Propensity Scores Analysis

[Sid Vaughn]

I am looking for annotated SPSS examples of the use of PSA

[Kristian Karlson]

Hi, as far as I know, SPSS can’t do PSA (I believe you think about Propensity Score Matching for estimating “causal” effects). However, I quit SPSS at version 15, so perhaps something changed.

Stata has excellent user-written commands for matching, especially -psmatch2- and also -pscore-.

Kristian

[Sid Vaughn]

Kristian,

Many thanks – I need to make the switch to Stata

[Jeremy Miles]

There are a few different ways to do the matching for propensity estimation – and most of these SPSS is capable of, you just need to do a little more work by hand. E.g. you can use logistic regression, and then use predicted probabilities to calculate the weights.

However, I use R, as it has a boosted regression based method which seems to do a better job. The program is called twang, and it (along with R) is free.

[Sid Vaughn]

Thanks Jeremy!

[Kristian Karlson]

Nice – I did not know of twang (until know). I have to have a look at that!

Thanks,
Kristian

[Author]

Posts