(e.g., large print, braille, response format) than those involving changes in administrative procedures (e.g., extra time, individual administration). Jayanthi and colleagues (1996) reported that most of the 401 general educators they surveyed believed it was unfair to provide testing adaptations only for students with identified disabilities, and that the adaptations teachers considered most helpful (simplified wording for items, individual help with directions) were not easy to make. Schumm and Vaughn (1991) also found that teachers believed all types of accommodations were more desirable than feasible. In that study, teachers favored accommodations that addressed motivational issues over those requiring curricular or environmental adaptations.
Clearly, more research on the validity of scores from accommodated testing is needed—in particular, research tailored directly to the particular assessments and inferences central to high-stakes decision making. In the interim, the existing research, although limited and based largely on different populations and types of assessments, suggests caution. The effects of accommodation cannot be assumed; they may be quite different from what an a priori logical analysis would predict. In addition, policies on the need to "flag" or mark score reports when accommodations are provided raise vexing legal, policy, and ethical questions (see Box 8-2).
The committee that wrote Educating One and All made the following broad recommendation to guide policymakers attempting to include more students with disabilities in standards-based assessment programs (p. 204):
Assessment accommodations should be provided, but they should be used only to offset the impact of disabilities unrelated to the knowledge and skills being measured. They should be justified on a case-by-case basis, but individual decisions should be guided by a uniform set of criteria.
A number of research studies are under way, including efforts to include more students with disabilities in the National Assessment of Educational Progress.4 A recent report from the National Center for Education Statistics describes many of these efforts (Olson and Goldstein, 1997); they are summarized in the appendix to this chapter.
Another problem raised in Educating One and All, which may make scores on large-scale assessments hard to interpret for some students with