Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: problem using "vselect"

Subject   Re: st: problem using "vselect"
Date   Fri, 28 Jun 2013 16:41:01 -0500

Matthew Hamilton <> has a problem using the -vselect-

>Hi, I'm using the vselect package as follows:
>vselect depvar varlist, fix(indepvar1 indepvar2) best
>"Varlist" stands for a long list of candidate predictors. This list does not
>include indepvar1, indepvar2, or depvar. My understanding is that the option
>fix(indepvar1 indepvar2) forces the algorithm to give the "best" model that
>contains those two variables for each quantity of predictors. For example:
>vselect logI2010pc logGCI7_LaborMarketEfficiency
>logGovcontrolofcorruption logLFPR2010, fix(logGDP2010pc logWA2010pc)
>However, I keep getting the error: "The Fixed variables overlap with
>the predictors/response."
>Based on my reading of vselect.ado, I am pretty sure it is supposed to
>give this error when varlist and the fix() vars are not disjoint.
>What's going on? Could this be a bug in vselect?

We were unable to replicate Matthew's error.  -vselect- was updated in 2011 to
correct a bug related to the -fix()- option.  Maybe Matthew's version of
-vselect- is out of date.   Matthew can get the current version of -vselect- 
by typing

  . adoupdate st0213, update

Charles Lindsey
Senior Statistician and Software Developer
*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index