Date of Award
Doctor of Philosophy (PhD)
The present investigation was essentially exploratory and descriptive in purpose. Utilizing the direct scaling method of magnitude estimation (with assigned modulus) the study attempted to: (1) determine if lawful relationships existed between the clinical judgment of licensed physicians and the results from nine numerical medical information indices, and (2) to quantitatively describe such relationships.
It was believed that direct estimation methodologies have been shown through empirical studies to be superior to the psychophysical models of Fechner and Thurstone. Previous work has also illustrated that the power law of S. S. Stevens' has provided a powerful methodology in studying the topic of clinical judgment.
In the present study 27 licensed physicians served as judges. They judged results from nine frequently used numerical medical information indices which were varied systematically and independently. Ah upper and lower limit for each of the nine indices was determined from the medical literature and medical consultants. Specific stimuli within these limits were spaced in equal logarithmic steps when feasible. Judgments were made relative to degree of concern for a contrived 35 year old patient's health status. The laboratory test-indices and the various levels of each test were presented in randomized orders. For four of the indices, levels above as well as below normal were included. In scale development, these were considered separately, thus 13 subjective scales were developed.
In general, the results indicate that for nine of these scales the relationships observed were curvilinear when degree of concern was plotted against the appropriate stimulus metric. A log-log transformation rectified the data so that straight lines offered reasonably good approximations of the observed trends. It was determined that a power function model was an appropriate description of these data. For four indices the relationships, when degree of concern was plotted against the stimulus continuum, were markedly linear in nature. It was suggested that: (1) the underlying continua for these four indices may be metathetic, or (2) that physicians view these four indices as some sort of ordered category measures even though the underlying stimulus measures are continuous in nature.
Implications for the direct estimation literature seem clear. This study represents one of the earliest successful extensions of these measurement methodologies into the topic of clinical judgment. It was suggested that direct estimation procedures are sufficiently sensitive to assist in the clarification of the many enigmatic ambiguities now existant in the clinical judgment literature.
Implications for medical education were also drawn. The development of scales similar to those produced in this inquiry could provide valuable communication vehicles whereby the "exigencies of the office would be brought into the classroom."
For several scales the predetermined standard was believed to be disparet from the intrinsic standard employed by the judges. This was believed to increase variability or noise in the measurement system. Inter-scale comparisons were also made and four of the indices seemed more potent in terms of eliciting concern. One index appeared to elicit relatively little concern. Data derived in connection with inter-scale comparisons holds potential for future research into this area.
Limitations of the present inquiry were discussed. For example, the sample was in no way random or systematic, and several standards which were employed seemed to be inappropriate. Suggestions for future research were also advanced.
Theye, Frederick W., "Physicians' Degree of Concern as a Function of Numerical Medical Information Indices" (1969). Theses and Dissertations. 3575.