Graphing interaction effects in R

Question

Is there an R equivalent of the Stata package Grinter? I would like to graph an interaction effect from a logistic regression equation in R, and I can't seem to get any of my graphs to work. If there were a command in R that could do what Grinter does in Stata, that would be extremely helpful.

Here's my equation:

svyglm(onset.new~Gini+preci.ma.30.CRU+lag.polity+
pop.growth+lpop+lag.lGDP.pc+Oil+ethfrac+lmtnest+Gini*preci.ma.30.CRU,
design=NEWD,na.action=na.omit,family=quasibinomial)

I'm interested in graphing the interaction term Gini*preci.ma.30.CRU (inequality times precipitation change) to see how the two variables combined affect conflict onset (onset.new). All the other variables are just controls.

Thank you!

There are many ways to graph interaction effects but to provide a worked example we need data, and all I see is code that is clearly dependent on a lot of prior work with the (unacknowledged) survey package. — IRTFM

Hammad Sheikh Hammad Sheikh · Accepted Answer · 2015-12-02T00:20:30

You can use the regression equation to calculate the predicted onset.new for the levels of Gini and precipitation you are intersted in (low/high Gini and low/high precipitation). You can then create a dataset with these predicted values and plot it as a bar or line graph. The dataset would look like this:

predicted.onset.new Gini precipitation .34 low low .23 high low .81 low high .28 high high

Graphing interaction effects in R

1 Answers