Effects of age, education, and sex on response bias in a recognition task.

This study examined age-related differences in decision criteria and the extent to which inconsistencies in earlier findings could be due to sampling artifacts, especially the underlying effects of educational level and sex. Male and female participants (N = 3,059) from 4 age groups (32, 42, 52, and 62 years) and a wide range of educational levels performed a word recognition task. Response bias was assessed with a nonparametric index derived from signal detection theory. The analyses revealed no age differences except for the most educated subjects, for whom increased age was associated with stricter decision criteria. Lower levels of education and men as compared with women were associated with a more conservative bias. Controlling for the level of sensitivity did not significantly change this pattern of results. This finding stresses the need for caution in generalizing age differences obtained from samples that are only partly representative or imbalanced with respect to education and sex.