rii [if] [in] , impvar(varname) [robust] : estimate command
rii is a prefix command that runs multiple imputations of a model based on the value of the multiple imputation variable. rii has been tested on: probit, tobit, cnreg, and regress. rii uses the repeated-imputation inference (RII) technique to derive OLS regression coefficients, estimates of the standard error of the coefficients, test statistics for individual coefficients and a model test statistic. RII is based on Bayesian theory, and is applicable to linear and nonlinear models, and to models estimated by both least squares and maximum likelihood. The RII technique incorporates the variability in the data due to missing data into the estimate of the standard error of the mean.
Note: The command submitted to rii cannot contain an if or in since rii subsets the data to each value of the imputation variable. if and/or in can be use in rii like so:
. rii if mvar <= 5 in 1/50000 , imp(mvar): regress y x1 x2 x3 x4 year89 year90
. rii if mvar <= 5 , imp(mvar): regress y x1 x2 x3 x4 year89 year90
. rii in 1/50000 , imp(mvar): regress y x1 x2 x3 x4 year89 year90
Paper rii is based on
Montalto, Catherine Phillips & Sung, Jaimie. (1996). Multiple imputation in the 1992 Survey of Consumer Finances. Financial Counseling and Planning, Vol. 7, pp. 133-46.
impvar(varname) is a required option that specifies a numeric variable that identifies the multiple imputations with which to run the submitted command.
rii saves the following in e():
Scalars e(N_imps) number of imputations e(N) number of observations
Macros e(depvar) name of dependent variable e(prefix) rii e(command) command run by rii e(cmdline) command as typed e(cmdname) name of command run by rii e(cmd) name of command run by rii e(properties) b V
Matrices e(b) coefficient vector e(V) variance-covariance matrix of the estimators e(obs_per_imp) number of observations per imputation
Functions e(sample) marks estimation sample
. rii , imp(mvar): probit x1 x2 x3, robust
. rii if mvar < 6, imp(mvar): tobit amtspt x1 x2 x3 x4, ll(0) ul(25000)
. rii , imp(mvar): cnreg y x1 x2 x3 x4 year89 year90, censored(cen)
. rii , imp(mvar): reg y x1 x2 [pweight=pop]
Instead of using dprobit use probit and then run mfx:
. rii , impvar(mvar): probit y x1 x2 x3
David T. Robinson Center for Entrepreneurship and Innovation Duke University's Fuqua School of Business firstname.lastname@example.org Dan Blanchette The Carolina Population Center University of North Carolina - Chapel Hill, USA email@example.com
On-line: probit, tobit, cnreg, and regress