A computer program randomly sampled ordered pairs of scores from known populations that departed from bivariate normal form and calculated correlation coefficients from sample values. Hypotheses were tested (1) that population correlations are zero using the t statistic; and (2) that population correlations have non-zero values using the r to z…

Each of the four classic multivariate analysis of variance (MANOVA) tests of statistical significance may lead a researcher to different decisions as to whether a null hypothesis should be rejected: (1) Wilks' lambda; (2) Lawley-Hotelling trace criterion; (3) Roy's greatest characteristic root criterion; and (4) Pillai's trace criterion. These…

In the educational research literature alpha, the a priori level of significance, and p, the a posteriori probability of obtaining a test statistic of at least a certain value when the null hypothesis is true, are often confused. Explanations for this confusion are offered. Paradoxically, alpha retains a prominent place in textbook discussions of…

This study is an analysis of the robustness of the Box-Tiao integrated moving averages model for analysis of time series quasi experiments. One of the assumptions underlying the Box-Tiao model is that all N values of alpha subscript t come from the same population which has a variance sigma squared. The robustness was studied only in terms of…

This paper develops a general method for comparing treatment magnitudes for research employing multiple treatment fixed effects analysis of variance designs, which may be used for main effects with any number of levels without regard to directionality. (Author/BS)

A recent study (Blair and Higgins, 1980) indicated a power advantage for the Wilcoxon W Test over student's t-test when calculated from a common mixed-normal sample. Results of the present study indicate that the t-test corrected for outliers shows a superior power curve to the Wilcoxon W.

A procedure was developed within hypothesis-testing logic that allows researchers to support a hypothesis that has traditionally been the statistical or null hypothesis. Four activities involved in attainment of this goal were discussed: (1) development of statistical logic needed to define the sampling distribution associated with the hypothesis…

An approach to statistical testing, which combines Neyman-Pearson hypothesis testing and Fisher significance testing, is recommended. The use of P-values in this approach is discussed in some detail. The author also discusses some problems which are often found in introductory statistics textbooks. The problems involve the definitions of…

Two easily computed test statistics are proposed for testing the equality of two correlated proportions when some observations are missing on both responses. The performance of these tests in terms of size and power is compared with other tests by means of Monte Carlo simulations. (Author/BS)

This paper discusses distribution theory and power computations for four common "tests of combined significance." These tests are calculated using one-sided sample probabilities or p values from independent studies (or hypothesis tests), and provide an overall significance level for the series of results. Noncentral asymptotic sampling…

The purpose of this study was to compare the results of an analysis of covariance design to a method developed by the authors called within-class regression. Application of this statistical method can be to any quasiexperimental design with a large number of groups. A test of statistical significance to use with the within-class regression…

The tests for zero partial association in a multiple contingency table have gained new importance with the introduction of graphical models. It is shown how these may be performed as exact conditional tests, using as test criteria either the ordinary likelihood ratio, the standard x squared statistic, or any other appropriate statistics. A…

This paper reports the results of a Monte Carlo investigation of Type I errors in the single group repeated measures design where multiple measures are collected from each observational unit at each measurement occasion. The Type I error of three multivariate tests were examined. These were the doubly multivariate F test, the multivariate mixed…

Data drawn from 30 journal articles and ERIC documents reporting on gender differences in natural science achievement were re-examined. Three meta-analysis methods were used: (1) vote counts and vote-counting estimation procedures; (2) tests of combined significance; and (3) analyses of effect sizes. The three methods produced seemingly…