Objectives: Telephone-administered cognitive assessments are a cost-effective and sometimes necessary alternative to face-to-face assessments. There is limited information in large studies concerning mode effects, or differences in cognition attributable to the assessment method, as a potential measurement threat. We evaluated mode effects on cognitive scores using a population-based sample of community-living older adults.
Methods: We used data from participants aged 65-79 in the 2014 Health and Retirement Study for whom the interview mode was randomized (n = 6,825). We assessed mode differences in test means, whether mode modifies associations of cognition with criterion variables, and formal measurement invariance testing.
Results: Relative to face-to-face assessment, telephone assessment was associated with higher scores for memory and calculation (0.06 to 0.013 standard deviations [SD]) and lower scores for nonmemory items (-0.09 to -0.01 SD). Cognition was significantly differentially related to instrumental activities of daily living difficulty depending on assessment mode. Measurement invariance testing identified evidence of mode differences in certain tests as a function of mode: adjusting for underlying cognition, the largest mode differences in memory and attention: immediate noun recall, delayed word recall, and serial-7s scores were higher given telephone administration.
Discussion: Differences by mode of administration are apparent in cognitive measurement in older adults, albeit to a small degree in our study, and most pronounced for tests of memory and attention. The importance of accounting for mode differences ultimately depends on one's research question and study sample: not all associations may be affected by mode differences, and such modification may only be apparent among those with lower cognitive functioning.
Keywords: Cognition; Mode effects; Psychometrics; Surveys; Telephone.
© The Author(s) 2022. Published by Oxford University Press on behalf of The Gerontological Society of America. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.