Comparing cognitive representations of test developers and students on a mathematics test with Bloom's taxonomy

An examination was conducted to determine whether the Taxonomy of Educational Objectives: Cognitive Domain (Bloom, Englehart, Furst, Hill, & Krathwohl, 1956) provided an accurate model to guide item writers for anticipating the cognitive processes used by students on a large-scale achievement test in mathematics. Thirty Grade 7 students were asked to think aloud as they solved problems on a mathematics achievement test. Students' cognitive processes were classified with a coding system based on Bloom's taxonomy, The overall match between the responses expected by the item writers and the responses observed from the students was 53.7%. The match score between the expected and the observed responses differed for the high and low mathematics achievers and also differed across the 2 content areas measured on the test. Agreements between the expected and the observed responses were further assessed by comparing loglinear models. The most parsimonious model contained an achievement group, cognitive level, and content area main effect, and, most important, a cognitive level by content area interaction. This finding indicated that the 2 dimensions assumed to be independent in the table of specifications, cognitive level and content area, were, in fact, dependent. The results of this study suggest that Bloom's taxonomy does not provide an accurate model for guiding item writers to anticipate the cognitive processes used by students. Implications for test design are discussed.