Evaluating the Use of Exploratory Factor Analysis in Developmental Disability Psychological Research

Abstract

Exploratory factor analysis (EFA) is a widely used but poorly understood statistical procedure. This paper described EFA and its methodological variations. Then, key methodological variations were used to evaluate EFA usage over a 10-year period in five leading developmental disabilities journals. Sixty-six studies were located and evaluated on multiple procedural variations. Only 35% (n = 23) of studies used EFA; principal components analysis was the model used most often (n = 40, 61%). Orthogonal rotation was used most often (n = 39, 59%). A large portion of studies ran analyses with a subject: item ratio larger than 5:1 (n = 49, 74%). Most researchers employed multiple criteria for retaining factors (n = 45, 68%). Overall, results indicated that published recommendations and guidelines for the use of EFA are largely ignored.

Validade e confiabilidade do índice de capacidade para o trabalho (ICT) em trabalhadores de enfermagem

“…Recomendações quanto às suposições de cada método e aplicabilidade na psicometria têm sido feitas 19,21 .…”

“…Esse mé-todo é mais apropriado para explorar dimensões latentes com dados que não apresentam distribuição normal, sendo indicado por diferenciar a parte da variância comum da específica, ao contrário de componentes principais que não fazem esta distinção 19,20,21 . Estipulou-se na interpretação do fator carga fatorial mínima de ±0,30 20 .…”

“…Também foram calculadas as comunalidades referentes a cada item definidas como a parcela da variância do item explicada pelos fatores retidos na análise. Valores maiores ou iguais a 0,40 são considerados satisfatórios 21 .…”

Portuguese validation of the Symptom Inventory of the M.D. Anderson Cancer Center

“…They are defined as the portion of item variance explained by the factors retained in the analysis. Values greater than or equal to 0.40 are considered satisfactory (19) . The scale accuracy was calculated by means of the Cronbach's alpha coefficient.…”

Problem behaviours and symptom dimensions of psychiatric disorders in adults with intellectual disabilities: An exploratory and confirmatory factor analysis

“…Although this statistical modelling approach using factor analysis improves on the tautological problems in the studies using a two group, depression/non-depression method (Charlot et al, 1993;Felce et al, 2009;Holden & Gitlesen, 2003;Hurley, 2008;Kishore et al, 2005;Marston et al, 1997;Moss et al, 2000;Tsiouris et al, 2011) both studies used Pearson correlations for the initial matrix, and principal components analysis, which are unsuited to the analysis of categorical data (Norris & Lecavalier, 2010) and have shown to produce unstable solutions in Monte Carlo simulation studies (Snook & Gorusch, 1989). Other limitations of these studies were the small sample size (n=92) in one of the studies (Tsiouris et al, 2003) and neither study validated the EFA findings, for example using confirmatory factor analysis (CFA).…”