Abstract
Laboratory paradigms have provided an empirical foundation for much of psychological science. Some have argued, however, that such paradigms are highly susceptible to idiosyncratic strategies and that many findings do not reflect fundamental cognitive principles but are instead artifacts of averaging across participants who employ different strategies. We developed a set of techniques to rigorously test the extent to which average data are distorted by such strategy differences and applied these techniques to free recall data from the Penn Electrophysiology of Encoding and Retrieval Study. Recall initiation showed evidence of subgroups: The majority of participants initiated recall from the last item in the list, but one subgroup showed elevated initiation probabilities for items 2 to 4 positions back from the end of the list, and another showed elevated probabilities for the beginning of the list. By contrast, serial position curves and temporal and semantic clustering functions were remarkably consistent, with almost every participant exhibiting a recognizable version of the average function, suggesting that these functions reflect fundamental principles of the memory system. The approach taken here can serve as a model for evaluating the extent to which other laboratory paradigms are influenced by individual differences in strategy use.