catherinen.close

Title of QTUG presentation:

Measuring the Speed of Numerical Reasoning

List of Author and co-authors for QTUG presentation:

Catherine N. Close, Robert Semmes, and Mark L. Davison

Abstract Measuring the speed of numerical reasoning has seen renewed interest because there might be a need to re-examine speeded tests for some academic disciplines. As an example, using knowledge in one domain (say arithmetic) to learn a second domain (say algebra) may require substantial proficiency in arithmetic, so that the task demands of arithmetic do not interfere with learning the algebra. We tested two hypotheses: A Level-Only Hypothesis and a Speed-Level Hypothesis. A person’s level can be broadly defined as the highest level of task difficulty the person is consistently and successfully able to manage without time constraints. 187 participants from a large Midwestern University were randomly administered 2 forms of a test each comprising 52 items from the GRE and DAT (Differential Aptitude Test). The data were analyzed in two groups using the HLM6 software.The results indicated a significant speed random effect, p-value < 0.0001, representing individual differences in the speed effect. Also, the correlation between level and speed dimensions was small (0.02) supporting the conclusion that these must be two separate dimensions. Finally, reliability estimates for the level dimension were high (> .90) but speed was not very reliably estimated (.31).