Wright B.D., Masters G.N. Rating Scale Analysis

Chicago: MESA Press, 1982. — 223 p.This is a book about constructing variables and making measures. We begin by outlining the requirements that a number must meet before it qualifies as a “ measure” of something. Our requirements for measurement may seem stringent, but they are not new. They characterize all the instances of useful scientific measurement that we have been able to find. We suspect that their neglect by social scientists has stemmed from the belief that such requirements can never be met in social science research. We do not share this belief. When the basic requirements of the method we describe are approximated in practice, they open the way to measuring systems which enable objective comparisons, useful predictions and the construction of firm bases for the study of psychological development.
The first requirement for making good measures is good raw material. When you construct a variable for yourself, you will have some control over the data you collect. Sometimes, however, the data you work with will be brought to you by someone else, and you will have to do your best with whatever you get. In either case, the materials you work with will have been gathered with a particular purpose in mind. At the heart of this purpose will be the
intention to make comparisons along a particular line of inquiry. To achieve this the data must contain the possibility of a single variable along which persons can be measured.