Jones, E. D., Barnette, J. J., & Callahan, C. M. (1983, April). Out-of-level testing for special education students with mild learning handicaps . Annual meeting of the American Educational Research Association (AERA), Montreal, Quebec (Canada). https://eric.ed.gov/?id=ED228328
Jones, E. D., Barnette, J. J., & Callahan, C. M. (1983, April). Out-of-level testing for special education students with mild learning handicaps. Annual meeting of the American Educational Research Association (AERA), Montreal, Quebec (Canada). https://eric.ed.gov/?id=ED228328
Downloadable from ERIC database: https://files.eric.ed.gov/fulltext/ED228328.pdf
Special education students took three tests; an in-level test, a test level that was one-level below and another test that was two levels below. The order was counterbalanced with 10 days between administrations. The authors explore the comparability of scaled scores across levels, guessing and accuracy of teacher vs. locator test assignment to test levels.
58 fifth and sixth grade special education students (91.3% had learning disabilities, 1.8% were EMR and 6.9% had emotional disabilities) from 11 rural and suburban elementary and middle schools. Selection criteria was the students received both regular and special instruction in reading and achievement was estimated to be below the 20th percentile. All participants were males. 74.1% were Caucasian and 25.9% were African Americans.
Reading subtest of California Achievement Test (CAT). The CAT was chosen because publishers recommend that students may be tested more as far as two levels out. The authors felt that this additional flexibility may be important in applications to special education populations where achievement would be expected to be lower than in Title I populations (who have been the primary participants in out-of-level testing research).
Statistically significant decline on vertical scale scores. (ANOVA). With a follow up Scheffe test only significant differences between means for in-level and two levels below (they interpret this to mean that the floor effect did not begin to disappear until two levels out). Reliability: The proportions of raw scores below the chance-level were significantly lower for the OOLT for fifth graders but not sixth graders. Guessing: K sample Binomial test for equal proportions revealed statistically significant decreases in proportions of guessed items for the out-of-level test.