Test-retest reliability, practice effects and reliable change indices for the recognition memory test.

Abstract

The Recognition Memory Test (RMT) is widely used; however, important characteristics such as reliability and stability over time were largely unknown. In this study, we document for the first time test-retest reliabilities, practice effects, and Reliable Change (RC) indices for this test. A sample of 206 normal adults (aged 40-70) were administered twice either the same version or two different versions of the RMT, with a 1-month interval between assessments. The normal sample was split into two groups; a young (aged 40-54) and an old (55-70) group. Test-retest reliabilities were modest when using either the same versions or different versions. Practice effects were abolished when different versions of the RMT were used. In contrast, practice effects were clearly present on the same version of the non-verbal subtest for both control groups. However, practice effects were present on the same versions of the verbal subtest only in the old group. RC indices were rather large when using the same or different versions. Although modest, the test-retest reliability of the RMT is no worse than those reliabilities reported for other commonly used recall memory tests. Thus, the inherent clinical advantages of using a recognition paradigm make its use desirable. Usage of different versions of the RMT enables us to avoid practice effects. However, the RC indices indicate that large changes in scores are needed to detect a significant improvement or decline in an individual's performance.