? Fluctuations of 0.1% to 0.2% are to be expected just from random sampling effects.
One possibility is that it does rattle but that the Bureau of Labor Statistics smooths the numbers based on some floating average. Another is that the bureau surveys the same folks every month -- a so-called longitudinal survey -- though that would be fraught with error from failing to adjust the sample correctly for changing employment. Still another possibility is that they adjust the result to fit a predetermined ethnographic and economic profile -- this many whites, that many Hispanics, so many more blacks, this many blue collars, that many white collars, so many professionals -- and so on, and this gives them enough control to produce a smoother result. But any such scheme introduces a different kind of error while it suppresses the sampling error: a general shift of the adjusted sample average away from the population average, known as bias.
Another possibility is that a political appointee somewhere up the food chain from the Bureau's statisticians puts his finger into the wind, then calls his boss, who puts his finger in wind and calls his boss, who puts his finger in the wind, and so on, and so on, till they reach the President.
Who puts his finger in the wind by looking at his re-election campaign's internal polling, and....
