This app runs the u-shape test introduced by Simonsohn (2017 .pdf).
It estimates an interrupted regression—two separate slopes—for the predictor hypothesized to have a u-shaped effect. The breakpoint is set using the “Robin Hood” algorithm, seeking higher power to detect a u-shape if it is present. If the resulting two slopes have opposite sign and are individually statistically significant, the test rejects the null that there is no u-shaped (nor inverted u-shaped) effect.
To proceed:
This free online app allows datasets up to 15 mb.
No file handy? Download this example datafile and upload it.
Uploaded data is deleted within 72 hours. Files are saved unencrypted in a public folder but given a temporary name, so they are hard to find but not impossible to locate. For confidential data, download the R code and run locally instead of uploading here.