Is the unemployment survey any better?

[Nords]

I stumbled across this short & old Business Week article:

"Up Front
Edited by Ira Sager
22 December 2003

Employment Is Up. No, Wait...
By Peter Coy
How much do we really know about the economy? The Bureau of Labor Statistics says that nonfarm payrolls 'rose slightly' in November 2003. But the BLS also concedes it's quite possible that payrolls actually fell -- or rose a lot.
What gives? The BLS surveys businesses that employ just 30% of all workers. There's a 10% chance it's wrong by 105,000 jobs on either side -- swamping the estimate of 57,000 new jobs. Says Robert Brusca, an economist at Native American Securities: 'The smart money doesn't react number to number.' No wonder."

Does anyone here know if the survey is still done the same way nearly three years later, or has BLS figured out a way to get below a six-figure error?

 

[Gone4Good]

I think the survey is still done pretty much the same way. Which is why all the attention to the monthly number is kind of silly. But in the larger sense, the survey is pretty accurate. With total employment around 144.3 million a 100,000 miss translates into an error of just 0.07% in the overall unemployment statistic.

 

[Nords]

I think the survey is still done pretty much the same way.  Which is why all the attention to the monthly number is kind of silly.  But in the larger sense, the survey is pretty accurate.  With total employment around 144.3 million a 100,000 miss translates into an error of just 0.07% in the overall unemployment statistic.

If it was a "total employment" number then I'd agree with your math.

The "total unemployment" number has a margin of error larger than its monthly variation. That little issue is usually ignored by the media and by people who should know better, which is why I was wondering if it's been fixed yet...