Standardized Mean Differences

[pink_poodle]

Hi friends,

I am doing the following:

           1. Obtaining propensity scores 

           2. Converting them to model weights

           3. Running the model

The model runs as follows:

PROC SURVEYREG DATA = ps;
	MODEL dyhosp = grp; 
	WEIGHT IPS;
RUN;

I need to compute Standardized Mean Differences (SMDs) as the fourth step.  Please advise.

Many thanks,

Eva

[StatDave · Accepted solution]

Standardized mean differences can be computed in PROC PSMATCH which performs propensity scoring. See the Getting Started example in the PSMATCH documentation.

[sbxkoenk]

Do a Google search:

Standardized Mean Difference site:communities.sas.com

BR, Koen

[Ksharp]

You could calculate Cohen's d statistic to get a standardized mean difference.
Check Rick's blog:

https://blogs.sas.com/content/iml/2025/08/11/cohens-d-statistic-in-sas.html