ABSTRACT
Evaluating metamnemonic accuracy is an inherently difficult enterprise as the theorist must contend with all of the usual variability inherent to normal memory behavior and additionally consider other sources that are relevant only to the metamnemonic aspects of the task. This chapter reviews the arguments motivating the use of the Goodman-Kruskal gamma coefficient γ in assessing metamnemonic accuracy and pits that statistic against a distance-based metric da derived from signal detection theory (Green & Swets, 1966). We evaluate the question of which potential measures of metamnemonic accuracy have the most desirable measurement characteristics and which measures support the types of inference that researchers commonly wish to draw. In doing so, we attempt to make general arguments without providing a detailed account of the underlying mathematics or statistics, but we do place appropriate references should those interested desire a more technical treatment of the issues that arise.
