aphalo
commented
6 months ago @aleungplants
The expected input is: y ~ x + I(x^2)
I recently added a warning, to the development version, highlighting that when writing the model formula the coefficients must be in order of increasing powers, or alternatively the model formula should be written using poly() and passing raw = TRUE. In the development version, based on an earlier request I have also added the possibility of having an equation label with the terms in order of decreasing powers by passing "decreasing = TRUE". However, even in this case the order in the model formula must remain increasing. I hope this solves the problem.
Ideally, the stats would check the syntax of the model formulas and reject those not matching the expectations with an error message. Writing code for this check I expect not to be simple, so it will take some time for me to write it, unless someone steps in to help with a pull request, or I find some example. Meanwhile, I hope the warning section I added in the help pages, will help.
Thanks for the report! I will keep this issue open but on hold, and edit the title.
This is the result of running an lm with a quadratic formula.
coef(lm(formula = mpg ~ I(wt^2) + wt, data = mtcars)) (Intercept) I(wt^2) wt 49.930811 1.171087 -13.380337
However, running this code returns a figure that says y = 49.9+1.17x+13.4x^2, when it should say 49.9+13.4x+1.17x^2.
ggplot(mtcars, aes(x = wt, y = mpg)) + geom_point() + ggpmisc::stat_poly_eq( formula = y ~ I(x^2)+I(x), ggpmisc::use_label(labels = c("eq")), label.x = "right")