An Investigation of the Nature of the Interaction of Reading and Computational Abilities in Solving Mathematics Word Problems

Abstract

This investigation was designed to provide evidence regarding the nature of the interaction between reading ability and computational ability when examinees respond to mathematics word problems. Scores on the Reading and Computation subtests of the Iowa Tests of Basic Skills (ITBS) for groups of fourth- and eighth-grade students were used as "pure" measures of reading ability and computational ability, respectively. These scores were used to identify three groups of students within each grade: (a) those very low in both reading and computational abilities, (b) those very high in reading ability but very low in computational ability, and (c) those very high in computational ability but very low in reading ability. The responses of each of these groups to the items on the ITBS Mathematics Problem-Solving subtest were used to describe the nature of the interaction between reading ability and computational ability. It was discovered that on each problem-solving test, examinees' item responses exhibited several different types of interaction. Some items appeared to measure primarily reading ability or primarily computational ability. Some clearly exhibited a compensatory type of interaction, whereas others exhibited a noncompensatory type of interaction. The presence of items as diverse as these would seem to have some implications relevant to applications of item response theory (IRT) with tests of this type.