Search code examples
rggplot2labelinteraction-plot

Customize label for an Interaction Plot

Suppose I have the following data which summarizes the values of a dependent variable (dv) at low and high vales of a predictor (x) and a moderator (z)

dfx <- data.frame(x = rep(c(-0.909, 0.909),2),
                  z = c(-0.676, -0.676, 0.676, 0.676),
                  dv = c(448.8, 464.4, 462.8, 472.1))

I would like to draw the dv as a function of the predictor and the moderator by customizing both the type of lines and the markers

library(ggplot2)
dfx$z <- as.factor(dfx$z) # turn the moderator as a factor

ggplot(dfx, aes(x=x, y = dv, colour=z, shape=z))+
  geom_line()+
  geom_point()+
  scale_y_continuous("dv", limits = c(440,480))+
  theme_classic()+
  scale_colour_manual  ("A title", values=c("black", "black"), labels=c("-1SD", "+1SD"))+
  scale_shape_manual   ("A title", values=c(2,15),labels=c("-1SD","+1SD"))

enter image description here

However, I would like to customize the linetype as well.

When I add:

ggplot(dfx, aes(x=x, y = dv, colour=z, shape=z, linetype=z)

the following plot emerges. Yet I cannot either hide the second label or customize it through scale_linetype_manual

enter image description here

Is there a way to fix this? I searched the community but could not find any method to overcome this problem. Also, when I suppress the second label, the legend key does not show the line type and mark. Any hint will be very much appreciated.

Solution

  • From your description, I'm guessing you want something like this:

    library(ggplot2)

ggplot(dfx, aes(x=x, y = dv, shape=z, linetype=z)) +
  geom_line()+
  geom_point()+
  scale_y_continuous("dv", limits = c(440,480))+
  theme_classic()+
  scale_linetype_manual("A title",
                      values=c(1, 3),
                      labels=c("-1SD", "+1SD")) +
  scale_shape_manual("A title",
                     values=c(2,15),
                     labels=c("-1SD","+1SD"))

    result