Procedures

All children underwent two testing sessions, Session A, followed by Session B. In Session A, we administered calculation fluency in a group setting, which included a 60-s time limit for each subtest. Tests took place in a quiet room at school and were conducted by two graduate students who received extensive training on how to implement the tasks. Session B was an individual session where each child met one-on-one with the experimenter for about 60 min. The session consisted of five tests which were math problem-solving, non-verbal matrices, expressive attention, N-back, and number line estimation. The sessions were not counter-balanced or separated across children for two reasons. First, the test time was unified by the school to fit teaching schedules, and second, the students could exchange answers after class if they were assigned into groups. All test results were cross-checked for scoring accuracy, with an interrater reliability of 0.99.