Hi there,

I am trying to develop tests of difference for 2 unpaired sets of data from cross-sectional surveys that both used clustering. I want to simply test whether there is a statistical difference from point 1 to point 2 (groups 1 and 2).

I modified the data and produced this output:


. prtest eb, by(svy) cluster(Cluster) rho(0.02)

Two-sample test of proportions
Cluster variable: Cluster

Group: 1 Group: 2
Number of obs = 334 Number of obs = 450
Number of clusters = 38 Number of clusters = 38
Avg. cluster size = 8.79 Avg. cluster size = 11.84
CV cluster size = 0.6676 CV cluster size = 0.5512
Intraclass corr. = 0.0200 Intraclass corr. = 0.0200

------------------------------------------------------------------------------
Group | Mean Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
1 | .5479042 .0302535 .4886085 .6071999
2 | .1866667 .0208523 .1457969 .2275364
-------------+----------------------------------------------------------------
diff | .3612375 .0367436 .2892214 .4332536
| under Ho: .0383799 9.41 0.000
------------------------------------------------------------------------------
diff = prop(1) - prop(2) z = 9.4122
Ho: diff = 0

Ha: diff < 0 Ha: diff != 0 Ha: diff > 0
Pr(Z < z) = 1.0000 Pr(|Z| > |z|) = 0.0000 Pr(Z > z) = 0.0000


It was suggested to use svy: tabulate twoway, however I was wondering if this would give a similar result. In the example given, I provided dummy ICC. What is the best way to obtain ICCs for the variables?

Thank you for your help in advance!
-Brett