I have a problem. my dependent variable is a relative frequency. i actually wanted to estimate a betaregression, but after checking my data i saw that 42% are either 0 or 1. i could weight these values to get them into the (0,1) interval but i think 42% is to much to estimate a betaregression. at least i cant find anything about a critical value (42% to high or not). so now i am looking for another method to estimate my regression. the values between the extrems are almost normal distributed.