Since we had discussed some of the issues with the Bureau of Labor Statistics’s employment situation reports, I thought this report was worth mentioning. At the New York Post John Crudele reports about some fraud on the part of Census Bureau employees who take the surveys used by the BLS in their report:
Let’s look at the one case that set Butler [Ed.: a whistleblower working with Crudele] off — the cheating done by Buckmon, who is now deceased.
Buckmon would complete more than 100 cases in 10 days — more than three times his peers.
It turns out that Buckmon wasn’t actually surveying people. He was making up data and collecting overtime to account for the time it would have taken to get those interviews.
And this one man alone was cheating on such a large scale that it could affect the national jobless numbers since the bureau’s Current Population Survey is scientifically weighted and each response counts as 5,000 households.
So Buckmon’s 100 cases equaled 500,000 households — and he wasn’t the only one caught faking data, Butler said.
The individual BLS employment situation reports don’t stand alone. Each ESR is dependent on those taken before. One bad survey means that those after it are questionable, maybe by just a little bit, maybe by a lot—keep in mind that the actual changes that we have been seeing have been relatively small. Lots of surveys being fraudulent calls the entire report into question. And with a sample size of 100 for a population of 500,000 the margin of error is already greater than 5% with a confidence level of 95%, suggesting that the confidence level they’re going for is lower than that.
That the fraud introduced by that one individual wasn’t detected in a routine review of the data suggests to me that the entire process is lacking.
