Application of Latent Variable Methods to the Study of Cognitive Decline When Tests Change over Time

Abstract

Background The way a construct is measured can differ across cohort study visits, complicating longitudinal comparisons. We demonstrated the use of factor analysis to link differing cognitive test batteries over visits to common metrics representing general cognitive performance, memory, executive functioning, and language. Methods We used data from three visits (over 26 years) of the Atherosclerosis Risk in Communities Neurocognitive Study (ARIC-NCS) (N=14,252). We allowed individual tests to contribute information differentially by race, an important factor to consider in cognitive aging. Using generalized estimating equations, we compared associations of diabetes with cognitive change using general and domain-specific factor scores vs. averages of equally weighted standardized test scores. Results Factor scores provided stronger associations with diabetes at the expense of greater variability around estimates (e.g., for general cognitive performance, −0.064 SD units/year, SE=0.015, vs −0.041 SD units/year, SE=0.014), which is consistent with the notion that factor scores more explicitly address error in measuring assessed traits than averages of standardized tests. Conclusions Factor analysis facilitates use of all available data when measures change over time, and further, it allows objective evaluation and correction for differential item functioning.

Cognitive Reserve in Midlife is not Associated with Amyloid-β Deposition in Late-Life

“…Three of these tests were also administered at visit 2. Latent variable methods were used to summarize these tests into a factor score representing general cognitive performance[8]. Scores were standardized such that a factor score of 1 represents general cognition that is 1 standard deviation above the mean.…”

Olfactory function and neurocognitive outcomes in old age: The Atherosclerosis Risk in Communities Neurocognitive Study

“…Using data from these tests in a factor analysis, factor scores for general cognitive performance, executive functioning/processing speed, memory and language were derived. 22 Briefly, factor analysis is a structured approach for identifying common covariation between specific indicators, in this case the cognitive tests, to reduce measurement error when combining data across multiple cognitive tests. The interpretations of factor scores are similar to that for z scores because they were scaled to have a mean of 0 and variance of 1 at ARIC visit 2 when the participant’s cognitive function was first tested.…”

A Latent Variable Modelling Approach for The Pooled Analysis of Individual Participant Data on The Association between Depression and Chlamydia Infection in Adolescence and Young Adulthood in the UK

“…Subsequently, by fitting an IRT‐graded response model to a calibration sample we allowed the relevant MFQ and harmonized PHQ‐2 items to be differently related to the underlying construct of depression—including DIF by study on one item. This is a substantial improvement over other potential methods that are used for scoring such as proportion, sum or z ‐score (Curran et al ., ; Gorter et al ., ; Griffith et al ., ; Gross et al ., ). Next, IRT longitudinal equating through incorporation of known item parameters for the harmonization and scoring of longitudinal and cross‐sectional ordinal data together with latent growth models, such as a piecewise linear model, were demonstrated that they can be implemented simultaneously.…”