The reliability of ERP measures was investigated in a sample covering the adult life span (n = 59, age 21-92). This sample was divided into a young and an old group. ERPs to an auditory two-stimuli oddball task were recorded in the sample at two occasions separated by 12-14 months (T1 and T2). The recordings of T1 were split in half, to assess within session reliability. Correlations were calculated for N1, P2 and P3 peak latency and amplitude, for average amplitude during 50 ms epochs within the defined P3-window 250-550, and for average amplitude in successive 15 ms epochs from 1 to 705 ms. The results show that amplitude measures were more reliable than the latency measures at all electrodes. Time window/epoch amplitude measures yielded reliabilities in the same range as peak amplitude. Reliabilities peaked around the conventionally studied N1, P2 and P3, and this is seen as a validation of the components. In general, the old group exhibited weaker P3 peak latency reliabilities than the young group. However, many of these differences did not reach statistical significance. Implications of the findings are discussed.
Copyright 2002 Elsevier Science B.V.