Random Group Variance Estimation --------------------------------
^rgroup^ ^"^command^"^ ^"^exp_list^"^ ^,^ ^rg^roups^(^varname^)^
^rgroup^ calculates the random group variance estimator described by Wolter (1985). This estimator is an alternative to other estimators of the variance of sample statistics for survey data e.g. the jackknife or taylor linearization. The application of the random group estimator requires, that random groups are already defined during data collection. This for example is the case in the GSOEP, the German Socio-economic Panel.
^rgroup^ runs the user-specified command for each group of the variable in ^rgroup()^ and uses the results to calculate the standard errors of the statistics in ^exp_list^.
As ^command^ any command may be used. Further all elements of that command are allowed.
Every accessible single statistic after that command may be used to build the exp_list. The expressions in exp_list must be separated by spaces and there must be no spaces within each expression. Note that command and exp_list must both be enclosed in double quotes.
The syntax of rgroup works like that of @bs@.
^rgroup^ stores its results r().
^rgroups(^varname^)^ specifies the variable, which contains the random groups.
. ^rgroup "sum xvar" "r(mean)", rg(rgroups)^ . ^rgroup "reg y x1 x2 x3" "_b[x1] _b[x2] _b[x3]", rg(rgroups)^ . ^rgroup "reg y x1 x2 x3" "e(r2)", rg(rgroups)^
Author ------ Ulrich Kohler ukohler@@sowi.uni-mannheim.de
Wolter (1985) "Introduction to Variance estimation", New York/Berlin (Springer)
Note for ^SOEP^ Users: The estimation is described in K.4
Acknowledgments --------------- rgroup.ado builds up on ranvar.ado, written by Frauke Kreuter, University of Konstanz and bs.ado, written by StataCorp.
Also see --------
Manual: ^[U] 23 Estimation and post-estimation commands^ ^[R] return, [R] saved results [R] estimation commands^ On-line: help for @est@, @postest@, @return@ , @ranvar@ (if installed)