Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | weddings@stata.com |
To | statalist@hsphsun2.harvard.edu |
Subject | Re: st: mvn impute - collinear imputation (dependent) variables detected |
Date | Mon, 17 Oct 2011 17:38:18 -0500 |
Raquel Guimaraes <raquelrguimadem@gmail.com> asked about using -mi impute mvn- with collinear imputation variables: > I need help using multiple imputation command for Multivariate Normal > Distribution. I did this before and worked perfectly. Now, unfortunately, I am > getting the following error message: collinear imputation (dependent) > variables detected. Below you may find the piece of do-file. (snip) > mi impute mvn: collinear imputation (dependent) variables detected > Well, what I am supposed to do? I have only one dependent variable (leitura), > and I can not drop it. The dependent variables in -mi impute mvn- are actually the variables to be imputed---the left-hand-side variables. Raquel will need to remove the collinear variable "climadisciplinar_mean" from the left-hand side. Even though the note in the output says that the variable was "omitted because of collinearity," the variable has not yet been omitted. The note is a generic note printed by the command -_rmcoll-, which checks for collinearity. Collinear variables on the right-hand side are automatically omitted, but left-hand-side variables are not. The note is printed to help users decide on their own which left-hand-side variables to remove. If Raquel would rather keep the variable "climadisciplinar_mean" in the model, she can instead drop other collinear imputation variables. Raquel can use the command -_rmcoll- on any list of variables to determine which variables in the list form a non-collinear set. --Wes weddings@stata.com * * For searches and help try: * http://www.stata.com/help.cgi?search * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/