Propensity Score Matching with multiple and continous treatments

[Ron Philip]

Hello,

I want to make a propensity score matching using four different treatments, which are all metric rather than binary.

Thus my first question:

I found the commands doseresponse and doseresponse2. Are there any differences in terms of timeliness or only in their ability to adjust for different settings?

Is there also a possibility to include all four treatments in a model at the same time, especially when all are not binary, but metric?

Thank you for your time and help!


And how do I interpret the results of the output of doseresponse(2)?
Can you recommend a paper which is very detailed here?

[David Benson]

Here are papers that are in Stata Journal, or are guides to use Stata:
1) A Stata package for the estimation of the dose–response function through adjustment for the generalized propensity score (2008). Link Uses gpscore and doseresponse

2) A Stata package for the application of semiparametric estimators of dose-response functions (2014). Link uses DRF

3) Estimating the dose-response function through the GLM approach.

• The working paper version (here) explains doseresponse2 and gpscore2 and how they differ from the original doseresponse and gpscore.
• The version published in Stata Journal (here) called them glmdose and glmgpscore

4) A tutorial on using ctreatreg (another Stata module for estimating Dose Response Treatment Models) here



Some papers actually implementing the models
Caliendo, M. and Kopeinig, S. (2008), SOME PRACTICAL GUIDANCE FOR THE IMPLEMENTATION OF PROPENSITY SCORE MATCHING. Journal of Economic Surveys, 22: 31-72. https://doi.org/10.1111/j.1467-6419.2007.00527.x

• Link to working paper version here

McCaffrey, D. F., Griffin, B. A., Almirall, D. , Slaughter, M. E., Ramchand, R. and Burgette, L. F. (2013), A tutorial on propensity score estimation for multiple treatments using generalized boosted models. Statist. Med., 32: 3388-3414. https://doi.org/10.1002/sim.5753 (This one is *very* detailed)

Meade, B., Steiner, B., Makarios, M., & Travis, L. (2013). Estimating a Dose–Response Relationship Between Time Served in Prison and Recidivism. Journal of Research in Crime and Delinquency, 50(4), 525–550. https://doi.org/10.1177/0022427812458928

Zanutto, E., Lu, B., & Hornik, R. (2005). Using Propensity Score Subclassification for Multiple Treatment Doses to Evaluate a National Antidrug Media Campaign. Journal of Educational and Behavioral Statistics, 30(1), 59–73. https://doi.org/10.3102/10769986030001059