ABSTRACT
The need for taking issues of reliability and validity seriously exists in all scientific endeavors and, of course, here also. This can be taken as self-evident and needs no special motivation. However, in the present context, there are two additional reasons for highlighting the importance of reliability and validity:
The approach presented here and the methods used to carry it out are considered by many “nonstandard.” The user of this approach may therefore encounter a certain skepticism that can be overcome by an especially careful analysis and presentation of aspects of reliability and validity.
Some of the technical methods used, foremost, cluster analysis, are not formulated in the form of a mathematical-statistical model, which incorporates errors of measurement and sampling errors. Hence, their effects are hard to evaluate. It is also documented that under certain conditions, results from cluster analysis can be distorted by such errors (see Milligan, 1980.)
