There are several easy and efficient R packages and functions to produce conditional effects plots (marginal effects) for linear regression models, such as interplot
(Solt and Hu 2021), ggeffects
(Lüdecke 2018), and JKmisc/ggintfun.R
(Karreth 2017). However, I could not find any package or function to produce marginal effects graphs for spatial models. In particular, I am conducting a SAR model using the command lagsarlm
from the package spdep
.
Consider, for example, the SAR model specification below:
Y = B1X1 + B2X2 + B3X3 + B4X1X2 + B5X3X2 + pWy
In this example, I would like to produce two marginal effects plots according to the interaction terms:
- The conditional effect of X1 on Y when X2 is increasing.
- The conditional effect of X3 on Y when X2 is increasing.
Is there any way to produce these graphs in R from the results generated using the command lagsarlm
(package spdep
)?