Removing selected regression lines from plots made with sjPlot

Question

I am trying to plot an interaction plot using the sjPlot package with the following code:

model <-lm(score ~ var1 + var2 + var3 + var4 + var1*var2,data=mydata)
plot1 <- plot_model(model,type="int",ci.lvl=NA, title="", legend.title="",
  axis.title=c("var2","score"))

The data I'm using is similar to this dataframe:

n <- 50
mydata<- data.frame(
  ID = 1:n,
  var1 = sample(c("rank1", "rank2", "rank3"), n, replace = TRUE),
  var2 = sample(0:2, n, replace = TRUE),
  var3 = runif(n, min = 1, max = 60),
  var4 = sample(c(0, 1), n, replace = TRUE)
)

Here are the first 10 rows:

   ID  var1 var2      var3 var4
1   1 rank2    0 11.252379    1
2   2 rank1    1 29.954015    0
3   3 rank2    2 42.397683    1
4   4 rank1    1 12.917567    1
5   5 rank3    1  8.095043    1
6   6 rank1    0 25.986603    0
7   7 rank3    2 10.349064    0
8   8 rank1    1 35.423613    1
9   9 rank2    0 24.232398    1
10 10 rank1    2 12.996389    1

var1 has 3 categories (rank1, rank2, rank3), so the interaction plot has 3 lines (score on the y-axis and var2 on the x-axis). I want to remove the line representing rank2 only, so the remaining plot will only have 2 lines.

TYIA

Answer 1

I used type="pred" rather than type="int" because it allows better control. You can then specify which is the term for the x axis, and which term (at which levels) to use as the moderator. See below:

plot_model(model,
                    type="pred",
                    ci.lvl=NA, 
                    title="", 
                    legend.title="",
                    axis.title=c("var2","score"),
                    terms=c("var2","var1 [rank2,rank3]")
                    )