The purpose of this study was to examine the potential impact of selected methodological factors on the validity of conclusions from reliability generalization (RG) studies. The study focused on four factors; (1) missing data in the primary studies; (2) transformation of sample reliability estimates; (3) use of sample weights for estimating mean…

An item response theory (IRT) model is used as a measurement error model for the dependent variable of a multilevel model where tests or questionnaires consisting of separate items are used to perform a measurement error analysis. The advantage of using latent scores as dependent variables of a multilevel model is that it offers the possibility of…

Single studies, by themselves, rarely explain the effect of treatments or interventions definitively in the social sciences. Researchers created meta-analysis in the 1970s to address this need. Since then, meta-analytic techniques have been used to support certain treatment modalities and to influence policymakers. Although these techniques…

A test which allows for errors of measurement is derived for the hypothesis that all the members of a population who possess a certain skill are a sub-set of the members who possess another skill. (Author)

Author supports the credibility of the propositions that: (1) the true component of a score is proportional to the number of equivalent elements that contribute to it. And, (2) the error component of a score is proportional to the square root of the number of equivalent elements that contribute to it. (Author/MB)

The standard error of an equipercentile equating is derived for four situations. Some numerical results are checked by Monte Carlo methods. Numerical standard errors are computed for two sets of real data. Standard errors of linear and equipercentile equating are compared. (Author)

Multivariate generalizability deals with the interpretation of a profile or composite of measurements. Using data from the Beginning Teacher Evaluation Study to illustrate the differing interpretations of viewing test data from universe and multivariate perspectives, this chapter concludes with computer programs that the reader can implement to…

A simple, usefully accurate approximation of the standard error of measurement is proposed for use by classroom teachers. An empirical comparison with Lord's approximation indicated that, though not as easy to calculate as Lord's, this approximation is more practical because it is useful at any point in the score distribution. (BW)

For tests used to make pass/fail decisions, the relevant standard error of measurement (SEM) is the SEM at the passing score. If the test is highly stratified, this SEM should be estimated by a split-halves approach. A formula and its derivation are provided. (Author)

A mathematical link between test reliability and test validity is derived, taking into account the correlation between error scores on a test and error scores on a criterion measure. When this correlation is positive, the "paradoxical" nonmonotonic relation between test reliability and test validity occurs universally. (Author/BW)

Glass's estimator of effect size, the sample mean difference divided by the sample standard deviation, is studied in the context of an explicit statistical model. The exact distribution of Glass's estimator is obtained and the estimator is shown to have a small sample bias. Alternatives are proposed and discussed. (Author/JKS)

Developed an alternative method of administering Halstead's category test using answer sheet and latent imager developer. There was lessened possibility of examiner error in providing reinforcement and in recording responses. Performance on alternative and standard methods by 50 subjects was the same. (Author)

Examines the prevalence of inflated Type I error in counseling research and recommends wider use of multivariate statistics to correct the problem. Type I error becomes inflated beyond acceptable levels when researchers perform individual univariate statistics on each of several dependent variables within a single project. (Author)

Assuming conditional normality and independence, standard estimation and hypothesis testing procedures for regression coefficients are the same whether using fixed x values, random x values or mixed fixed and random values. Contrary to Sockloff's suggestion, polynomial regression is useful with either fixed or random models. (See EJ 142 050). (CP)

Methods by which structural models correct for the effects of attenuation due to measurement error are reviewed, and implications of such disattenuation for interpreting the results of structural equation models are considered. Recommendations are made for improving the practice of disattenuation, and caution is urged in drawing inferences based…