Welcome (name, who presentation is for etc), 15 minute presentation. I work in Justice at the Scottish Government but the work for this presentation is not in the name of SG. Originally a part of research for Napier University.
A thought â many of you will have been to a particular university or school. Holding your circumstances and background constant, have you wondered what would have been the result if you had attended somewhere else ?
Here is the first part of the process flow which does the propensity score. When you click âRun process flowâ it provides a dialog box where you choose several options. These have been set up using parameters as macro variables â next slide
It shows that I have chosen the Caliper method and a distance of 0.0001 between scores. It will only match if the scores between treatment and control are âthis closeâ
The source dataset looks like this. It has 11 variables, a product field and a result field
So in PSM the first step is to use SAS Proc Logistic to obtain scores. Then sort the scores.
In PSM the next step after scoring, is to match (the scores in the treatments to the scores in the controls).
Here is the part of the process flow which does the matching. The parameterised code is used in the âmatching macroâ SAS code
Here are the results. The caliper method with a distance of 0.0001 has found that only 303 out of 980 can be matched between treatment and control.