Repeating the scalogram analyses using two alternative
measures of involvement in problem drinking produced little
difference either in the relative position of problem drinking
along the dimension or in the magnitude of the Guttman
properties of the resulting scales.