The disease burden of depression among older populations is high. Detecting changes in late-life depression is predicated on the seldom-examined assumption of longitudinal measurement invariance (MI). Therefore, we investigated longitudinal MI of the 8-item Center for Epidemiological Studies Depression Scale in core members repeatedly assessed in the English Longitudinal Study of Aging, a nine-wave representative study of the English population above 50 years of age (initial N = 11,391). Based on prior literature, we tested MI of a one-factor solution, a one-factor solution with correlated errors of reversely coded items, and a two-factor solution (depressed affect/somatic complaints). For all factor solutions, residual MI was confirmed across nine waves and gender. Sum score models (i.e., all factor loadings constrained to equity) had a good fit. Depression scores correlated with psychiatric diagnoses, ill health, lower life quality, and female gender. Associations slightly differed depending on the factor solutions, signifying their applicability across contexts.
Keywords: Center for Epidemiological Studies Depression Scale; depression; longitudinal measurement invariance; older-aged populations.