Developing Multidimensional Likert Scales Using Item Factor Analysis

Abstract

This study compares the performance of two approaches in analysing fourpoint Likert rating scales with a factorial model: the classical factor analysis (FA) and the item factor analysis (IFA). For FA, maximum likelihood and weighted least squares estimations using Pearson correlation matrices among items are compared. For IFA, diagonally weighted least squares and unweighted least squares estimations using items polychoric correlation matrices are compared. Two hundred and ten conditions were simulated in a Monte Carlo study considering: one to three factor structures (either, independent and correlated in two levels), medium or low quality of items, three different levels of item asymmetry and five sample sizes. Results showed that IFA procedures achieve equivalent and accurate parameter estimates; in contrast, FA procedures yielded biased parameter estimates. Therefore, we do not recommend classical FA under the conditions considered. Minimum requirements for achieving accurate results using IFA procedures are discussed.

18REST: a short RIASEC-interest measure for large-scale educational and vocational assessment

“…Decision-making as to factor retention was guided by parallel analysis with data permutation and the Hull method (Lorenzo-Seva, Timmerman, and Kiers, 2011). Given the ordered categorical nature of variables, exploratory factor model parameters were estimated using robust weighted least squares mean − and variance−adjusted (WLSMV), as this has been recommended in recent simulation studies for the analysis of Likert−type data (Asún, Rdz-Navarro, and Alvarado, 2015). Information functions of each resulting scale were inspected to examine the achieved latent variable coverage.…”

Evidências de Validade da Versão Adaptada para o Português do Questionário TPACK Survey for Meaningful Learning

“…O uso de máxima verossimilhança pode subestimar as cargas fatoriais e as correlações entre os fatores durante análises confirmatórias, sobretudo em estudos cujas variáveis são categóricas ordenadas de natureza discreta em decorrência do uso de escalas de Likert ( Hauck-Filho, 2016). Estudos recentes indicam que estimadores, como quadrados mínimos não ponderados e quadrados mínimos ponderados robustos, podem ser mais robustos do que a máxima verossimilhança (Asún, Rdz-Navarro, & Alvarado, 2016). No entanto, no presente estudo, o uso da máxima verossimilhança resultou em cargas fatoriais superiores a 0,5 para todos os itens.…”

Validación de la escala de creencias de eficacia en la enseñanza de la matemática (ECEEM) y caracterización de las creencias de estudiantes de pedagogía básica

“…La expresión visual del modelo se puede observar en la Figura 1. El AFC se realizó mediante el procedimiento de estimación de mínimos cuadrados no ponderados (ULS) sobre las correlaciones tetracóricas entre los ítems, pues se ha demostrado que ese procedimiento de estimación (Forero, Maydeu-Olivares & GallardoPujol, 2009) y ese tipo de correlaciones (Flora & Curran, 2004; Holgado-Tello, Chacón-Moscoso, Barbero-García & Vila-Abad, 2010) permiten obtener resultados óptimos cuando se trabaja con variables ordinales, al reconocer la naturaleza ordinal de las variables observadas (Asún, Navarro & Alvarado, 2015). Esta versión del AFC es llamada también Análisis Factorial de Ítems (AFI) por algunos autores (Wirth & Edwards, 2007).…”