Discussion

It is best to treat the proximities as ordinal variables, because there is great improvement in the stress measures. As a next step, you may want to “untie” the ordinal variables—that is, allow equivalent values of the original variables to obtain different transformed values. For example, in the first source, the proximities between Aunt and Son, and Aunt and Grandson, are 85. The “tied” approach to ordinal variables forces the transformed values of these proximities to be equivalent, but there is no particular reason for you to assume that they should be. In this case, allowing the proximities to become untied frees you from an unnecessary restriction.