Recent studies suggest that older adults' simple arithmetic fact knowl
edge may be superior to that of college students, as evidenced by the
older adults' more frequent use of direct memory retrieval (versus com
putation) as an answer generation mechanism. Whereas previous studies
assessed strategy selection via self-report and/or reaction time, we h
ave adopted the ''generation effect'' paradigm-better memory for items
that are subject-generated versus those that are simply read. The mem
orial advantage of generation depends in part on the degree of effort
involved in generating versus reading an item. Because direct retrieva
l is less effortful than computation, we expected qualitative age diff
erences in answer generation strategies to manifest themselves as age
differences in the magnitude of the generation effect, especially for
problems with larger answers. With simple multiplication problem mater
ials, the expected Age X Problem Size interaction was found. In a verb
al materials comparison condition, the size of the generation effect d
id not differ across adult age.