Hi, I need to perform a robust regression with several dummy independent variables (i.e., binary), which are paired (clustered). The outcome (dependent variable) is continuous. I have tried using the robcov function with the next syntaxis: out1 <- ols(y ~ x1 + x2 + x3 , x=T) robustclust <- robcov(out1, pair) (where, as stated 'y' is continuous, 'x1', 'x2' and 'x3' are binary and 'pair' contains cluster numbers). Is this the best way to do it? How to know it? Should 'bootcov' be used instead of 'robcov'? Regards, [[alternative HTML version deleted]]