Printer Friendly Version (pdf, 125K)

Validating Assessments for Students With Disabilities

The Validity Argument

The validity argument states that some tests will produce scores that can be interpreted validly as measures of student achievement and used validly in a manner that is stated publicly (Kane, 1992). Making this argument involves assumptions about the causal connections between student learning and out-of-school factors, such as family background characteristics (socioeconomic status, mobility, etc.) and in-school factors, such as quality and quantity of instruction, learning environment, opportunity to learn, and school leadership. We also determine whether the student's disabilities interfere in some way with the validity of any assessment of his or her learning.

Part of the validity argument is that interpretations of test results for students with disabilities are valid if certain conditions are met satisfactorily. A student's disability should not interfere with the assessment of his or her learning. For instance, a student with mild mental retardation and low reading comprehension may have difficulty reading a mathematics test that features items from the domain of complex mathematics problem-solving tasks. Administering an unaltered mathematics problem-solving test prevents the student from performing even if he or she has the ability to solve the problems. Addressing mathematical problem-solving in a manner that is more suitable for a student with this type of cognitive functioning might require simplification of the problem or its cognitive demand. In other words, accommodations are developed to remove factors that obscure a valid assessment of each student's learning.

Abedi (2004) presents examples of linguistic modifications of math test items in which, although used with English language learners, simplified language would provide similar advantages to students with disabilities. In simplifying linguistic features, Abedi removes idioms and words that are long or unfamiliar in context. Complex sentences are simplified by removing the passive voice and subordinate, conditional, and adverbial clauses. As Abedi, Hofstetter, & Lord (2004, p. 17) note, these changes narrow the performance gap of English language learners and other students by "modifying the language of the test items to reduce the use of low-frequency vocabulary and complex language structures that are incidental to the content knowledge being assessed."

Validity Claim

An argument about the validity of an interpretation based on a test of student achievement is subject to analysis, evaluation, and approval by state policymakers. This may be a public encounter in which all constituencies take part. Participation is one contributing form of validity evidence (Kane, 2002). After all is said and done, the sponsor of the tests—the state—needs to make the claim that the use of the test scores for students with disabilities is sufficiently valid. At the end of this validation process, the evidence should support the claim. Even though the validation is intended to support the argument and claim, the state has a duty to seek out evidence that may not support either the argument or claim. By doing this, the state shows that its work was done with integrity. In addition, identifying evidence that threatens validity provides critical data for recommending changes in the testing program that can potentially eliminate or reduce these threats. Validity research is a key to uncovering these threats to validity (Haladyna, in press).