Subjects recalled cmly an average of 42.5% of the nonreferent words compared tn 56.D1% of the self-referent terms. A 2 X 2 (Sex X Type of Adjective [self-referent vs. nonreferent]) mixed analysis of variance yielded only a main effect for type of adjective. F ( I , 56) = 13.82. p < .DOI.
DiscussionThe procedures used in this demonstration, although rudimentary, me sufficiently sensitive to document the selfreference effect: Subjects' memory for self-referent items was superit~r to their memory for items that were ntit selfreferent. T h e procedure is also a practical one. Recause it does not require individual testing sessions or reaction time assessment, it can he used during class with a large proup of students.The demonstration also 6aciilitates the analysis of several methodological and theoretical issues concerning schematic processing of infnrmation. Initially, students maintained that their incidental recall scores were shaped primarily hy the vividness of the trnir terms. Atrenti<~n-gettina trait terms, they contended, were hetrer reniemhered than Inore pallid terms. The exercise, however, convincingly demonstrated to them the impact of their self-schemas hecause more memorable words were als~l morc self-descriptive words. The discussion also proved usefill in illusrratina remarch design and data analysis. Some students, for example, failed to correct fbr initial frequencies of self-descriptive terms when they first explored the effect. Students uften recalled more self-referent words than nonreferent words, hut rhis difference cannot he interpreted until scores are adjusted to reflect the numher of items initially selected as self-referent.The demonstration also facilitated the analysis ofdepth-(rf-processing and schema-hased memory models of social cognition (Klein. Loftus, & Burton, 1989). First, depth-ofprocessing theory maintains that self-referent ininrmation is processed at a deeper level than nonreferenr infcirmation. If, fnr example, the students were asked "Ihes the word have more than two syllahles?." they could respond without processing the word very deeply. Such shallow processing would not lead to particularly durahle memories. In contmst, selfreferent encoding requires much deeper processing (Klein & Kihlstrom, 1986). Second, schema theories suggest that the more elahorate the schema that will hold the inconling infimnation, the hetter our ability to recall that inh~rmntion. Self-schemas may he the most complex and intricate associative networks in our memory system, so selferefrrent information is particularly rnemorahle. k r t h of these theories could hedemonstrated in theclassroom by varying the initial question posed to si~hjects. Although some students could answer the question "Does the word descrihe yc~u?," orhers could he asked "Does the wordhave more than twosyllahles" or "Lhes the word descrihe your psychology reacher?" (Rellezra, 1984).Depending on interest, the exercise could also he used to exnlore rhe coenitive cnnseauences of eender identirv. Re-