Contact: Jason Bryer ([email protected])
Early draft of the Applied Propensity Score Analysis with R book.
Run the PSA Shiny App:
devtools::install_github('jbryer/psa')
psa::psa_shiny()
The use of propensity score methods (Rosenbaum & Rubin, 1983) for estimating causal effects in observational studies or certain kinds of quasi-experiments has been increasing in the social sciences (Thoemmes & Kim, 2011) and in medical research (Austin, 2008) in the last decade. Propensity score analysis (PSA) attempts to adjust selection bias that occurs due to the lack of randomization. Analysis is typically conducted in two phases where in phase I, the probability of placement in the treatment is estimated to identify matched pairs or clusters so that in phase II, comparisons on the dependent variable can be made between matched pairs or within clusters. R (R Core Team, 2012) is ideal for conducting PSA given its wide availability of the most current statistical methods vis-à-vis add-on packages as well as its superior graphics capabilities.
This workshop will provide participants with a theoretical overview of propensity score methods as well as illustrations and discussion of PSA applications. Methods used in phase I of PSA (i.e. models or methods for estimating propensity scores) include logistic regression, classification trees, and matching. Discussions on appropriate comparisons and estimations of effect size and confidence intervals in phase II will also be covered. The use of graphics for diagnosing covariate balance as well as summarizing overall results will be emphasized.
- Theretical overview of propensity score methods
- Phase I of PSA - Adjusting for selection bias by modeling treatment placement
- Stratification using logistic regression and classification trees; also random forests
- Propensity score matching
- Checking covariate balance
- Phase II of PSA - Estimating effects re: response variables
- Dependent sample tests and confidence intervals.
- Visualizing results
- Advanced Topics
- Sensitivity analysis
- PSA with Missing Data
- Bootstrapping for PSA (using the
PSAboot
package) - Analysis of non-binary treatments (using the
TriMatch
package) - Analysis of multilevel data (using the
multilevelPSA
package)
- Workshop at University at Albany, Division of Educational Psychology & Methodology -- April 30 and May 7, 2014 Download Slides
- NEAIR Conference Talk -- November 11, 2013 Download Slides
- Pre-conference workshop for the 2013 useR! conference -- July 9, 2013 [Download Slides](Slides/useR 2013/Slides.pdf?raw=true)
Before getting started, you should download and install R and RStudio.
This workshop assumes minimal knowledge of R and will not cover the basics. If you are new to R, there are many great resources for learning.
- Quick-R - A great general purpose reference website covering basic and advanced R topics.
- R in Action - This book by Kabacoff is originated from the Quick-R website and is a fantastic book for learning R.
- ggplot2 - Most of the visualizations in this workshop were created with the
ggplot2
package which is an implementation of Wilkinson's (2005) Grammar of Graphics. - The Art of R Programming - This is a nice companion book to Kabacoff's R in Action and may be preferred if you come to R with a programming background.
- OpenIntro Statistics - This is a free and opensource introductory statistics textbook that uses R for all its examples and exercises.
- The R Journal - Journal that publishes articles related to R.
- Journal of Statistical Software - An American Statistical Association journal focused on statistical software that regularly publishes articles about R packages.
- R FOR SAS AND SPSS USERS - If you already know SAS or SPSS, Muenchen's book is very well regarded for learning R.
- R Bloggers - This is an aggregate of many blogs about R and an excellent resource for learning about new developments around R.
- Twitter - The
#Rstats
hashtag on Twitter is a great way to see what people are saying about R. - StackOverflow - This website is a great place to go to ask questions and get help on R related topics.
The following R scripts will outline how to conduct propensity score analysis.
- Setup.R - Install R packages. This script generally needs to be run once per R installation.
- IntroPSA.R - Conducts propensity score analysis and matching, summarizes results, and evaluates balance using the National Supported Work Demonstration and Current Population Survey (aka lalonde data).
- IntroPSA-Tutoring.R - Conducts propensity score analysis and matching, summarizes results, and evaluates balance using data from a study examining student use of tutoring services in an online introductory writing class (from the
TriMatch
package). - Sensitivity.R - Conduct a sensitivity analysis.
- Missingness.R - How to evaluate whether data is missing at random.
- BootstrappingPSA.R - Boostrapping PSA.
- NonBinaryPSA.R - Analysis of three groups (two treatments and one control)
- MultilevelPSA.R - Multilevel propensity score analysis.
There are a number of R packages available for conducting propensity score analysis. These are the packages this workshop will make use of:
MatchIt
(Ho, Imai, King, & Stuart, 2011) Nonparametric Preprocessing for Parametric Causal InferenceMatching
(Sekhon, 2011) Multivariate and Propensity Score Matching Software for Causal InferencemultilevelPSA
(Bryer & Pruzek, 2011) Multilevel Propensity Score Analysisparty
(Hothorn, Hornik, & Zeileis, 2006) A Laboratory for Recursive PartytioningPSAboot
(Bryer, 2013) Bootstrapping for Propensity Score AnalysisPSAgraphics
(Helmreich & Pruzek, 2009) An R Package to Support Propensity Score Analysisrbounds
(Keele, 2010) An Overview of rebounds: An R Package for Rosenbaum bounds sensitivity analysis with matched data.rpart
(Therneau, Atkinson, & Ripley, 2012) Recursive PartitioningTriMatch
(Bryer, 2013) Propensity Score Matching for Non-Binary Treatments
Rosenbaum, P.R., & Rubin, D.B. (1983). The central role of the propensity score in observational studies for causal effects. Biometrika, 70(1), 41-55.
Rosenbaum, P.R. (2010). Design of Observational Studies. New York: Springer.
Austin, P. C. (2011). Comparing paired vs non-paired statistical methods of analyses when making inferences about absolute risk reductions in propensity-score matched samples. Statistics in Medicine, 30.
Bryer, J. (2011). multilevelPSA: Multilevel propensity score analysis [Computer software manual]. Retrieved from http://github.com/jbryer/multilevelPSA
Bryer, J., & Pruzek, R.M. (2011). An international comparison of private and public schools using multilevel propensity score methods and graphics (Abstract). Multivariate Behavioral Research, 46(6), 1010-1011.
Helmreich, J. E., & Pruzek, R. M. (2009). PSAgraphics: An R package to support propensity score analysis. Journal of Statistical Software, 29(6). Available from http://www.jstatsoft.org/v29/i06/paper
Ho, D.E., Imai, K., King, G., and Stuart, E.A (2011). MatchIt: Nonparametric Preprocessing for Parametric Causal Inference. Journal of Statistical Software 42(8).
Hothorn, T., Hornik, K., & Zeileis, A. (2006). Unbiased Recursive Partitioning: A Conditional Inference Framework. Journal of Computational and Graphical Statistics, 15(3), 651--674.
R Core Team (2012). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0.
Rosenbaum, P.R. (2005). Sensitivity analysis in observational studies. In B.S. Everitt & D.C. Howell Encyclopedia of Statistics in Behavioral Science, pp. 1809-1814. Chichester: John Wiley & Sons.
Rosenbaum, P.R. (2012). Testing one hypothesis twice in observational studies. Biometrika.
Sekhon, J.S. (2011). Multivariate and Propensity Score Matching Software with Automated Balance Optimization: The Matching Package for R. Journal of Statistical Software, 42(7), 1-52.
Shadish, W.R., Clark, M.H., & Steiner, P.M. (2008). Can nonrandomized experiments yield accurate answers? A randomized experiment comparing random and nonrandom assignments. Journal of the American Statistical Association, 103(484). 1334-1356.
Stuart, E. A. (2010). Matching methods for causal inference: A review and a look forward. Statistical Science, 25, 1-21.
Stuart, E.A., & Rubin, D.B. (2007). Best practices in quasi-experimental designs: Matching methods for causal inference. Chapter 11 (pp. 155-176) in J. Osborne (Ed.). Best Practices in Quantitative Social Science. Thousand Oaks, CA: Sage Publications.
Therneau, T., Atkinson, B., & Ripley, B. (2012). rpart: Recursive Partitioning. R package version 4.0-1.
Thoemmes, F. J., & Kim, E. S. (2011). A systematic review of propensity score methods in the social sciences. Multivariate Behavioral Research, 46, 90-118.