The results of six major projects are discussed including a comprehensive mathematical and statistical analysis of the problems caused by errors of measurement in linear models for assessing change. In a general matrix representation of the problem, several new analytic results are proved concerning the parameters which affect bias in observed-score regression statistics. The bias in ordinary least squares estimators is expressed as a function of covariances among true scores, among the measurement errors, and sample size. The first two projects were employed to create an algorithm for assessing the potential bias due to the unreliability of measures. The algorithm was implemented as a FORTRAN program to improve the design of investigations of change and minimize potential errors of inference. A review is presented of statistical methods which have been developed in several disciplines to estimate the parameters of true change by correcting the observed-score regression estimates for unreliability. A series of Monte Carlo experiments which evaluated the performance of the methods are discussed. The advantages and general superiority of estimators proposed by Fuller are examined. The relevance of a special linear functional relation (LFR) model and models devised for estimating the parameters of LFRs are compared. (Author/CM)