Reducing Proportion of Chance Scores in Inner-City Standardized Testing Results: Impact on Average Scores

Abstract

The article by Doscher and Bruno, Simulation of Inner-City Standardized Testing Behavior: Implications for Instructional Evaluation (1981), indicates, through a simulation study, that reported test scores will be an overstatement of actual performance when a large proportion of scores are included. This is precisely the result obtained when, between the 1974 and 1975 testing periods, the public schools in Chicago switched from grade-level testing to functioning-level testing. In 1974, and before, students were assigned a test according to their grade level. The test used was the Iowa Tests of Basic Skills (ITBS) (Hieronymous & Lindquist, 1971). That is, every grade 8 student took Level 14 of the ITBS, every grade 7 student Level 13, and so forth. Under functioning-level testing, the student was assigned a test the difficulty level of which, in the teacher's opinion, best matched the student's current functioning reading level. The reason for the change was the observation in 1974 that fully 42 percent of the students taking the ITBS scored at the equivalent of a chance level-25 percent or less in terms of raw score. As Doscher and Bruno correctly assert, the proportion is highest at low performance levels. The proportion ranged from 5 percent in the high-performing schools to 85 percent or more in the schools at the bottom of the distribution. Table I shows the impact on the citywide average test scores. The table indicates that the change caused lower scores, and that this negative impact increased with grade. The impact on average scores was negligible for schools which in 1974 had scored fairly close to, or above, the national averages. The lowering of the average citywide score at grade 8 was about .4 standard deviation.