pci2s: Regression-based proximal causal inference using two-stage-least-squares

We provide estimation and inference for regression-based proximal causal inference with a variety data types of outcome variables and negative control outcomes. The supported models for the distribution of the outcome variables include linear regression model (p2sls.lm), log-linear regression model (p2sls.loglin), negative binomial regression model (p2sls.negbin), logistic regression model (p2sls.logitreg), and additive hazards regression model (p2sls.ah). For details of implementation, please refer to the documentation of the individual functions. Theoretical details can be found at Liu et al. (2024) and Li et al. (2024). An introduction of proximal causal inference can be found at Shi et al. (2020) and Tchetgen Tchetgen (2024).

Installation

Install from GitHub:

# install.packages("pak")
pak::pak("KenLi93/pci2s")

References

Liu, J., Park, C., Li, K. and Tchetgen Tchetgen, E.J., 2024. “Regression-based proximal causal inference.” American Journal of Epidemiology p.kwae370.

Li, K., Linderman, G.C., Shi, X. and Tchetgen, E.J.T., 2024. “Regression-based proximal causal inference for right-censored time-to-event data.” arXiv preprint arXiv:2409.08924.

Shi, X., Miao, W. and Tchetgen, E.T., 2020. “A selective review of negative control methods in epidemiology.” Current epidemiology reports 7, pp.190-202.

Tchetgen Tchetgen, E.J., Ying, A., Cui, Y., Shi, X. and Miao, W., 2024. “An introduction to proximal causal inference.” Statistical Science 39(3), pp.375-390.