Akihito Kamata

A Note on the Relation Between Factor Analytic and Item Response Theory Models

The relations among several alternative parameterizations of the binary factor analysis model and the 2-parameter item response theory model are discussed. It is pointed out that different parameterizations of factor analysis model parameters can be transformed into item response model theory parameters, and general formulas are provided. Illustrative data analysis is provided to demonstrate the transformations.

College women with eating disorders: self-regulation, life satisfaction, and positive/negative affect.

Abstract The authors examined the self-regulatory strategies and subjective well-being of students recently diagnosed with eating disorders, at-risk students, and individuals without eating disorders. Fifty-six college students were individually interviewed regarding their use of self-regulatory strategies to lose and maintain their weight; they also completed the Extended Satisfaction with Life Scale (V. C. Alfonzo, D. B. Allison, D. E. Rader, & B. S. Gorman, 1996) and the Positive and Negative Affect Scale (D. Watson, L. A. Clarck, & A. Tellegen, 1988). As hypothesized, students with eating disorders reported more self-regulated strategies for managing their weight, a lower level of life satisfaction, and higher levels of negative affect than did at-risk students or individuals with normal weights. At-risk students reported higher levels of self-regulation and negative affect than did the students with normal weights. These findings may be useful for parents and health practitioners providing care to college students, who must be made aware of the signs and symptoms of these disorders.

The Role of Collective Efficacy in Adventure Racing Teams

This study examines the relationship between collective efficacy and performance in a single competition of adventure racing. Adventure racing is a team-based sport that requires the multidisciplinary tasks of trekking, mountain biking, canoeing, and climbing to navigate through a preplanned racecourse. Seventeen teams competing in an adventure race completed measures of prior performance, preparation effort, and a collective efficacy assessing perceptions of their teams functioning in six performance areas. Three in-race measures of collective efficacy and environmental factors-conditions are taken at various checkpoints. A correlational analysis indicates a positive relationship between preparation effort and initial perceptions of collective efficacy. A repeated measures analysis reveals the dynamic nature of collective efficacy and the reciprocal relationship between efficacy and performance. The results are consistent with D. L. Feltz and C. D. Lirggs (1998) examination of collegiate teams and A. Banduras (2000) contention that collective efficacy fosters a sense of motivational investment and an increased sense of staying power.

Affect‐related performance zones: An idiographic method for linking affect to performance

Abstract A new method for establishing distinct affect‐related performance zones (APZs) using an ordinal logistic regression is presented. The method is illustrated on three female collegiate tennis players observed during the 2001/2002 collegiate tennis season. APZs were determined for arousal and pleasantness dimensions of affect and their perceived functionality for performance. Once APZs were established for each player across the entire season, affect dimensions and their perceived functionality for each game were profiled separately to classify performance quality for each match (i.e., second‐order APZ). After each match the players were administered the positive‐negative affect scale (PNA; Hanin, 2000) and the flow state scale (FSS; Jackson & Marsh, 1996). Results revealed that APZs were found to be distinct and individual. When APZ was “primarily in the optimal zone,” affect was perceived as helpful to performance. However, during “mostly moderate, somewhat poor” APZ, affect was perceived as somewhat helpful and pleasant, but also to some extent unpleasant and harmful. Furthermore, flow was experienced to a greater extent during optimal and near optimal APZs. Autotelic experience and possessing clear goals were the dimensions of strongest intensity during primarily optimal and somewhat moderate/optimal APZs. The article offers both researchers and practitioners a convenient new direction in the study and implementation of affect‐performance linkage through the use of nonintrusive measures. It emphasizes an idiographic approach to the study of affect in sport as a sound alternative to the correlational methods, which were limited in both theoretical and practical aspects.

Multilevel Rasch Models

Over the past few years, several studies have investigated and demonstrated the relationships between generalized linear mixed models (GLIMM) and item response modeling. Some benefits associated with this GLIMM-based modeling framework include the modeling of nested structure of data, such as examinees nested within schools (Kamata, 2001), of multidimensional measures (Cheong & Raudenbush, 2000), and of wider class of item response models, such as 2PL item response model (Rijmen et al., 2003).

Increasing the Precision of Subscale Scores by Using Out-of-Scale Information:

In this study, the precision of subscale score estimates was evaluated when out-of-scale information was incorporated. Procedures that incorporated out-of-scale information and only information within a subscale were compared through a series of simulations. It was revealed that more information (i.e., more precision) was always provided for subscale score estimates when out-of-scale information was used. The degree of the information gain depended on the number of out-of-scale items, the magnitude of item discrimination power, and the magnitude of subscale-trait correlation. Also, the accuracy of subscale score estimates was evaluated. Contrary to precision, subscale score estimates were somewhat more biased with out-of-scale information when there were more out-of-scale items and/or when out-of-scale items had high item discrimination power. This tendency was more apparent when the correlation between subscale traits was low. It was concluded that subscale-trait correlation is an important factor to be considered when out-of-scale information is used.

A Bifactor Multidimensional Item Response Theory Model for Differential Item Functioning Analysis on Testlet-Based Items.

A differential item functioning (DIF) detection method for testlet-based data was proposed and evaluated in this study. The proposed DIF model is an extension of a bifactor multidimensional item response theory (MIRT) model for testlets. Unlike traditional item response theory (IRT) DIF models, the proposed model takes testlet effects into account, thus estimating DIF magnitude appropriately when a test is composed of testlets. A fully Bayesian estimation method was adopted for parameter estimation. The recovery of parameters was evaluated for the proposed DIF model. Simulation results revealed that the proposed bifactor MIRT DIF model produced better estimates of DIF magnitude and higher DIF detection rates than the traditional IRT DIF model for all simulation conditions. A real data analysis was also conducted by applying the proposed DIF model to a statewide reading assessment data set.

Modeling Nonlinear Growth with Three Data Points: Illustration with Benchmarking Data.

The purpose of this article is to demonstrate ways to model nonlinear growth using three testing occasions. We demonstrate our growth models in the context of curriculum-based measurement using the fall, winter, and spring passage reading fluency benchmark assessments. We present a brief technical overview that includes the limitations of a growth model with three time points, and how nonlinear growth can be modeled and the associated limitations. We present results for a piecewise growth mixture modeling approach to model nonlinear growth for 1 to 3 classes, as well as to further explain individual differences and to capture heterogeneity of growth patterns. We discuss our interpretation of these results, as well as the implications of different methods for modeling nonlinear growth with three occasions.

Differential item functioning due to gender between depression and anxiety items among Chilean adolescents

Background: Although much is known about the higher prevalence of anxiety and depressive disorders among adolescent females, less is known about the differential item endorsement due to gender in items of scales commonly used to measure anxiety and depression. Aims: We conducted a study to examine if adolescent males and females from Chile differed on how they endorsed the items of the Youth Self Report (YSR) anxious/depressed problem scale. We used data from a cross-sectional sample consisting of 925 participants (mean age = 14, SD 1.3, 49% females) of low to lower-middle socioeconomic status. Methods: A two-parameter logistic (2PL) IRT DIF model was fit. Results: Results revealed differential item functioning (DIF) by gender for six of the 13 items, with adolescent females being more likely to endorse a depression item while males were found more likely to endorse anxiety items. Conclusions: Findings suggest that items found in commonly used measures of anxiety and depression symptoms may not equally capture the true levels of these behavioural problems in adolescent males and females. Given the high levels of mental disorders in Chile and the surrounding countries, further attention should be focused on increasing the number of empirical studies examining potential gender differences in the assessment of mental health problems among Latin American populations to better aid our understanding of the phenomenology and determinants of these problems in the region.

Centering, Scale Indeterminacy, and Differential Item Functioning Detection in Hierarchical Generalized Linear and Generalized Linear Mixed Models

In this article, we discuss and illustrate two centering and anchoring options available in differential item functioning (DIF) detection studies based on the hierarchical generalized linear and generalized linear mixed modeling frameworks. We compared and contrasted the assumptions of the two options, and examined the properties of their DIF estimates with a simulation study. For reference purposes, the results were compared to those obtained from using the Mantel-Haenszel procedure as well. Finally, we discuss some implications regarding the choice of model parameterizations for DIF detection using these frameworks.