plot_model() keeps returning NULL in theme features

Question

I am making use of the plot_model() predictive function for a multiple linear regression and am not able to modify aesthetics. I have read both that it is a part of the sjplot package as well as a wrapper for ggplot. While I have gotten pretty proficient in modifying/debuggin ggplot aesthetics, I am finding this not working here.

It plots just fine in my code as is. I've put placeholder names for the variables.

library(sjPlot)
library(performance)
library(ggplot2)
library(sjmisc)

setwd("C:/Users/directory/")

df=read.csv('example.csv')
df$id=factor(df$id)
df$course=factor(df$course)

lm_mod=lm(data=df, score~var1+var2+var3)
summary(lm_mod)

prediction = plot_model(lm_mod, 
          type="pred", 
          show.data=TRUE,
          dot.size = 1,
          )

prediction

As soon as I add themes, it returns NULL.

prediction + set_theme(base = theme_minimal())

I want to adjust the axis scale / limits to match my other ggplot made figures. The text size of the axis I would like to fix also. Image below, with the variable names removed

If I try anything from sjplot or ggplot to modify the aesthetics/themes, they all universally return "NULL." Some examples below:

> prediction + font_size(axis_title.x=28)
NULL
> prediction + font_size(28)
NULL
> prediction + set_theme(base = theme_minimal())
NULL
> prediction + theme_sjplot(base_size = 24, base_family="")
NULL
> prediction + theme(text=element_text(size=22),  
+                    axis.title = element_text(size=28, 
+                                              face="bold" ),  
+                    axis.text = element_text(size=24), 
+                    axis.line = element_line(colour = "black", 
+                                             size = 1, 
+                                             linetype = "solid"),
+                    plot.title = element_text(size=28,
+                                              face="bold",
+                                              vjust=2,
+                                              hjust=0.5))
NULL

I have read related posts. Please help! I am very stuck trying to fix this.

Answer 1

If you look at the help page ?plot_model, under Value it states:

Depending on the plot-type, plot_model() returns a ggplot-object or a list of such objects.

Here's a simple example:

library(sjPlot)
library(ggplot2)

lm1 <- lm(mpg ~ wt, data = mtcars)

p1 <- plot_model(lm1)
p2 <- plot_model(lm1, type = "pred")

class(p1)
[1] "gg"     "ggplot"

class(p2)
[1] "list"

So p1 is a ggplot object and you could do this:

p1 + theme_bw()

But p2 is a named list. It gets its name from the independent variable in the model. You'll get NULL if you try to add themes to a list.

names(p2)
[1] "wt"

p2 + theme_bw()
NULL

So to access the ggplot part of the list, you can do this:

p2$wt + theme_bw()

or this:

p2[[1]] + theme_bw()

With more than one independent variable, you'll get a list element for each one. In your case:

prediction$var1
prediction$var2
prediction$var3