Several experiments were conducted to check the reliability and validity of the CORMAS scoring procedure.
Scores yielded by different variants of the CORMAS scoring procedure (different methods of choosing examples and counterexamples) were correlated with each other, and with the scores yielded by the two other scoring methods.
Finally, as an additional indication of construct validity, the CORMAS scores were found to differentiate between different memory conditions, yielding for example lower scores for reports provided after a longer delay between exposure and testing, and lower scores for the reports of subjects exposed to "contaminating" post-event misinformation relative to control subjects who were not exposed.
The main conclusions from the CORMAS strand of the project can be summarized as follows:
The CORMAS procedure yields reliable and valid memory scores which capture both the overall amount of correct information (number of correct statements) scores and a more global evaluation of the overall correspondence between the contents of a free-narrative memory report, and the original events and details that actually occurred.
The CORMAS method is standardized, in the sense that it involves a minimal amount of human intervention (in the production of examples and counterexamples), and is quite robust across minor variations in the nature of this intervention.
CORMAS provides a new and convenient research tool that can be used to evaluate the overall quality of free-narrative memory reports in a standardized and reliable way.
We found that several of the QAP measures that were based on a particular participant's memory and metamemory performance on a word-list study task and on a task involving memory of a different crime film could in fact be used to predict the quality of that same participant's free-narrative account of the criterion crime episode, indexed by the CORMAS score assigned to that text.
For instance, the CORMAS software described in this paper was used to fill in a gap in the existing QAP methodology for measuring memory accuracy, namely the evaluation of the overall correspondence between verbal free-narrative reports of witnesses and the actual events that are being described in these reports.