The relationship between cognitive abilities and depression in the context of old age: Testing for measurement invariance using local structural equation modeling

Recent work revealed that measures of mental health might lack measurement invariance across the levels of cognitive abilities. This finding raises concerns for studies on both cognition and mental health, as measurement invariance is a fundamental requirement for meaningful comparisons of variables across a given characteristic. The aim of this study was to test the measurement invariance of the Center for Epidemiological Studies – Depression (CES-D) scale across four cognitive abilities: crystallized knowledge, quantitative reasoning, verbal fluency, and episodic memory. Data from the Health and Retirement Study (HRS) were used. The sample analyzed consisted of 19,475 participants aged 50 years or older. The results showed that CES-D lacked measurement invariance across all four cognitive abilities measured. A prominent pattern emerged, with unstandardized item thresholds increasing as cognitive abilities increased. The CES-D scores of individuals at different levels of cognitive abilities are not comparable, which may lead not only to incorrect inferences about the relationship between cognitive abilities and CES-D scores but also affects any other research using CES-D where the impact of cognitive abilities on the scales' observed scores is uncontrolled for. The increase in thresholds might suggest that individuals with higher cognitive abilities are more likely to mask their symptoms.