Using a Multinomial Model to Differentiate Alzheimer's Disease, Vascular Disease, and Elderly Controls Based on an Immediate Recall Task

Can the ability to freely recall words immediately after reading them from a list differentiate early stages of Alzheimer's Disease (AD), Vascular Dementia (VD), and healthy, elderly controls? Previous studies have suggested that immediate free recall ability does not help to differentiate early AD from controls, while performance on delayed recall can (e.g., Welsh et al., 1991; 1992). However, the methods used typically analyzed only part of the available information, namely, the mean number of items recalled per trial. General Processing Tree (GPT) modeling is a form of multinomial modeling (described in Riefer & Batchelder, 1988) that analyzes more of the available information and may help differentiate these diagnoses. The GPT method provides estimates and confidence intervals for parameters reflecting the probabilities of hypothesized, underlying cognitive processes.