Arnold <[email protected]> asks about comparing a subpopulation mean
to the population mean:
> I would like to estimate a population mean and subpopulation means
> simultaneously, using -svymean- , in order to -svytest- selected subpop
> means against the population mean. Is there a straightforward way to do
> this? Alternatively, can results from one -svymean- estimation be preserved
> for -svytest- while running a second -svymean- estimation?
David <[email protected]> then makes the following suggestion:
> I have a more conceptual comment and suggestion rather than any Stata
> commands.
> I'm assuming that even the whole is a sample.
> It is hard for me (and probably for software) to figure out the right error
> term when comparing a part to the whole. However, I believe the answer to
> the question "Is the mean of this part different from the mean of the whole"
> is the same as the answer to the question "Is the mean of this part
> different from the mean of the rest of the whole?" That second question is
> easy to test. That's how I've done it and suggest doing it.
I can think of two ways that the above can be done:
1. -svymean, by()- followed by -test-
2. -svyregress-
Here is a short example of each in Stata 8 [mis]using the auto data; -price-
is the variable of interest, and -foreign- is the subpopulation indicator
variable:
***** BEGIN
. sysuse auto, clear
(1978 Automobile Data)
. svyset [pw=weight], psu(rep78)
pweight is weight
psu is rep78
. svyregress price for
Survey linear regression
pweight: weight Number of obs = 69
Strata: <one> Number of strata = 1
PSU: rep78 Number of PSUs = 5
Population size = 209210
F( 1, 4) = 0.18
Prob > F = 0.6957
R-squared = 0.0012
------------------------------------------------------------------------------
price | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
foreign | -263.5 626.6204 -0.42 0.696 -2003.277 1476.277
_cons | 6617.084 379.1009 17.45 0.000 5564.531 7669.637
------------------------------------------------------------------------------
. svymean price, by(for)
Survey mean estimation
pweight: weight Number of obs = 69
Strata: <one> Number of strata = 1
PSU: rep78 Number of PSUs = 5
Population size = 209210
------------------------------------------------------------------------------
Mean Subpop. | Estimate Std. Err. [95% Conf. Interval] Deff
---------------+--------------------------------------------------------------
price |
Domestic | 6617.084 379.1009 5564.531 7669.637 .661794
Foreign | 6353.584 270.4661 5602.65 7104.519 .2080977
------------------------------------------------------------------------------
. test _b[Domestic] = _b[Foreign]
Adjusted Wald test
( 1) [price]Domestic - [price]Foreign = 0
F( 1, 4) = 0.18
Prob > F = 0.6957
***** END
--Jeff
[email protected]
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
