Search code examples
rggplot2linear-regression

Add +- 2 SE in regression line R

I was wondering if it is possible to plot a regression line in R including +- 2* Standard Error. For example, with iris :

library(ggplot2)
ggplot(iris, aes(x = Petal.Length, y = Sepal.Width))+geom_smooth(method=lm, color="black")+ theme_bw()

the output is:

enter image description here

How would it possible (if possible) to expand the grey area for values twice the standard error?

Solution

  • One option to achieve your desired result would be to manually add the confidence bands via stat_smooth and making use of after_stat:

    library(ggplot2)
ggplot(iris, aes(x = Petal.Length, y = Sepal.Width)) +
  geom_smooth(method = lm, color = "black", se = FALSE) +
  stat_smooth(aes(ymin = after_stat(y - 2 * se), ymax = after_stat(y + 2 * se)), geom = "ribbon", method = lm, fill = "grey60", alpha = .4) +
  theme_bw()
#> `geom_smooth()` using formula 'y ~ x'
#> `geom_smooth()` using formula 'y ~ x'