This study builds on prior research on adaptive testing by examining the performance of item calibration methods in the context of multidimensional multistage tests with within-item multidimensionality. Building on the adaptive module-level approach, where test-takers proceed through customized modules based on their initial performance, this…

Exact testing was evaluated as a method for conducting Mantel-Haenszel differential item functioning (DIF) analyses with relatively small samples. A series of computer simulations found that the asymptotic Mantel-Haenszel and the exact method yielded very similar results across sample size, levels of DIF, and data sets. (SLD)

Computer simulations under three conditions of polytomous differential item functioning (DIF) compared the ability of three different statistical procedures to detect nonuniform DIF. The procedures were a nominal and an ordinal extension of the Mantel-Haenszel statistic, and logistic discriminant function analysis. Results showed that only the…

An item response theory-based parametric procedure proposed by N. S. Raju, W. J. van der Linden, and P. F. Fleer (1995) known as differential functioning of items and tests (DFIT) can be used with unidimensional and multidimensional data with dichotomous or polytomous scoring. This study describes the polytomous DFIT framework and evaluates and…

Hierarchical Bayes procedures for the two-parameter logistic item response model were compared for estimating item and ability parameters through two joint and two marginal Bayesian procedures. Marginal procedures yielded smaller root mean square differences for item and ability, but results for larger sample size and test length were similar.…

A didactic example is provided, using a Monte Carlo method, of how differential item functioning (DIF) can be eliminated (and thus better understood) when the complete latent space is used. The main source of DIF is that the matching single criterion used in some DIF procedures, Mantel Haenszel or Simultaneous Item Bias (SIBTEST), does not account…

The impact of capitalizing on chance in item selection on the accuracy of test information functions was studied through simulation, focusing on examinee sample size in item calibration and the ratio of item bank size to test length. (SLD)

Type I error rates of the likelihood ratio test for the detection of differential item functioning (DIF) were investigated using Monte Carlo simulations. The graded response model with five ordered categories was used to generate data sets of a 30-item test for samples of 300 and 1,000 simulated examinees. All DIF comparisons were simulated by…

The results obtained from two basic types of pre-equatings of tests were compared: the item response theory (IRT) pre-equating and section pre-equating (SPE). The simulated data were generated from a modified three-parameter logistic model with a constant guessing parameter. Responses of two replication samples of 3000 examinees on two 72-item…

Item parameter estimation errors in test development are highlighted. The problem is illustrated with several simulated data sets, and a conservative solution is offered for addressing the problem in item response theory test development practice. Steps that reduce the problem of capitalizing on chance in item selections are suggested. (SLD)

Errors in item parameter estimates have a negative impact on the accuracy of item and test information functions. The estimation errors may be random, but because items with higher levels of discriminating power are more likely to be selected for a test, and these items are most apt to contain positive errors, the result is that item information…

The purpose of this study was to compare two non-parametric procedures, the Mantel-Haenszel (MH) procedure and the simultaneous item bias (SIB) procedure, with respect to their Type I error rates and power, and to investigate the conditions under which asymptotic distributional properties of the SIB and MH were obtained. Data were simulated to…

Equating tests from different calibrations under item response theory (IRT) requires calculation of the slope and intercept of the appropriate linear transformation. Two methods have been proposed recently for equating graded response items under IRT, a test characteristic curve method and a minimum chi-square method. These two methods are…

A two-stage procedure for estimating item bias was examined with six indices of item bias and the Mantel-Haenszel statistic. Results suggest that the two-stage procedure is not very useful when the number of biased items is small and bias magnitude is weak. (SLD)

In this article some results are presented relating to the dimensionality of instruments containing polytomously scored as well as dichotomously scored items, concentrating on the 1992 National Assessment of Educational Progress' (NAEP) mathematics and reading assessment data and several simulated datasets. The maximum likelihood factor analytic…