A general model for prediction and association is described for the situation in which both criterion and predictor(s) are discrete variables. The approach can be employed for the general case involving any number of predictors, and the related measures of multiple and partial association are described. (Author/GK)

A procedure for inferring the validity of a selection test as a predictor of some criterion when the available data are limited due to prior selection is described. (Author/JKS)

Models proposed by Cleary, Thorndike, Cole, Einhorn and Bass, and Darlington for analyzing bias in the use of tests in a selection strategy are surveyed. Six additional models for test bias are also introduced. The purpose is to describe, compare, contrast, and evaluate these models while, at the same time, extracting such useful ideas as may be…

Confirmatory factor analysis was applied to test second- and higher-order factor models in the areas of structure of abilities, allometry, and the separation of specific and error variance estimates. The estimation of validity and reliability, second-order models within factor analysis models, and the concept of discriminability were also studied.…

The use of analysis of covariance in conjunction with the multivariate approach to analyzing repeated measures designs is considered for designs involving between- and within-subject factors, one dependent variable, and one observation per subject on the covariate. (Author/RL)

An alternative definition has been developed of the delta scale of item difficulty used at Educational Testing Service. The traditional delta scale uses an inverse normal transformation based on normal ogive models developed years ago. However, no use is made of this fact in typical uses of item deltas. It is simply one way to make the probability…

Eight formula-based estimates of population squared cross-validity and nine estimates of factor scores were compared with estimates from the conventional cross-validation procedure. Burket's and Rozeboom's formula-based estimates of population squared cross-validity in combination with a factor score method offer the most for practitioners…

When either regression models or subjectively-weighted models are used as aids in making placement decisions, the discriminant validity of these models is questioned. The validity of several regression models and of subjectively weighted models was investigated in two experiments. (Author/JKS)

This supplementary report identifies and provides individual descriptions and reviews of 71 retirement forecasting models. Composed of appendices, it is intended as a source of more detailed information than that included in the main volume of the report. Appendix I is an introduction. Appendix II contains individual descriptions of 32 models of…

The impact of misspecification on the estimation, testing, and improvement of structural equation models was assessed via a population study in which a prototypical latent variable model was misspecified. Results provide insights into the maximum likelihood estimator versus a limited two-stage least squares estimator in LISREL. (TJH)

Prediction of success in a special program often guides program participant selection and treatment. If the statistical analysis used to develop such predictions is based on the special selectivity built into the program, the predictions can be misleading and the validation tests uninformative. Federal district court pretrial release data is used…

The use of two-stage least squares (2SLS) for the estimation of feedback linkages is inappropriate for nonorthogonal data sets because 2SLS is extremely sensitive to multicollinearity. It is argued that what is needed is use of a different estimating criterion than the least squares criterion. Theoretically the variance normalization criterion has…

The unbiased estimate of a "treatment effect" reached by analysis of covariance in a nonrandomized experiment would often require that a different covariate be used in each treatment. A sufficient but unlikely condition for an unbiased estimate is that the covariate for each treatment is (1) the complete covariate that predicts the outcome as…

The effect of test length on predictive validity is examined empirically. For four tests, the curve of validity against test length had a very gentle slope for the longer tests and all tests could be reduced by more than 60 percent without appreciable decreases in validity. (Author/BW)

Applies the Box-Jenkins time series methodology to enrollment data for the Virginia community college system. Describes the enrollment data set, the Box-Jenkins approach, and the forecasting results. Discusses the value of one-quarter ahead enrollment forecasts and implications for practice. Provides a technical discussion of the model. (DMM)