Illustration from thinkCausal software used to teach the foundations of causal inference through interactive tools.
The overarching goal of the thinkCausal project was to create a user-friendly software package to support researchers in understanding the methods embedded in the existing but prohibitively technical bartCause causal inference software package. thinkCausal is highly scaffolded, not only to enable the R “engine” to be seamlessly utilized, but to bolster the user’s knowledge about the methods and underlying assumptions. This will allow researchers to make better decisions about implementation as well as more knowledgeably discuss their analyses and results.
The current software, thinkCausal, scaffolds the researcher through the data analytic process including: uploading data, modeling choices (which variables to include, what estimands to target), exploratory data analysis (plots of raw data, visualization of balance and overlap), modeling fitting, post-hoc analysis (examination of moderation), and interpretation of results. In addition, interactive educational components are interspersed throughout the tool to provide researchers with the opportunity to gain a deeper understanding of the underlying causal inference foundations (potential outcomes, causal estimands, assumptions, researcher degrees of freedom, multiple comparisons). To aid in reproducibility, the software provides automatic documentation of the analytic choices made. In this section of the report we discuss the development and refining of the analysis portion of thinkCausal.
Project Team
Publications
Hill, J., Perrett, G., Hancock, S., Bergner, Y., and L. Win (2024) “Causal Language and Statistics Instruction: Evidence from a randomized experiment” Statistics Education Research Journal, 23(1) [ERIC Accession number: ED660558]
DOI: https://doi.org/10.52041/serj.v23i1.673
Hill, J., Perrett, G. and V. Dorie (2023) “Machine Learning for Causal Inference,” in J.R. Zubizarreta, E.A Stuart, D.S. Small, and P.R Rosenbaum (Eds.) Handbook of Multivariate Matching and Weighting for Causal Inference (pp. 416-443). Chapman & Hall/CRC: Boca Raton, FL [ERIC Accession number: ED660568]
DOI: https://doi.org/10.1201/9781003102670
Dorie, V., Perrett, G., Hill, J.H., and B. Goodrich (2022) "Stan and BART for Causal Inference: Estimating Heterogeneous Treatment Effects Using the Power of Stan and the Flexibility of Machine Learning" Entropy, 24, no. 12: 1782. [ERIC Accession number: ED626397]