Computer-based testing is an emerging method to evaluate students' mathematics learning outcomes. However, algebra problems impose a high cognitive load due to requiring multiple calculation steps, which might reduce students' performance in computer-based testing. In order to understand students' cognitive load when answering algebra questions in a computer-based testing environment, three perspectives, element interactivity, practice effect, and individual differences, were investigated in this study. Seven levels of algebra exam questions were created using unary and simultaneous linear equations, and the inverse efficiency scores were employed as a measure of cognitive load in the study. Forty undergraduate and graduate students were tested. There were four findings: (1) As the element interactivity of test materials increased, the cognitive load increased rapidly. (2) The high-efficiency group had a lower cognitive load than the low-efficiency group, suggesting that the high-efficiency group had an advantage in a computer-based testing environment. (3) "Practice" has a considerable effect on reducing cognitive load, particularly in level 6 and 7 test items. (4) The low-efficiency group can reduce but not eliminate the gap with the high-efficiency group; they may require additional experience in a computer-based testing environment in order to improve reducing their cognitive load.
Keywords: cognitive load; computer-based testing; element interactivity; individual differences; mathematics algebra questions; practice effect.