Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | "Ariel Linden, DrPH" <ariel.linden@gmail.com> |
To | <statalist@hsphsun2.harvard.edu> |
Subject | re: st: Pstest bias reduction |
Date | Mon, 26 Mar 2012 10:35:05 -0400 |
Nick, now that you have been properly chastised for following the rules of the statalist, let me respond to your question. -pstest- (a user written program by Edwin Leuven & Barbara Sianesi and found on ssc) uses the logic set forth in the paper by Rosenbaum and Rubin (1985) listed below by Daniel. Specifically, the standardized residual is first calculated on the total unadjusted sample using the formula: difference in means/pooled standard deviation. Next, the standardized residual is then calculated on the reduced subset (matched set of treatment vs controls) using the new difference in means as the numerator and the original (total sample) pooled standard deviation in the denominator. Thus, holding the variability constant, the difference in the first and second calculations represents the " bias reduction." I hope this helps Ariel Date: Sun, 25 Mar 2012 19:04:13 -0400 From: Nick Sauers <nck0208@gmail.com> Subject: st: Pstest bias reduction I used Stata for propensity score matching.  After completing my match, I ran the Pstest command, and I'm unable to interpret the column that reports bias. How is bias calculated in Pstest? I've tried multiple calculations, and my numbers don't align with the  output. Thanks! Nick Sauers nck0208@gmail.com Date: Mon, 26 Mar 2012 02:50:00 +0200 From: daniel klein <klein.daniel.81@googlemail.com> Subject: st: Pstest bias reduction Nick, as recently stated (http://www.stata.com/statalist/archive/2012-03/msg00513.html) there is a convention on Statalist to use - - to indicate Stata code or commands, which in your case means write -pstest- not Pstest. More important, you are also asked to explain where user-written software comes from (http://www.stata.com/support/faqs/res/statalist.html#stata). -pstest- comes with the -psmatch2- package from SSC. The FAQ also asks you to read the online help before posting a question (http://www.stata.com/support/faqs/res/statalist.html#before). In your case the help file for -pstest- states "(b) [...] The standardised bias is the difference of the sample means in the treated and non-treated (full or matched) sub-samples as a percentage of the square root of the average of the sample variances in the treated and non-treated groups (formulae from Rosenbaum and Rubin, 1985)." along with the full reference Rosenbaum, P.R. and Rubin, D.B. (1985), "Constructing a Control Group Using Multivariate Matched Sampling Methods that Incorporate the Propensity Score", The American Statistician 39(1), 33-38. The article is found here: http://www.stat.duke.edu/~banks/130-lectures.dir/readbyNov22.pdf If the article does not provide an answer to your question, consider a more detailed explanation of what exactly you did when trying "multiple calculations". Best Daniel * * 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/