The Impact and Consequences of Correcting for Acquiescence When Correlated Residuals are Present.

Identifier:  imarina:9435587

Handle: https://hdl.handle.net/20.500.11797/imarina9435587

Authors:  Hernández-Dorado A; Ferrando PJ; Vigil-Colet A

Abstract:
In spite of the interest generated by controlling variance unrelated to the content in personality measures, few studies have assessed the combined impact of acquiescence (ACQ) and correlated residuals in the factor-analytic structural solutions that serve as a basis for item calibration. This article compares three control procedures in databases that simultaneously exhibit acquiescence (ACQ) and correlated residuals: the SIREN method (Navarro-Gonzalez, et al., 2024; ACQ control), the MORGANA method (Ferrando et al., 2022; 2023; correlated residuals control), and a combined double control method. A 'control' procedure was also examined in which the presence of both determinants was ignored. The findings reveal significant differences between the three control methods, with the ACQ control method and the combined method proving more effective. Moreover, when the residual variance shifts to the factor loadings, it appears to be captured by the ACQ correction method. The findings suggest using a mixed correction procedure in order to obtain less biased item parameter estimates when both sources of unwanted variance are operating.