and some states had “coding problems” (Quality Counts, p. 76). Like Thurlow, the authors note that compilation of these data was not straightforward, and they note that “differences in participation rates across states reflect, in part, the fact that states do not count students the same way when calculating such data” (p. 76). They cite the following as sources of differences:
If states did not have tests in place in the targeted grades, participation rates were based on tests for the next closest grade level.
If states did not have results for the 2002-2003 school year, participation rates were based on the most recent results available.
While all states count students with disabilities who take state tests without accommodations or with “standard” accommodations in their participation rates, only 26 states and the District of Columbia count those who take state tests with modifications.1
Fourteen of the states include students who took out-of-level tests in their participation rates.
While most states counted students who took alternate assessments in their participation rates, California and Indiana excluded them from their participation rates.
The participation rate data reported above all pertain to students with disabilities. We were unable to obtain data that would permit calculations of participation rates for English language learners.
The provision of accommodations has clearly increased the overall participation of students with special needs in NAEP, but significant variations in accommodation policies, both among the states and between states and NAEP, remain an important issue to consider in evaluating the comparability of data about students with disabilities and English language learners. Nevertheless, state assessment programs vary in the constructs they are measuring, both from one another and from NAEP, and these differences account for some of the variation in policies. To the extent that the rates are significantly different, inferences made from comparisons of results from NAEP and state-level assessments for these two groups must be limited. While other differences between NAEP and state-level assessments limit the kinds of inferences that can be made from comparisons in any case (National Research Council, 1999b), it is nevertheless true that gross differences in performance on NAEP and a state assessment are often cited as reasons to further explore the state assessment results and possible reasons for the