Content area
Abstract
The purpose of this study was to examine the extent of empirical support for several commonly-held beliefs and disputes about testing accommodations for students with disabilities. Testing accommodations are changes in standardized test procedures that are intended to facilitate the measurement of targeted skills and knowledge. Although research on testing accommodation effects for students with disabilities is accumulating, questions remain about which accommodations should be considered appropriate. Decisions about which accommodations should be made available to students are therefore often based on beliefs and opinions, and have been shown to vary across states. In order to examine the extent of support for these beliefs, as well as identify the degree of support for several widely disputed accommodations, existing student performance data from three statewide achievement tests and standardization sample data from a published test were obtained and analyzed. This study focused on the following beliefs: (a) the appropriateness of accommodations changes as a function of the student's disabling condition (e.g., accommodating sensory/physical disabilities vs. cognitive disabilities), (b) the appropriateness of several individual accommodations depends on the content area and skills tested (e.g., reading aloud a math test vs. reading aloud a reading test), and (c) accommodations fall on a continuum in terms of how they affect test validity (e.g., some accommodations are highly appropriate, some accommodations are moderately appropriate, and some accommodations are highly inappropriate). Measurement characteristics (item-total correlations, coefficient alphas, and IRT item parameters) for a variety of accommodation groups were compared to a reference group of non-accommodated students without disabilities. Substantial differences in item-level characteristics were considered indicative of poorer measurement for students in the corresponding accommodation group. The third belief was the only belief for which empirical support was evident across several datasets. Even so, results for specific accommodation groups did not always follow the patterns expected (e.g., some accommodations that are commonly considered highly appropriate exhibited substantial differences in measurement characteristics). Empirical support for the first and second beliefs was limited to one or two datasets. Limitations of the study are discussed, and direction for future research in the area of inclusive assessment systems is provided.
