Studies the comparative componential skill analysis across both reading proficiency levels and processing demand conditions. Finds that initial predictions were generally supported: significant variance contribution models were found for comprehension disabled readers; skill components were significant predictors; and sensitive eye movement…

Describes a study that examined the usefulness of undergraduate and graduate grade point averages, Graduate Record Examination general test scores, and years since graduation as predictors of professional salaries for librarians. Correlations were found for men and persons employed in some types of libraries. Suggestions for further research are…

Challenges the prevailing view that educational effectiveness in Third World countries is related to school-based factors rather than student background variables. Uses a study of academic achievement in 32 Zimbabwean schools to illustrate the use of a multilevel regression model for school effectiveness research. Contains 42 references. (SV)

Outlines some of the errors and biases in thinking that psychologists have documented in recent years, including the eyewitness fallacy, underutilization of statistics, confirmation bias, misperceptions of risk, sample errors and biases, and misunderstanding of regression. Argues that journalism educators need to bring these to the attention of…

School effectiveness indices based on residuals from regressing test performance onto prior test performance and socioeconomic status were obtained for two consecutive years for 431 elementary schools. The resulting indices were found to be stable, and it was found that grade-within-school effects dominated global school effects. (TJH)

The study with over 2,500 children (ages 9-14) found the two approaches identified somewhat different groups of children as reading disabled despite little difference in performance on the Halstead-Reitan Neuropsychological Battery for Children. (Author/DB)

Canonical analysis of two convex polyhedral cones involves looking for two vectors whose square cosine is a maximum. New results about the properties of the optimal solution to this problem are presented. The convergence of an alternating least squares algorithm and properties of limits of calculated sequences are discussed. (SLD)

Several different methods for projecting grade-progression ratios were tested with 12 years of grade-specific membership data for Virginia. In nearly all cases, projections made by exponential smoothing produced more accurate results than series produced by averages, moving averages, or regression methods. (SLD)

A method is presented for extending full versus restricted model testing to predictive discriminant analysis when the criterion of model accuracy is classification hit rate. An example is presented involving the dropout or school persistence of 76 5th graders, and a computer program for the analysis is introduced. (SLD)

A study of the ability of different models--including the classification techniques of discriminant analysis, logistic regression, and neural networks--to predict the academic success of master's degree students in business administration suggests that prediction is difficult, but that classification and nonparametric techniques may be…

An equation derived by W. A. Nicewander and J. M. Price relating statistical power to reliability of dependent variable measures when true score regression is homogeneous across treatment conditions is enhanced through overcoming the problem of directly estimating the squared linear correlation between true scores for X and Y. (SLD)

The Regression Discrepancy Model, intended to increase accuracy in assessing severe discrepancy between IQ and achievement scores in students with learning disabilities, is examined. The model's initial equation is shown to produce results which bias the detection of severe discrepancy at lower IQ levels. Methods to minimize or remove this bias…

A method is presented for estimating the item characteristic curve (ICC) using polynomial regression splines. Estimation of spline ICCs is described by maximizing the marginal likelihood formed by integrating ability over a beta prior distribution. Simulation results compare this approach with the joint estimation of ability and item parameters.…

Because they used an inappropriate statistical procedure, J. H. Kranzler and A. R. Jensen (1991) have not demonstrated that a factor of general intelligence ("g") depends on several independent factors. A factorial reanalysis of their data suggests that speed and efficiency of information processing are important in "g." (SLD)

The hypothetical idea of a perfectly pure psychometric "g" is empirically unattainable. Because the unity of "g" cannot be proved or disproved by factor analysis, the unitary "g" hypothesis represents a parsimonious assumption. J. B. Carroll's (1991) analysis demonstrates the relationship between psychometric and…