Reply #11: potential bias - the clumping by geo area must be reflected in the

[papau]

weightings for the stratified sample -

A Nat'l poll has fewer hurdles to jump to get to a proper weighting - this is a bit more work

The bias (a bad word no doubt) is the fact more work means more chance of error.

In anycase what you are saying is that the 4% error - undercount - that always pops up (1980, then 1990, and I forget 2000) is based on a household survey like this one for jobs? interesting.

the employment picture - the jobs growth - is always taken from the payroll survey (and is quoted in histories of Reagan Admin, Clinton Admin, etc)- and the look backs that I am aware of do not show a change in workforce jobs to a higher percentage of folks making a real income from home in the 90's. The "put a job on your resume when the hiring sucks" is the only explanation that rings true - to me - as an explanation of the higher household survey count at this point in time.