I'm compiling some comparisons
of 2006-08 ACS and 2000 Census data, and I'm looking for some guidance in
estimating standard errors for 2000 CTPP tables. For the place of residence
data, I have applied the estimation procedures outlined in the 2000 SF3
documentation:
Unadjusted
standard error:
SE(Y) = sqrt(5*Y*(1-Y/N)),
where N =
size of publication area (total population) and Y = estimate of characteristic
total
This value is then multiplied
by a design factor determined by the state, sample rate and
characteristic.
Are these methods also
applicable to the workplace (CTPP part 2) tables? If so, what assumptions can I
make about sample rates, and are the standard error design factors applicable to
place of work statistics?
Any advice will be greatly
appreciated.