[Test-retest reliability of category-bound word production in psychogeriatrics]

Fifty patients with a dementia syndrome were tested and retested with a category fluency task. On each occasion subjects retrieved items from two categories: articles of clothing and fruits. The test-retest interval was two or three weeks. Test-retest correlation coefficients for the two categories were .84 and .83 respectively. The standard error of measurement was computed for investigating the statistical significance of score changes for individual subjects. The quantitative stability of successive retrieval from semantic memory contrasted with qualitative variability. Many category examples (57% of articles of clothing and 43% of fruits) retrieved during the second testing had not been retrieved during the first testing. This is consistent with a hypothesis of access deficits during retrieval from an otherwise intact knowledge base.