Variation in Initial Procedures and the Stability of Dimensions Extracted from Free Response Data

The purpose of the research was to examine the stability of dimensions extracted from a body of free response data under conditions of variation in the initial data reduction procedures. Data consisted of over 1500 expressions of concern and questions raised by an elderly population. Complete sequences of analytic steps were conducted on the data by two groups of experimenters working independently. These sequences included determining representative units to be scaled, developing instructions for obtaining proximity measures, recruiting subjects, and administering the sorting tasks necessary to obtain proximity measures. In spite of dramatic differences in procedures at each of these steps, the structures of the resulting multidimensional configurations obtained by the two groups were identical. Selected, independently measured properties were used to interpret the configurations. In both cases, a three dimensional solution defined by the properties Health Related, Quality of Lire, and Individual versus Societal Responsibility appeared optimal.