The effectclass
package helps interpreting effects and visualising uncertainty.
It classifies the effects by comparing a coverage interval with a reference, lower and upper threshold. The result is a 10 scale classification of the effect. You can reduced it to a 4 scale classification. effectclass
provides stat_effect()
and scale_effect()
to visualise the effects as points with different shapes.
The Bank of England visualises uncertainty by using a fan plot1. Instead of displaying a single coverage interval, they recommend to display a bunch of coverage intervals with different levels of transparency.
You can install the released version of effectclass
from GitHub with:
# installation requires the "remotes" package
# install.package("remotes")
remotes::install_github("inbo/effectclass")
Classifying effect for usage in a table
library(effectclass)
z <- data.frame(
effect = c("unknown\neffect", "potential\npositive\neffect",
"potential\nnegative\neffect", "no effect", "positive\neffect",
"negative\neffect", "moderate\npositive\neffect",
"moderate\nnegative\neffect", "strong\npositive\neffect",
"strong\nnegative\neffect"),
estimate = c( 0, 0, 0, 0, 1, -1, 0.5, -0.5, 1.5, -1.5),
lcl = c(-2, -0.9, -2, -0.9, 0.1, -2, 0.1, -0.9, 1.1, -2),
ucl = c( 2, 2, 0.9, 0.9, 2, -0.1, 0.9, -0.1, 2, -1.1)
)
classification(z$lcl, z$ucl, threshold = c(-1, 1), reference = 0)
Adding a classification to a plot
library(ggplot2)
ggplot(z, aes(x = effect, y = estimate, ymin = lcl, ymax = ucl)) +
stat_effect(threshold = c(-1, 1), reference = 0, size = 3)
Creating a fan plot
Britton, E.; Fisher, P. & J. Whitley (1998). The Inflation Report Projections: Understanding the Fan Chart. Bank of England Quarterly Bulletin. Retrieved 2019-05-22.↩︎