How do you make percent distributions by subgroups with the package Survey? Given a continuous variable X, this command calculates a mean for X. svymean( ~ X, design=Sdescript, na.rm=T) If I use cut() to create a factor, Xcat out of X, this command gives a percent distribution. svymean( ~ Xcat, design=Sdescript, na.rm=T) This command give means of X by sex. svyby( ~ X, ~sex, design=Sdescript, svymean, na.rm=T) But this command, issued in hopes of percent distributions by sex, gives and error. svyby( ~ Xcat, ~sex, design=Sdescript, svymean, na.rm=T) The ultimate objective here is to get percent distributions and standard errors for subgroup factors like sex=(M,F) by urban/ rural=(U,R) for factors from a stratified cluster survey. /Wilbur Hadden [[alternative HTML version deleted]]