Demonstrates the utility of assigning unique data sets to students. Students in this statistics course typically received unique data sets generated by a computer program, minimizing their ability to engage in unauthorized collaboration. On one of 11 assignments, however, a random group of students received common rather than the usual unique data sets. Students with shared data sets earned statistically significant lower grades on corresponding two exam questions that assessed conceptual and computational aspects of the topic. These finding suggest that at least some students with common data engaged in unauthorized collaboration and, as a consequence, learned that material less well. No significant difference was detected between students with unique versus common data on the overall exam grades or final course grades, presumably because, by design, the assessed material accounted for only a trivial amount of either grade. Consistent use of shared data sets, however, conceivably could interfere with learning.
Back to publications