PROPOSING AN ACHIEVEMENT SIMULATION METHODOLOGY TO ALLOW THE ESTIMATION OF INDIVIDUAL IN CLINICAL TESTING CONTEXT

Abstract

Information from the population should be directly transposed to the individual level only under strict conditions of stationarity and homogeneity. In general, psychological phenomena are neither stationary nor homogeneous. Furthermore, the individual parameters must be estimated. The usual techniques of estimating the individual are rarely attainable. This occurs because in order to get valid estimates of individual parameters, many occasions of the same test must be performed. In this article, we propose an approach (simerg algorithm) to estimate the individual, so that a proper clinical testing for the individual in the context of testing setting that require the respondents' performance becomes viable. We present the rationale of simerg and apply it to estimate the verbal comprehension ability of three individuals, which performed, in 90 different occasions, the 60 items of this ability. In this application, we show  how simerg uses the initial empirical performance of these persons and simulates their whole  performance. We evaluate the validity of simerg using different measures to predict the empirical performance. We conclude that simerg produces valid estimates. Further studies in a variety of populations and tests that require respondents' performance are needed to enable a stronger conclusion about the validity of simerg.