It is observed that many sorts of difficulties may preclude the uneventful construction of tests by a computerized algorithm, such as those currently in favor in Computerized Adaptive Testing (CAT). In this essay we discuss a number of these problems, as well as some possible avenues of solution. We conclude with the development of the “testlet,” a bundle of items that can be arranged either hierarchically or linearly, thus maintaining the efficiency of an adaptive test while keeping the quality control of test construction that is possible currently only with careful expert scrutiny. Performance on the separate testlets is aggregated to yield ability estimates.

This paper focuses on whether a consistent difference by ethnicity existed in the clinical diagnosis of children and adolescents in two behavioral health service environments and reviews plausible explanations for such a difference. Key measures were clinical diagnosis and ethnicity, abstracted from the administrative dataset of a New Jersey behavioral health care organization during 2000-2002, and a data collection conducted for the State of Indiana during 1991-1992. Sample sizes were 5,394 and 10,437, respectively. Only primary diagnoses were used in this study, classified into externalizing versus internalizing disorders. Logistic regression was performed for the dependent variable of presence/absence of an externalizing disorder or internalizing disorder. A main effect for ethnicity was found; African American youth received more externalizing diagnoses than did European American youth (odds ratio 2.01 (CI: 1.73-2.33) in one sample and 1.67 (CI: 1.44-1.94) in the other); African American youth also received fewer internalizing diagnoses than European American youth (odds ratio 0.55 (CI: .48-.63) in one sample and 0.75 (CI:.64-.88) in the other. Potential explanations for these findings include: 1. Biopsychosocial origin; 2. Clinician bias; 3. Discordant normative behavioral expectations between parents and service providers; and 4. Interaction between differential expression of underlying pathology and tolerance for such expressions.

No abstract

No abstract

It is observed that many sorts of difficulties preclude the uneventful construction of tests by a computerized algorithm, such as those currently in favor in Computerized Adaptive Testing (CAT). In this essay we discuss a number of these problems as well as some possible avenues of solution. We conclude with the development of the ‘testlet’, a bundle of items that can be arranged either hierarchically or linearly, which maintains the efficiency of an adaptive test while simultaneously keeping the quality control of test construction that is only possible currently with careful expert scrutiny.