Artifact Corrections and Model Assumptions
As long as the data satisfy the model assumptions, the estimates should be satisfactory and plausible. No restrictions need to be placed on how the model parameters ρ j and their estimates are viewed. They may be regarded as a product of other parameters if desired. However, in many practical testing settings, measurement unreliability and range restriction are important considerations and cannot be regarded as negligible. The obvious question arises as to how the model or procedures can be altered to allow for corrections of these influences. There are several possible approaches to the problem. Because corrections for unreliability and corrections for range restriction present different kinds of problems, each will be considered separately.
KeywordsMixture Model Empirical Distribution Function Bivariate Normal Distribution Normal Mixture Validity Coefficient
Unable to display preview. Download preview PDF.