Changing the interval ribbon in interact

I am creating some interaction plots and want to add a 95% confidence interval for my interaction terms. However, using the interact_plot() function from the interactions package adds a geom_ribbon() for the confidence interval, which I find difficult to read. I would like to simply have dashed lines to represent the upper and lower bounds of the confidence interval.

It seems to me this involves updating or changing the layers from the list that is stored for my plot, but I am not quite sure how to do this.

Here is an example:

library(lme4)
library(tidyverse)
library(interactions)
data(VerbAgg)
mv <- glmer(r2 ~ Anger * mode   (1 | item), data = VerbAgg,
             family = binomial,
             control = glmerControl("bobyqa"))
p = interact_plot(mv, pred = Anger, modx = mode, vary.lty = FALSE, interval = TRUE, types = c("solid", "solid"))
p

Instead of having the ribbon, I want to just have dashed lines representing the lower and upper bounds of the confidence interval as said above.

Looking at the summary of the plot object yields:

summary(p)

data: r2, item, mode, Anger, ymax, ymin, modx_group [200x7]
mapping:  x = ~Anger, y = ~r2, colour = ~mode, group = ~mode, linetype = NULL
scales:   colour, fill 
faceting: <ggproto object: Class FacetNull, Facet, gg>
    compute_layout: function
    draw_back: function
    draw_front: function
    draw_labels: function
    draw_panels: function
    finish_data: function
    init_scales: function
    map_data: function
    params: list
    setup_data: function
    setup_params: function
    shrink: TRUE
    train_scales: function
    vars: function
    super:  <ggproto object: Class FacetNull, Facet, gg>
-----------------------------------
geom_path: lineend = butt, linejoin = round, linemitre = 10, arrow = NULL, na.rm = FALSE, size = 1
stat_identity: na.rm = FALSE
position_identity 

mapping: x = ~Anger, ymin = ~ymin, ymax = ~ymax, fill = ~mode, group = ~mode, colour = ~mode, linetype = NA 
geom_ribbon: na.rm = FALSE, orientation = NA, outline.type = both, alpha = 0.2, flipped_aes = FALSE
stat_identity: na.rm = FALSE
position_identity

You'll notice that the mapping layer mapping: x = ~Anger, ymin = ~ymin, ymax = ~ymax, fill = ~mode, group = ~mode, colour = ~mode, linetype = NA has linetype = NA. My thought is that this needs to change to accomplish my goal, but I do not know how I can change the layer.

Any insight would be greatly appreciated! It does not seem that interact_plot() has an argument that will allow me to do this directly, so I think it requires some wrangling.

CodePudding user response：

One option to achieve your desired result would be to

Add lines for the confidence intervals using two geom_line
Get rid of the geom_ribbon by setting the fill colors to transparent
Add the colors back where I used the default colors used by interact_plot (Note: interact_plot uses one scale for both fill and color. Hence, when adding scale_fill_manual we also lose the colors).

library(lme4)
library(ggplot2)
library(interactions)

p  
  geom_line(aes(Anger, ymin), linetype = "dashed")  
  geom_line(aes(Anger, ymax), linetype = "dashed")  
  scale_fill_manual(values = c("transparent", "transparent"), aesthetics = "fill")  
  scale_color_manual(values = jtools::get_colors("CUD Bright", 2), aesthetics = "color")
#> Scale for 'colour' is already present. Adding another scale for 'colour',
#> which will replace the existing scale.