Test‐Retest Reliability of Self‐Reported Disability Measures in Older Adults

S. Crawford,A. Jette,S. Tennstedt

DOI: https://doi.org/10.1111/j.1532-5415.1997.tb00950.x

1997-03-01

Journal of the American Geriatrics Society

Abstract:he assessment of disability has become an essential part T of health studies among older adults. In most epidemiologic surveys, the assessment of disability is usually approached by using self-report measures of activities of daily living (ADLs).’” A growing number of papers have reported disability incidence rates, rates of change in disability, and predictors of change all based on repeated application of self-report disability mea~ures.~-’’ Few studies, however, have paid sufficient attention to psychometric issues such as source and extent of measurement error in repeated administration of self-report disability measures.’ The situation is compounded by the frequent use of a mixed-mode field approach in gerontological research that combines two or more modes of data collection, i.e., mail, telephone, and face-to-face inter~iewing.”-’~ Mixed-mode health surveys have become particularly attractive for studies of older populations because they increase response and retention rates and because the literature suggests that the quality of resulting responses does not differ in any substantial way across the two We are unaware, however, of any research that has examined the direction and magnitude of test-retest measurement error when a mixedmode approach is used to assess self-reported physical disability. This study examined agreement regarding ADL dilliculty and dependence between two self-reports, a telephone interview, and a subsequent home visit for 13 separate ADLs and three summary measures. We assessed the variation in the level of agreement across different subgroups defined by ssubject characteristics, including ethnicity and gender. The extent of a trend in the direction of disagreements was also determined.