Studied a two-stage methodology for evaluating differential item functioning (DIF) in large-scale assessment data using a sample of 60,000 students taking a large-scale assessment. Findings illustrate the merit of iterative approached for DIF detection, since items identified at one stage were not necessarily the same as those identified at the…

Investigated potentially biased scale items on the Center for Epidemiological Studies Depression scale (CES-D; Radloff, 1977) in a sample of 600 adults. Overall, results indicate that the scoring method has an effect on differential item functioning (DIF), and that DIF is a property of the item, scoring method, and purpose of the assessment. (SLD)

Presents an overview of nonparametric regression as it allies to differential item functioning analysis and then provides three examples to illustrate how nonparametric regression can be applied to multilingual, multicultural data to study group differences. (SLD)

Data from over 50,000 people in 50 occupational groups were used to revise the Strong Interest Inventory. New General Reference Samples containing over 18,000 people were used to construct scales, and nearly every scale was revised. (SK)

Examined whether reviewers on item-review committees can identify accurately test items that exhibit a variety of flaws. Results with 39 reviewers of a 75-item test show that reviewers exhibit fairly high accuracy rates overall, with statistically significant differences in judgmental accuracy among reviewers. (SLD)

Studied the appropriateness and implications of using a generalizability theory approach to estimating the reliability of scores from tests composed of testlets. Analyses of data from two national standardization samples suggest that manipulating the number of passages is a more productive way to obtain efficient measurement than manipulating the…

Compared the person-response function (PRF) method for identifying examinees who respond to test items in a manner divergent from the underlying test model to the "l(z)" index of Drasgow and others (1985). Although performance of the "l(z)" index was superior in most cases, the PRF was useful in some conditions. (SLD)

Proposes an algorithm for the assembly of multiple test forms in which the multiple-form problem is reduced to a series of computationally less intensive two-form problems. Illustrates how the method can be implemented using 0-1 linear programming and gives two examples. (SLD)

Demonstrates a method for studying differential item functioning (DIF) that can be used with dichotomous or polytomous items and that is valid for data that follow a partial credit Item Response Theory model. A simulation study shows that positively biased Type I error rates are in accord with results from previous studies. (SLD)

Studied whether a change of test item sequencing, intended to increase test complexity, would cause increased involvement of general intelligence using a sample of Swedish military recruits who received heterogeneous (n=1,778) or homogeneous (n=363) tests. Items presented homogeneously showed higher general intelligence ("G") loadings.…

Generating a test from an item bank using a criterion based on classical test theory parameters poses considerable problems. A mathematical model is formulated that maximizes the reliability coefficient alpha, subject to logical constraints on the choice of items. Theorems ensuring appropriate application of the Lagragian relation techniques are…

Simulation studies indicate that, for responses to complete tests, construction of Rasch measures from observational data, followed by principal components factor analysis of Rasch residuals, provides an effective means of identifying multidimensionality. The most diagnostically useful residual form was found to be the standardized residual. (SLD)

The local dependence of item pairs is investigated through a conditional covariance function estimation procedure. The conditioning variable used is obtained by a monotonic transformation of total score on the remaining items. Conditional covariance functions are estimated by using kernel smoothing. Several models of local dependence are…

Using item mean squares to evaluate fit to the Rasch model was studied, also considering the transformed version of the item fit statistics. Simulations demonstrate that the critical value for the mean square used to detect misfit is affected by the type of mean square and the number of persons in the calibration. (SLD)

Presents the hierarchical generalized linear model (HGLM) as an explicit two-level formulation of a multilevel item response model. Shows that the HGLM is equivalent to the Rasch model, and that a characteristic of the HGLM is that person ability can be expressed as a latent regression model with person-characteristic variables. Shows that the…