A method for detecting and interpreting disturbances of the local-independence assumption among items that share common stimulus material or other features is presented. Dichotomous and polytomous Rasch models are used to analyze structure of the learning outcome superitems. (SLD)

The construction of parallel tests from item response theory (IRT) based item banks is discussed. Tests are considered parallel whenever their information functions are identical. After the methods for constructing parallel tests are considered, the computational complexity of 0-1 linear programming and the heuristic procedure applied are…

Explored are two theoretical approaches that attempt to cope with omitted responses, that is, when an examinee omits (fails to respond to) an item and therefore the item response formula cannot be used. Preliminary considerations are discussed, and it is shown that a conveniently simple application of equivalent items leads to internal…

A "maximin" model for item response theory based test design is proposed. In this model only the relative shape of the target test information function is specified. It serves as a constraint subject to which a linear programming algorithm maximizes the information in the test. In the practice of test construction there may be several…

Statistical procedures for measuring heterogeneity of test item distractor distributions, or cultural variation, are presented. These procedures are based on the notion that examinees' responses to the incorrect options of a multiple-choice test provide more information concerning cultural bias than their correct responses. (Author/PN)

A closed form approximation is given for the variance of examinee proficiency estimates in the Rasch model for dichotomous items, under the condition that only estimates, rather than true values, of item difficulty parameters are available. The term that must be added to the usual response-sampling variance is inversely proportional to both the…

The Rasch model of test theory is described in general terms, compared with latent trait theory, and shown to have interesting applications for the measurement of affective as well as cognitive traits. Three assumption of the Rasch model are stated to support the conclusion that calibration of the items and tests is independent of the examinee…

The interest in the application of large-scale computerized adaptive testing has served to focus attention on issues that arise when theoretical advances are made operational. Some of these issues stem less from changes in testing conditions and more from changes in testing paradigms. One such issue is that of the order in which questions are…

Two fundamental problems in mental test theory are to estimate true score and to estimate the amount of error when testing an examinee. In this report, three probability models which characterize a single test item in terms of a population of examinees are described. How these models may be modified to characterize a single examinee in terms of an…

Several methods of assessing test item bias are described, and the concept of fair use of tests is examined. A test item is biased if individuals of equal ability have different probabilities of attaining the item correct. The following seven general procedures used to examine test items for bias are summarized and discussed: (1) analysis of…

The potential of item response theory (IRT) for solving a number of testing problems in the Maryland Functional Reading Program would appear to be substantial in view of the many other promising applications of the theory. But, it is well-known that the advantages derived from an IRT model cannot be achieved when the fit between an item response…

On well-constructed multiple-choice tests, the most serious threat to measurement is not variation in item discrimination, but the guessing behavior that may be adopted by some students. Ways of ameliorating the effects of guessing are discussed, especially for problems in latent trait models. A new item response model, including an item parameter…

A strategy is described for dealing with measurement situations in which certain categories of responses are null, that is, persons do not respond in certain categories to certain items. The method is described for the partial credit model while maintaining the integrity of the original response framework. (SLD)

Factor analysis is the traditional method for studying the dimensionality of test data. However, under common conditions, the factor analysis of tetrachoric correlations does not recover the underlying structure of dichotomous data. The purpose of this paper is to demonstrate that the factor analyses of tetrachoric correlations is unlikely to…

A method is proposed for the detection of item bias with respect to observed or unobserved subgroups. The method uses quasi-loglinear models for the incomplete subgroup x test score x item 1 x ... x item k contingency table. If the subgroup membership is unknown, the models are the incomplete-latent-class models of S. J. Haberman (1979). The…