connorjmccabe/modglm: Computing interaction effects for nonlinear probability and counts

This package computes interaction effects and their standard errors for logistic, Poisson, and negative binomial models. It also provides graphical utilities for viewing marginal effects corresponding with these interactions. Effects are computed using cross-partial derivatives and discrete difference approaches described by Ai & Norton (2003) and by the accompanying manuscript to this package (McCabe et al., 2021). This package allows users to evaluate interaction functions conditional on user-specified values of predictors. It also computes interactions for each observation in the data row-wise, as well as the average interaction effect observed in the data.

install.packages("remotes")
remotes::install_github("connorjmccabe/modglm")