This article examines simple and multiple regression analysis as forecasting tools, and details the process by which multiple regression analysis may be used to increase the accuracy of the technology forecast. (CT)

The type I error rate in stepwise regression analysis deserves serious consideration by researchers. The problem-wide error rate is the probability of selecting any variable when all variables have population regression weights of zero. Appropriate significance tests are presented and a Monte Carlo experiment is described. (Author/CTM)

Three methods of handling disproportionate cell frequencies in two-way analysis of variance are examined. A Monte Carlo approach was used to study the method of expected frequencies and two multiple regression approaches to the problem as disproportionality increases. (Author/JKS)

A statistical analysis procedure is developed, based on the notion that many educational programs are dynamic interventions in natural growth processes, and is called value-added analysis. The theory of value-added analysis, and several applications are presented. (Author/JKS)

Computer programs for analysis of covariance use classical experimental, regression, or hierarchical methods of least squares. In a 3 X 3 factorial experiment with equal cell frequencies, three solutions yielded different sums of squares for main effects although correlation between variables was negligible and cell frequencies were equal.…

Three traditional methods of selection of variables to be included in a "best" regression equation are compared to a method designed to maximize weight validity. Implications for constructing regression equations for prediction are discussed, with consideration of the weight validity maximization method recommended in crucial situations.…

When sample sizes and/or X intervals are unequal, the analysis of variance computations for trend analysis become quite complicated. This article shows how multiple regression/correlation analysis may be applied in order to accomplish with great simplicity trend analysis under "irregular" conditions. (Author/RL)

A distinction is drawn between redundancy measurement and the measurement of multivariate association between two sets of variables. Several measures of multivariate association between two sets of variables are examined. (Author/JKS)

Since actual selection can be different from the selection as it is intended, a method is described for clarifying "restriction of range" problems in developing selection/prediction equations. The application of the method is illustrated in a case study. (Author/JKS)

Fifteen examples using different formulas for calculating degrees of freedom for power analysis of multiple regression designs worked out by Cohen are presented, along with a more general formula for calculating such degrees of freedom. (Author/JKS)

With respect to each of the seven decision-task variables, there was no significant difference between the two groups (of 100 men and 100 women executives). The multiple regression analysis did not show sex of executives as a factor influencing confidence, dogmatism, and decision latency as related to decision accuracy. (Author/IRT)

The hypothesis that two variables have a perfect disattenuated correlation and hence measure the same trait, except for errors of measurement, is discussed. Equivalently, the underlying variables, the true scores, are related linearly. It is shown that previously proposed ad hoc tests are, in fact, likelihood ratio tests. (Author/JKS)

The utility of a nonlinear transformation of the criterion variable in multiple regression analysis is established. A well-known law--the Pythagorean Theorem--illustrates the point. (Author/JKS)

Estimators which are optimal under assumptions of normality are shown to be vulnerable to the effects of outliers. A survey of robust alternatives is presented. Included are alternatives to the mean, standard deviation, product-moment correlation, t-test, analysis of variance, multivariate techniques, and schemes for outlier detection. (Author/JKS)

A multiple regression approach is used to assess the feasibility of reciprocal prediction between the Sixteen Personality Factor Questionnaire scales and the California Psychological Inventory scales (i.e., the prediction of each 16PF scale from the CPI scales and of each CPI scale from the 16PF scales). (RC)