Brent Fulton asked:--
> > I've run the following and can examine if the e.g., 95% CI's overlap,
but
> > would like to calculate the p-value that the means are equal.
> > .svy: mean y
> > .svy: mean y, subpop(Michigan_dummy)
> >
> > Is there a post-estimation test that can compare the survey-based
means
> > above?
How about -suest- followed by -test-. Below is an example comparing the
proportion obese (0=nonobese, 1=obese) for the US vs. Michigan. The data
come from a national telephone survey. The test is first done using
-svy:regress- (constant only model), then using -svy:logit- (constant only
model).
svyset [pweight=finalwt2];
pweight: finalwt2
VCE: linearized
Strata 1: <one>
SU 1: <observations>
FPC 1: <zero>
. svy: reg bmi2;
(running regress on estimation sample)
Survey: Linear regression
Number of strata = 1 Number of obs = 2764
Number of PSUs = 2764 Population size =
1.240e+08
Design df = 2763
F( 0, 2763) = .
Prob > F = .
R-squared =
0.0000
------------------------------------------------------------------------------
| Linearized
bmi2 | Coef. Std. Err. t P>|t| [95% Conf.
Interval]
-------------+----------------------------------------------------------------
_cons | .1924716 .0085171 22.60 0.000 .175771
.2091722
------------------------------------------------------------------------------
. estimates store a1;
. svy, subpop(mi): reg bmi2;
(running regress on estimation sample)
Survey: Linear regression
Number of strata = 1 Number of obs = 2764
Number of PSUs = 2764 Population size =
1.240e+08
Subpop. no. of obs = 96
Subpop. size =
3992188.9
Design df = 2763
F( 0, 2763) = .
Prob > F = .
R-squared =
0.0000
------------------------------------------------------------------------------
| Linearized
bmi2 | Coef. Std. Err. t P>|t| [95% Conf.
Interval]
-------------+----------------------------------------------------------------
_cons | .1587098 .0388779 4.08 0.000 .0824771
.2349425
------------------------------------------------------------------------------
. estimates store a2;
. suest a1 a2;
Simultaneous survey results for a1, a2
Number of strata = 1 Number of obs = 2764
Number of PSUs = 2764 Population size =
1.240e+08
Design df = 2763
------------------------------------------------------------------------------
| Linearized
| Coef. Std. Err. t P>|t| [95% Conf.
Interval]
-------------+----------------------------------------------------------------
a1 |
_cons | .1924716 .0085171 22.60 0.000 .175771
.2091722
-------------+----------------------------------------------------------------
a2 |
_cons | .1587098 .0388779 4.08 0.000 .0824771
.2349425
------------------------------------------------------------------------------
. test [a1]_cons=[a2]_cons;
Adjusted Wald test
( 1) [a1]_cons - [a2]_cons = 0
F( 1, 2763) = 0.77
Prob > F = 0.3812
. svy: logit bmi2;
(running logit on estimation sample)
Survey: Logistic regression
Number of strata = 1 Number of obs = 2764
Number of PSUs = 2764 Population size =
1.240e+08
Design df = 2763
F( 0, 2763) = .
Prob > F = .
------------------------------------------------------------------------------
| Linearized
bmi2 | Coef. Std. Err. t P>|t| [95% Conf.
Interval]
-------------+----------------------------------------------------------------
_cons | -1.43403 .0547986 -26.17 0.000 -1.54148
-1.326579
------------------------------------------------------------------------------
. estimates store a3;
. svy, subpop(mi): logit bmi2;
(running logit on estimation sample)
Survey: Logistic regression
Number of strata = 1 Number of obs = 2764
Number of PSUs = 2764 Population size =
1.240e+08
Subpop. no. of obs = 96
Subpop. size =
3992188.9
Design df = 2763
F( 0, 2763) = .
Prob > F = .
------------------------------------------------------------------------------
| Linearized
bmi2 | Coef. Std. Err. t P>|t| [95% Conf.
Interval]
-------------+----------------------------------------------------------------
_cons | -1.667859 .2911746 -5.73 0.000 -2.238801
-1.096918
------------------------------------------------------------------------------
. estimates store a4;
. suest a3 a4;
Simultaneous survey results for a3, a4
Number of strata = 1 Number of obs = 2764
Number of PSUs = 2764 Population size =
1.240e+08
Design df = 2763
------------------------------------------------------------------------------
| Linearized
| Coef. Std. Err. t P>|t| [95% Conf.
Interval]
-------------+----------------------------------------------------------------
a3 |
_cons | -1.43403 .0547986 -26.17 0.000 -1.54148
-1.326579
-------------+----------------------------------------------------------------
a4 |
_cons | -1.667859 .2911746 -5.73 0.000 -2.238801
-1.096918
------------------------------------------------------------------------------
. test [a3]_cons=[a4]_cons;
Adjusted Wald test
( 1) [a3]_cons - [a4]_cons = 0
F( 1, 2763) = 0.66
Prob > F = 0.4172
****************************************************************
Michael R. Frone, Ph.D.
Senior Research Scientist
Research Institute on Addictions
State University of New York at Buffalo
1021 Main Street
Buffalo, New York 14203
Office: 716-887-2519
Fax: 716-887-2477
E-mail: [email protected]
Internet: http://www.ria.buffalo.edu/profiles/frone.html
****************************************************************
*
* 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/
