The diagnostic-prescriptive model: a new approach to an old problem through multidimensional scaling

Abstract

This investigation used Individual Differences Scaling Analysis (INDSCAL) as a tool in the definition of a diagnostic-prescriptive model from a systems perspective. The study also evaluated INDSCALs' portrayal of treatment change with time. An educational facility for emotionally disturbed children was the object of investigation. The teachers responsible for treatment implementation made pairwise similarity judgements of nine clients in the program and eleven of the treatment techniques. Judgements were also made on all client-treatment pairs: each teacher judged how well client X was matched with treatment Y. These judgements were rendered every two weeks. In all, there were five data collection sessions. INDSCALs of each data session failed to display the dynamics of treatment change with time. The source of this failure was twofold: 1) the teachers did not share a common perceptual space; and 2) judgements of inter-client similarity masked the client-treatment matches. To circumvent these problems, the teachers' data were divided into two groups on the basis of a Q Components Analysis. Then, a separate solution was computed for each client. That is, only one client and eleven treatments were used for each analysis. Hence, ninety solutions were computed. This piecemeal approach permitted a two dimensional view of treatment dynamics with time. Changes in the client-treatment space with time took on an interesting form. The client maintained a relatively static position, while the treatments approached and retreated. A few client did, however, move from one quadrant of the configuration to another. The treatment configuration was thus a dynamic one, accommodating clients in different ways, at different points in time. The small sample size made the development of a treatment prescription rule difficult. Hence, another set of judgements was solicited to increase the client sample. The teachers were provided with a graphic display (Map) of the group solution from data session IV. Judgements about the location of the remaining clients in the program were then made on this two-dimensional map. Distances between all points were measured directly and then analyzed several ways with the INDSCAL program. The approach yielded a final client sample of twenty-five members. External validation of the INDSCAL solutions was accomplished with the aide of a behavioral checklist and several self-report attitude inventories, e.g., a locus of control scale. Correlations between these measures and the INDSCAL dimensions were computed. A multiple discriminant analysis also assisted in the external validation. A summary of the findings resulting from these two approaches follows. Aggressive children were treated with a combination of positive and aversive behavior management principles during the early weeks of residence. In the later weeks Glasser's Reality Therapy principles, e.g., student commitment, were used as well. Positive behavior management and an early introduction of Reality Therapy principles were used with the withdrawn children. These are general statements which apply to the teachers as a group. One set of teachers relied primarily on behavior management, since they did not view the Reality Therapy principles as effective. The INDSCAL provided a first step in the definition of a diagnostic-prescriptive model from a systems perspective. Cross validation with another sample is required. Problems with the approach and future usage are discussed.