My understanding of a qqplot was it was the sorted values for a variable on the y against the theoretical values from a normal (or whatever distribution) distribution on the x. We can get the y values from our sample and the x values from looking up the p value in a given distribution and getting the corresponding nromal value. Now to get the p value I thought we divide the 1 through length of the sample by n + 1. I have never actually tried to do this and used qqplot or the same geom_qq from ggplot. But today I tried to make the qqplot by hand and I don't get the same scale for the theoretical values and am unsure why. The points look the same but the scale is different. Why?
x <- c(3.89, 4.75, 4.75, 5.2, 5.78, 5.8, 6.33, 7.21, 7.9) par(mfrow = c(2, 1)) plot(qnorm((1:length(x))/(1 + length(x))), sort(x)) qqnorm(x)