Fundamental Measurement for Health Services Research

摘要

This study describes and demonstrates the mathematical properties of additive formulations of probabilistic conjoint measurement (PCM), which provide the necessary and sufficient conditions for fundamental measurement. PCM was applied to data from 23,767 persons responding to AHRQ's Medical Expenditure Panel Survey. Five hypotheses were tested: (1) that the measured quality of care (QOC) variable is quantitative, (2) that the data's ordering of the survey items and rating categories is meaningful, (3) that the respondent measurement order remains constant over item subsamples, and that the item calibration order remains constant over respondent subsamples, (4) that there are no substantive differences among the measures associated with different forms of insurance coverage, and (5) that each of three approaches to measurement (PCM, Item Response Theory, and the method of summated ratings) are appropriate for general use as scientific measurement methods. None of the hypotheses were rejected, though (3) and (5) are strongly qualified. The results indicate two general conclusions: (1) PCM offers new insights into theories of QOC, and (2) It is possible to construct fundamental measurement systems of QOC variables. Since QOC fundamental measurement systems are viable, it should be feasible to scale different QOC surveys in the same metric.