Confirmatory factor analysis with missing data in a small sample: cognitive reserve in people with Down Syndrome

Cristina Cañete-Massé*, Maria Carbó-Carreté, María Dolores Figueroa-Jiménez, Guillermo R. Oviedo, Myriam Guerra-Balic, Casimiro Javierre, Maribel Peró-Cebollero, Joan Guàrdia-Olmos

*Corresponding author for this work

Research output: Indexed journal article Articlepeer-review

Abstract

The presence of missing data and small sample sizes are very common in social and health sciences. Concurrently to present a methodology to solve the small sample size and missing data, we aim to present a definition of Cognitive Reserve for people with Down Syndrome. This population has become an appealing focus to study this concept because of the high incidence of dementia. The accidental sample comprised 35 persons with DS (16–35 years). A total of 12 variables were acquired, four of them had missing data. Two types of multiple imputation were made. Confirmatory factor analysis with Bayesian estimations was performed on the final database with non-informative priors. However, to solve the sample size problem, two additional corrections were made: first, we followed the Jiang and Yuan (2017) schema, and second, we made a Jackknife correlation correction. The estimations of the confirmatory factor analysis, as well as the global fit, are adequate. As an applied perspective, the acceptable fit of our model suggests the possibility of operationalizing the latent factor Cognitive Reserve in a simple way to measure it in the Down Syndrome population.

Original languageEnglish
Pages (from-to)3363-3377
Number of pages15
JournalQuality and Quantity
Volume56
Issue number5
DOIs
Publication statusPublished - Oct 2022

Keywords

  • Bayesian structural equation models
  • Cognitive reserve
  • Down syndrome
  • Missing data
  • Small sample

Fingerprint

Dive into the research topics of 'Confirmatory factor analysis with missing data in a small sample: cognitive reserve in people with Down Syndrome'. Together they form a unique fingerprint.

Cite this