Cross-validation is a common statistical procedure applied to problems that are otherwise computationally intractable. It is often employed to assess the effectiveness of prediction procedures. In this report, cross-validation is discussed in terms of "U"-statistics. This approach permits consideration of the statistical properties of…

We derive formulas for the differential item functioning (DIF) measures that two routinely used DIF statistics are designed to estimate. The DIF measures that match on observed scores are compared to DIF measures based on an unobserved ability (theta or true score) for items that are described by either the one-parameter logistic (1PL) or…

The finite population correction (FPC) factor is often used to adjust variance estimators for survey data sampled from a finite population without replacement. As a replicated resampling approach, the jackknife approach is usually implemented without the FPC factor incorporated in its variance estimates. A paradigm is proposed to compare the…

This report presents results on a parallel implementation of the expectation-maximization (EM) algorithm for multidimensional latent variable models. The developments presented here are based on code that parallelizes both the E step and the M step of the parallel-E parallel-M algorithm. Examples presented in this report include item response…

In this study, we examined the potential impact of item misfit on the reported scores of an admission test from the subpopulation invariance perspective. The target population of the test consisted of 3 major subgroups with different geographic regions. We used the logistic regression function to estimate item parameters of the operational items…

Quantitative information about test score reliability can be presented in terms of the distribution of equated scores on an alternate form of the test for test takers with a given score on the form taken. In this paper, we describe a procedure for estimating that distribution, for any specified score on the test form taken, by estimating the joint…

Several factors could cause variability in item response theory (IRT) linking and equating procedures, such as the variability across examinee samples and/or test items, seasonality, regional differences, native language diversity, gender, and other demographic variables. Hence, the following question arises: Is it possible to select optimal…

A general program for item-response analysis is described that uses the stabilized Newton-Raphson algorithm. This program is written to be compliant with Fortran 2003 standards and is sufficiently general to handle independent variables, multidimensional ability parameters, and matrix sampling. The ability variables may be either polytomous or…

In operational equating, frequency estimation (FE) equipercentile equating is often excluded from consideration when the old and new groups have a large ability difference. This convention may, in some instances, cause the exclusion of one competitive equating method from the set of methods under consideration. In this report, we study the…

The purpose of this paper is to extend von Davier, Holland, and Thayer's (2004b) framework of kernel equating so that it can incorporate raw data and traditional equipercentile equating methods. One result of this more general framework is that previous equating methodology research can be viewed more comprehensively. Another result is that the…

This paper describes an empirical evaluation of a Bayesian procedure for equating scores on test forms taken by small numbers of examinees, using collateral information from the equating of other test forms. In this procedure, a separate Bayesian estimate is derived for the equated score at each raw-score level, making it unnecessary to specify a…

The purpose of this study was to empirically evaluate the impact of loglinear presmoothing accuracy on equating bias and variability across chained and post-stratification equating methods, kernel and percentile-rank continuization methods, and sample sizes. The results of evaluating presmoothing on equating accuracy generally agreed with those of…

At present, although the percentages of students with disabilities (SDs) and/or students who are English language learners (ELL) excluded from a NAEP administration are reported, no statistical adjustment is made for these excluded students in the calculation of NAEP results. However, the exclusion rates for both SD and ELL students vary…

In item-response theory, if a latent-structure model has an ability variable, then elementary information theory may be employed to provide a criterion for evaluation of the information the test provides concerning ability. This criterion may be considered even in cases in which the latent-structure model is not valid, although interpretation of…

Confidence intervals are an important tool to indicate uncertainty of estimates and to give an idea of probable values of an estimate if a different sample from the population was drawn or a different sample of measures was used. Standard symmetric confidence intervals for proportion estimates based on a normal approximation can yield bounds…