Unidimensional factor models imply weaker partial correlations than zero-order correlations

DOI10.1007/S11336-018-9607-ZMaRDI QIDQ725296zbMATH OpenOpenAlexWikidataFDO

Authors Riet van Bork, Raoul P. P. P. Grasman, Lourens J. Waldorp

Publication date 1 August 2018

Published in Psychometrika (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1610.03375, https://pure.uva.nl/ws/files/36294562/10.1007_s11336_018_9607_z.pdf

zbMATH Keywords

factor modelspartial correlationszero-order correlations

Mathematics Subject Classification ID

Applications of statistics to psychology (62P15)

Abstract: In a unidimensional factor model it is assumed that the set of indicators that loads on this factor are conditionally independent given the latent factor. Two indicators are, however, never conditionally independent given (a set of) other indicators that load on this factor, as this would require one of the indicators that is conditioned on to correlate one with the latent factor. Although partial correlations between two indicators given the other indicators can thus never equal zero (Holland and Rosenbaum, 1986), we show in this paper that the partial correlations do need to be always weaker than the zero-order correlations. More precisely, we prove that the partial correlation between two observed variables that load on one factor given all other observed variables that load on this factor, is always closer to zero than the zero-order correlation between these two variables.

Recommendations

• The manifest association structure of the single-factor model: insights from partial correlations
• An inequality for correlations in unidimensional monotone latent variable models for binary variables
• On the Correlation of Common Factors with Variance Not Accounted for by the Factor Model
• Miscellanea. On the identification of a single-factor model with correlated residuals
• A test of multivariate independence based on a single factor model

Cites work

• Adjustment of an Inverse Matrix Corresponding to a Change in One Element of a Given Matrix
• An inequality for correlations in unidimensional monotone latent variable models for binary variables
• Conditional association and unidimensionality in monotone latent variable models
• Image theory for the structure of quantitative variates
• Model Selection and Model Averaging
• Multiple rectilinear prediction and the resolution into components
• Testing the conditional independence and monotonicity assumptions of item response theory
• The manifest association structure of the single-factor model: insights from partial correlations

Cited in

(2)

• The manifest association structure of the single-factor model: insights from partial correlations
• On the Correlation of Common Factors with Variance Not Accounted for by the Factor Model

This page was built for publication: Unidimensional factor models imply weaker partial correlations than zero-order correlations

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q725296)