Consider the characteristics of a prototypical study in the area of cognitive aging: 20 young adults recruited from a pool of college students are contrasted with 20 older adults with a mean age of between 60 and 75 years, often recruited from a pool of individuals who have participated in prior studies; a single task is administered with two (or more) condi-tions assumed to differ with respect to one or more theoretically interest-ing components; a single measure of performance is obtained in each condition, usually with no indication of its reliability; and the results are analyzed with an analysis of variance (ANOV A), with the greatest interest in the Age x Condition interaction. If the interaction is statistically sig-nificant, then the researcher usually infers that the age effects are espe-cially pronounced on the theoretical component corresponding to the variable with the largest age difference. Frequently the researcher con-cludes that there are no age differences in a component in which the age difference in the relevant measure was not statistically significant (i.e., the null hypothesis is accepted). Each aspect of the design, analysis, and interpretation in this proto-typical approach involves numerous important, but often unrecognized, assumptions. The primary goal of this chapter is to examine assumptions implicit in current research practices in cognitive aging. In the process, possible consequences of violations of the assumptions will be identified, as well as ways in which the validity of the assumptions can be investi-gated.
Stored only when both papers are in the corpus — 24 citations exist upstream.