We investigate two non-iterative estimation procedures for Rasch models, the pair-wise estimation procedure (PAIR) and the Eigenvector method (EVM), and identify theoretical issues with EVM for rating scale model (RSM) threshold estimation. We develop a new procedure to resolve these issues--the conditional pairwise adjacent thresholds procedure…

We surveyed researchers in the health and exercise sciences to explore different areas and magnitudes of bias in researchers' decision making. Participants were presented with scenarios (testing a central hypothesis with p = 0.06 or p = 0.04) in a random order and surveyed about what they would do in each scenario. Participants showed significant…

Null distributions of permutation tests for two-sample, paired, and block designs are simulated using the R statistical programming language. For each design and type of data, permutation tests are compared with standard normal-theory and nonparametric tests. These examples (often using real data) provide for classroom discussion use of metrics…

Standardized testing has been part of the American educational system for decades. Controversy from the beginning has plagued standardized testing, is plaguing testing today, and will continue to be controversial. Given the current federal educational policies supporting increased standardized testing, psychometricians, educators and policy makers…

Two strategies, derived from J. P. Schaffer (1986), were compared as tests of significance for a complete set of planned orthogonal contrasts. The procedures both maintain an experimentwise error rate at or below alpha, but differ in the manner in which they test the contrast with the largest observed difference. One approach proceeds directly to…

Whether one should use null hypothesis testing, confidence intervals, and/or effect sizes is a source of continuing controversy in educational research. An alternative to testing for statistical significance, known as equivalence testing, is little used in educational research. Equivalence testing is useful in situations where the researcher…

A computer-based learning simulation was developed at Florida State University which allows for high interactive responding via a time-sharing terminal for the purpose of demonstrating descriptive and inferential statistics. The statistical simulation (STATSIM) is comprised of four modules--chi square, t, z, and F distribution--and elucidates the…

Five methods for testing the hypothesis of independence between two sets of variates were compared through simulation. Results indicate that two new methods, based on robust measures reflecting the linear association between two random variables, provide reasonably accurate control over Type I errors. Drawbacks to rank-based methods are discussed.…

When examining a repeated measures design with independent groups for a significant group by trial interaction, classical analysis of variance or multivariate procedures can be used if the assumptions underlying the tests are met. Neither procedure may be justified for designs with small sample sizes and dichotomous dependent variables. An omnibus…

Details of a reasonably precise normalization technique for coefficient alpha are outlined, along with methods for estimating the variance of the normalized statistic. These procedures lead to the K-sample significance test. (RC)

Stability of Type I error rates and power are investigated for three forms of the Box test and two forms of the jackknife test with equal and unequal sample sizes under conditions of normality and nonnormality. The Box test is shown to be robust to violations of the assumption of normality when sampling is from leptokurtic populations. The…

The problem of spurious significance in multivariate exploratory research is discussed. When a very large number of statistical tests are performed, many tests will be significant on the basis of chance alone. To counter this problem, the use of two sign tests to analyze sets of results has been suggested; the chance expectation [CE test] assesses…

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…