On the evils of group averaging: commentary on Nevin's "Resistance to extinction and behavioral momentum".

Except under unusually favorable circumstances, one can infer from functions obtained by averaging across the subjects neither the form of the function that describes the behavior of the individual subject nor the central tendencies of descriptive parameter values. We should restore the cumulative record to the place of honor as our means of visualizing behavioral change, and we should base our conclusions on analyses that measure where the change occurs in these response-by-response records of the behavior of individual subjects. When that is done, we may find that the extinction of responding to a continuously reinforced stimulus is faster than the extinction of responding to a partially reinforced stimulus in a within-subject design because the latter is signaled extinction.