Original Data
Replication Data
Software
The R package provides functions for estimating and making inference of treatment effects from time-series cross-sectional data with the Bayesian dynamic multilevel latent factor models.
Cite: Xun Pang, Licheng Liu and Yiqing Xu, “A Bayesian Alternative to Synthetic Control for Comparative Case Studies,” Political Analysis, Vol.30, No.2, 2022.
The R package provides functions for identifying and estimating time-varying network interdependence by controlling for unobserved homophy and common exposures with observational longitudinal network data.
Cite: Licheng Liu and Xun Pang , “A Bayesian Multifactor Spatio-Temporal Model for Estimating Time-Varying Network Interdependence”
The R package contains functions to estimate the GLMM-AR(p) model for analyzing discrete time-series cross-sectional data via Markov Chain Monte Carlo simulation. The simulation is done only with the R language. The model returns draws of the parameter posteriors selected by the user in a list format. Each parameter chain is returned as a matrix. The user is responsible to summarize the mcmc output by using the coda package. GLMMarp also contains several useful utility functions, including an independent function for computing the Bayes factor with GLMM-AR(p) output, a function to recover the random coefficients at the individual level, and a function to do prediction by using the posterior distributions. The package also contains a library of supporting functions for the MCMC simulation and Bayes factor estimation. In this version, no tools for visualization are provided. To use the functions in this package, the user needs to load the following two packages by her panel and bayesSurv, because the two packages have no namespace.
Cite: Xun Pang, “Modeling Heterogeneity and Serial Correlation in Binary TSCS Data: A Bayesian Multilevel Model with AR(p) Errors,” Political Analysis, Vol.18, No.4, pp.470-498, 2010.
SpikeSlab
This package provides function to use spike and slab prior distributions for simultaneous Bayesian hypothesis testing, model selection, and prediction based on Bayesian model averaging, of nonlinear outcomes.
Cite: Xun Pang and Jeff Gill , “Spike and slab prior distributions for simultaneous Bayesian hypothesis testing, model selection, and prediction, of nonlinear outcomes”, working paper.