Math Ability

Participants completed difficult mental arithmetic problems adapted from the Kit of Factor-Referenced Cognitive Tests^50,51. Trials included all four basic arithmetic operations: addition (three 2-digit numbers; e.g., 45 + 72 + 87), subtraction (a 2-digit or 3-digit minuend and a 2-digit or 3-digit subtrahend; e.g., 354–87), multiplication (one 2-digit number and one 1-digit number; e.g., 64 × 6), and division (a 1-digit divisor into a 2-digit or 3-digit dividend; e.g., 432 ÷ 9). Problems were open-ended (i.e., not verification); hence, participants responded by typing their answers using the number pad on the keyboard. They were required to calculate the answer mentally—that is, pencil and paper or other devices were not permitted to aid with calculation. As such, the task was relatively difficult for arithmetic (mean accuracy = 81.2%, mean RT = 9.91 s). Operation types were presented in separate blocks, and in each block, participants completed as many problems as they could in 3 min. Participants were not aware that there was a time limit, and the block ended once a participant completed the trial they were at once 3 min had passed (this final trial was omitted from analysis). A math ability score was computed for each participant by summing the total number of problems answered correctly across all four operation types, where higher scores indicate greater math ability. Past work has shown that performance on this task is correlated with performance on several basic numerical tasks (including numerical ordering and numerical comparison tasks⁵¹). Internal reliability for this task was computed using participants’ scores for each of the four operation types; Cronbach’s α was .89.

While we acknowledge that mathematics as a whole comprises far more than even the most difficult mental arithmetic task, here we chose a challenging arithmetic task as a measure of objective math ability for several reasons. First, this ensured that all participants would have the requisite knowledge to complete the chosen task (e.g., it is not a given that all students would have taken the necessary course for more specialized math topics, such as geometry or calculus). Second, not all STEM courses will require the same type of specialized math or even specialized math at all. That is, arithmetic is likely to be one of the more universally tapped math skills across a wide range of STEM courses. Finally, prior empirical work has shown that arithmetic is a reliable predictor of more advanced math skills^45,46.