Education Bias in the Mini-Mental State Examination

Abstract

Education is correlated with cognitive status assessment. Concern for test bias has led to questions of equivalent construct validity across education groups. Following the work of previous researchers, we submitted Mini-Mental State Examination (MMSE) responses to external validation analyses. Subjects were older participants in the Epidemiologic Catchment Area study (age 50-98). Little evidence for test bias against those with low education was found. The correlation of MMSE scores and age was equivalent across high-and low-education groups (−.29 vs. −.27, p = .48), as was the correlation of MMSE scores and activities of daily living (ADL) functioning (−.23 vs. −.27, p = .42). The MMSE displayed significantly higher internal consistency reliability in the low-education group (.75 vs .72, p = .04). The MMSE did not predict functional decline over 1 year or mortality over 13 years differently by level of educational attainment. Evidence for sex bias was found. The MMSE was more highly correlated with age among women than among men (−.28 vs. −.21, p < .001). The MMSE was more highly correlated with ADL impairment among women than among men (−.30 vs. −.17, p = .01). The MMSE predicted mortality differently according to participant sex (p = 0.53). The lack of evidence for bias provides little support to proposals to adjust MMSE scores according to level of education.