In the present study, equivalence‐based instruction was used to teach 2 4‐member classes representing high and low statistical variability to 10 college students. Computerized equivalence‐based instruction with multiple‐exemplar training was used to teach the classes. A pretest–training–posttest design evaluated performances on both computer‐based tests and written multiple‐choice tests. Scores improved from pretest to posttest on both the computerized and the multiple‐choice tests for all students following equivalence‐based instruction. Class‐consistent selections also generalized from training to novel stimuli and to a novel context (i.e., written test). Finally, class‐consistent performances maintained 1 week after equivalence‐based instruction was completed. The study demonstrated that equivalence‐based instruction can be used to teach labeling of statistical variability and that a selection‐based teaching protocol administered on a computer can promote the emergence of responses to a written selection‐based testing protocol.