Hello, Lets assume I have an ordinal response variable representing the D<-c(D0,D1,D2,D3,D4) where D0 is no damage and D4 is collapse which I want to correlate with a continuous predictor variable, wind speed at the location of each building. is there a function in R which I can use to estimate the sample size of buildings with a given power if I want to perform an ordinal logistic regression? Is this sample the same if I want to use kernel smoothing? Is this possible to estimate the sample size in D0,.,D4? Collapse is rather rare. Regards, Yanna [[alternative HTML version deleted]]