pubmed.ncbi.nlm.nih.gov

Investigating the performance of exploratory graph analysis and traditional techniques to identify the number of latent factors: A simulation and tutorial - PubMed

Investigating the performance of exploratory graph analysis and traditional techniques to identify the number of latent factors: A simulation and tutorial

Hudson Golino et al. Psychol Methods. 2020 Jun.

Abstract

Exploratory graph analysis (EGA) is a new technique that was recently proposed within the framework of network psychometrics to estimate the number of factors underlying multivariate data. Unlike other methods, EGA produces a visual guide-network plot-that not only indicates the number of dimensions to retain, but also which items cluster together and their level of association. Although previous studies have found EGA to be superior to traditional methods, they are limited in the conditions considered. These issues are addressed through an extensive simulation study that incorporates a wide range of plausible structures that may be found in practice, including continuous and dichotomous data, and unidimensional and multidimensional structures. Additionally, two new EGA techniques are presented: one that extends EGA to also deal with unidimensional structures, and the other based on the triangulated maximally filtered graph approach (EGAtmfg). Both EGA techniques are compared with 5 widely used factor analytic techniques. Overall, EGA and EGAtmfg are found to perform as well as the most accurate traditional method, parallel analysis, and to produce the best large-sample properties of all the methods evaluated. To facilitate the use and application of EGA, we present a straightforward R tutorial on how to apply and interpret EGA, using scores from a well-known psychological instrument: the Marlowe-Crowne Social Desirability Scale. (PsycInfo Database Record (c) 2020 APA, all rights reserved).

PubMed Disclaimer

Figures

Figure F1.
Figure F1.

Effect Size - Multidimensional Structures

Figure 1.
Figure 1.

Simulated five factor model with loadings of .70 and 5,000 observations with interfactor correlation of .70 (top) and zero (bottom). The left side shows the population correlation matrix plotted as a network of zero-order correlations, while the left side shows the EGA estimation of the population correlation matrix. Nodes represent variables, edges represent correlations, and the node colors indicates the simulated factors.

Figure 2.
Figure 2.

New EGA algorithm for unidimensional and multidimensional structures

Figure 3.
Figure 3.

A depiction of a network tetrahedron (left) and a tetrahedron drawn so that no edges are crossing (right)

Figure 4.
Figure 4.

A depiction of how TMFG constructs a network. Starting with the tetrahedron, the node with the largest sum to three other nodes in the network is added (top left). This process continues until all nodes are included in the network.

Figure 5.
Figure 5.

Accuracy per sample size, factor loadings and number of variables (NVAR) for unidimensional factors with continuous (A) and dichotomous (B) data.

Figure 6.
Figure 6.

Mean Absolute Error (MAE) per sample size, factor loadings and number of variables (NVAR) for unidimensional factors with continuous (A) and dichotomous (B) data.

Figure 7.
Figure 7.

Accuracy per sample size, factor loadings and number of variables (NVAR) for multidimensional factors with continuous (A) and dichotomous (B) data.

Figure 8.
Figure 8.

Boxplot comparing the percentage of correct estimates between EGA, PApaf and PApca in multidimensional structures with dichotomous data by interfactor correlation and factorloadings.

Figure 9.
Figure 9.

Mean Absolute Error (MAE) per sample size, factor loadings and interfactor correlation for unidimensional factors with continuous (A) and dichotomous (B) data.

Figure 10.
Figure 10.

EGA dimesional structure of the Social Desirability Scale.

Similar articles

Cited by

References

    1. Anderson H, T.W. & Rubin. (1958). Statistical inference in factor analysis. In Proceedings of the 3rd berkeley symposium on mathematics, statistics, and probability (Vol. 5, pp. 111–150).
    1. Auerswald M, & Moshagen M (2019). How to determine the number of factors to retain in exploratory factor analysis: A comparison of extraction methods under realistic conditions. Psychological Methods, 24, 468–491. 10.1037/met0000200 - DOI - PubMed
    1. Aust F, & Barth M (2018). papaja: Create APA manuscripts with R Markdown. Retrieved from https://github.com/crsh/papaja
    1. Barfuss W, Massara GP, Di Matteo T, & Aste T (2016). Parsimonious modeling with information filtering networks. Physical Review E, 94(6), 062306. - PubMed
    1. Beierl B, E. T. (2018). Is that measure really one-dimensional? Nuisance parameters can mask severe model misspecification when assessing factorial validity. Methodology, 14(4), 188–196.

MeSH terms