February 2020 Newsletter
|
|
We hope this newsletter finds you well. We are pleased to share updates from the Causal Tools Project. We have been busy over the past 18 months, developing new tools and methods to support causal inferences from observational studies. This newsletter will catch you up on recent activities with an overview of what’s available online now, what’s new, and what’s coming up!
|
|
|
|
Latest Tool
|
We have successfully ported our TWANG package command for binary treatments to a Shiny app. This is a user-friendly, menu-driven application (app) that can be used to estimate propensity score weights for binary treatments. It also estimates the needed treatment effect estimates using those propensity score weights. The app includes a tutorial, which demonstrates use of the TWANG Shiny app through an illustrative example and toy dataset.
The source code is also available for more advance R/Shiny users who want to personalize the app. If you do, we would love to hear back from you about what changes you made to personalize the app.
|
|
|
Forthcoming Tools
We have additional tools developed for estimating causal effects that are currently being beta-tested. We’re looking for additional beta-testers to help with these tools. If you are interested in beta testing any of our forthcoming tools, please email us at twang-project@rand.org. Beta testers who can return feedback within two weeks will receive a $50 gift card.
R Tools
- Continuous exposures/treatment tool using GBM (pscont) – This tool estimates generalized propensity score weights for continuous exposures/treatments using generalized boosted models (GBM).
- Omitted variable sensitivity analysis tool (OVtool) – This tool assesses the sensitivity of treatment effects and statistical significance to an unobserved confounder.
- Causal mediation tool (wgtmed) – This tool implements causal mediation analyses.
- Selection bias decomposition tool (SBdecomp) – This tool quantifies the bias due to each observed confounders in causal treatment effect estimates for binary treatments.
- Twang 2.0 – Our updated TWANG package expands the original package to include xgboost for estimation of propensity score weights with the goal of improving the performance of the TWANG package for big data.
- Entropy balancing for binary & continuous trts (entbal) – This tool estimates causal treatment effects for binary and continuous treatments using entropy balancing (EB).
- Estimation of optimally balanced Gaussian process propensity scores (gpbalancer) – This tool estimates weights for use in estimating causal treatment effects for binary and multivariate treatment settings.
|
Shiny Tools
- TWANG MNPS Shiny app – This user-friendly, menu-driven app estimates propensity score weights for more than two treatment groups using GBM.
- TWANG IPTW Shiny app – This user-friendly, menu-driven app estimates inverse probability of treatment weights for time-varying treatments using GBM.
|
Stata Tools
- IPTW Macro – This macro estimates inverse probability of treatment weights for time-varying treatments using GBM.
|
|
|
Currently Available Tools
All software tools are available to download here: http://www.rand.org/statistics/twang/downloads.html
SAS Tools
- TWANG Macro Package – This package contains commands for estimating propensity score weights for two groups (using the %ps command) and for more than two groups (using the %mnps command).
|
Stata Tools
- Support Files for PS and MNPS - This set of files supports estimating propensity score weights for two groups (using the ps command) and more than two groups (using the ps and mnps command).
|
R Tools
The twang R library on CRAN contains:
- ps command — to estimate propensity score weights for two groups
- mnps command — to estimate propensity score weights for more than two groups
- iptw command — to estimate inverse probability of treatment weights for time-varying treatments.
|
|
|
Tutorials
|
All tutorials are available to download here: http://www.rand.org/statistics/twang/tutorials.html
- TWANG Short Course/Educational Videos: This series of three training videos provides researchers and analysts with an overview of causal inference and the role of propensity scores in analysis. The videos provide step-by-step procedures for implementing propensity score analyses involving two or more treatment groups using the TWANG (Toolkit for Weighting and Analysis of Nonequivalent Groups) data analysis package.
- R Tutorials for using ps, mnps, and iptw commands
- SAS Tutorials for %ps and %mnps commands
- Stata Tutorials for ps and mnps commands
|
|
|
|
Upcoming Workshops
|
A list of workshops can be found here: http://www.rand.org/statistics/twang/workshops.html
Members of the TWANG project team will be conducting a series of workshops in 2020 to help those without any experience in propensity scores, as well as more advanced users, gain hands-on experience to estimate propensity score weights using boosted models in R, SAS, and Stata.
American Statistical Association (ASA) Traveling Short Course Series
Guidelines for Using State-of-the-Art Methods to Estimate Propensity Score and Inverse Probability of Treatment Weights When Drawing Causal Inferences
Presenters: Beth Ann Griffin, Daniel McCaffrey, Lane Burgette, Matt Cefalu, Donna Coffman, Brian Vegetabile.
Locations: TBA
Dates: 2020
Causal Inference Using the R TWANG Package for Mediation and Continuous Exposures
Presenters: Donna L. Coffman
Location: Eastern North American Region conference, Nashville TN
Date: March 23, 2020
Estimating Propensity Scores for Binary, Multinomial, and Continuous Exposures Using TWANG
Presenters: Donna L. Coffman & Megan S. Schuler
Location: Society of Epidemiological Research Annual Meeting, Boston, MA
Date: June 16-19, 2020
Can’t Shake That Feeling You Forgot Something? Assessing Sensitivity of Findings to Omitted Variable Bias
Presenters: Beth Ann Griffin
Location: College on the Problems of Drug Dependence, Hollywood, FL
Date: June 20-24, 2020
Estimating Causal Effects of Continuous Exposure: Comparing the Relative Performance of Commonly Used Methods
Presenters: Donna Coffman, Brian Vegetabile, Beth Ann Griffin
Location: Invited poster session, Joint Statistical Meetings, Philadelphia, PA
Date: August 1-6, 2020
Can't Shake That Feeling That You're Missing Something?
Presenters: Lane Burgette, Beth Ann Griffin
Location: Invited session, Joint Statistical Meetings, Philadelphia, PA
Date: August 3, 2020 at 10:30 a.m.
|
|
|
|
|
Related Publications
|
A full list of related publications can be found here: http://www.rand.org/statistics/twang/publications.html
- Antonelli J, Cefalu M, and Agniel D (2018). Doubly robust matching estimators for high dimensional confounding adjustment, Biometrics, 74: 1171-1179.
- Coffman, D. L., Zhou, J., Cai, X., & Graham, J. W. (2018). Addressing missing data in confounders when estimating propensity scores for continuous exposures. Health Services and Outcomes Research Methodology, 18(4), 265-286.
- Grant, S., Agniel, D., Almirall, D., Burkhart, Q., Hunter, S. B., McCaffrey, D. F., Pedersen, E. R., Ramchand, R. & Griffin, B. A. 2017. Developing adaptive interventions for adolescent substance use treatment settings: protocol of an observational, mixed-methods project. Addiction Science & Clinical Practice, 12(1), pp 35.
- Griffin, B. A., McCaffrey, D. F., Almirall, D., Burgette, L. F. & Setodji, C. M. 2017. Chasing balance and other recommendations for improving nonparametric propensity score models. Journal of Causal Inference, 5(2).
- Miles, J., Parast, L., Babey, S., Griffin, B.A. & Saunders, J. 2017 The truth about dogs and cats? A propensity matched population based study of the health benefits of dogs and cats for children. Anthrozoos.
- Parast, L., McCaffrey, D. F., Burgette, L. F., de la Guardia, F. H., Golinelli, D., Miles, J. N. & Griffin, B. A. 2017. Optimizing variance-bias trade-off in the TWANG package for estimation of propensity scores. Health Services and Outcomes Research Methodology, 17(3-4), pp 175-197.
- Robbins, M., Griffin, B.A., Shih, R.A. & Slaughter, M.E., Robust estimation of the effect of neighborhood socioeconomic status on cognitive function. In press. Statistics in Medicine.
- Schnitzer M and Cefalu M (2018). Collaborative targeted learning using regression shrinkage, Statistics in Medicine, 37: 530–543.
- Setodji, C. M., McCaffrey, D. F., Burgette, L. F., Almirall, D. & Griffin, B. A. 2017. The Right Tool for the Job: Choosing Between Covariate-balancing and Generalized Boosted Model Propensity Scores. Epidemiology (Cambridge, Mass.), 28(6), pp 802-811.
- Shanahan, M., Hughes, C. E., McSweeney, T. & Griffin, B. A. 2017. Alternate policing strategies: Cost-effectiveness of cautioning for cannabis offences. International Journal of Drug Policy, 41(140-147).
- Vegetabile B.G., Gillen D.L., & Stern H.S.. 2020. Optimally balanced Gaussian process propensity scores for estimating treatment effects. J. R. Stat. Soc. A. 183: 355-377.
|
|
|
|
|
Funding
|
Funding for these efforts have been supported by two grants from the National Institute of Drug Abuse (NIDA): R01DA034065 and R01DA045049 (PIs: Griffin & McCaffrey).
|
|
|
|
|
Feedback
|
|
RAND is a nonprofit institution that helps improve policy and decisionmaking through research and analysis.
|
|
|
Privacy statement
|
 |
RAND. RAND® is a registered trademark.
|
|
|
|
|
|
|