Acquiescence bias: exploring the applicability of ipsative scoring and acquiescence modeling

Acquiescence bias, defined as the tendency to agree with an item irrespective of the content is an important methodological issue in cross-cultural psychological research. The present study compared two methods to control for acquiescence bias, namely ipsatizing and modeling an acquiescence factor using confirmatory factor analysis (CFA). Simulated data sets are being used to compare raw data estimation methods, ipsative estimation methods and the acquiescence modeling method on how well they can reproduce the true theoretical correlations between underlying factors of an uncorrelated bidimensional model. Three conditions were manipulated: factor strength, acquiescence strength and factor intercorrelation. Depending on the specific model conditions, correlations and CFA using raw data and ipsative estimation methods show big under- or overestimations. The acquiescence modeling method reproduces the true theoretical correlations very well, irrespective of the specific conditions. Estimating relationships between constructs by means of acquiescence modeling in CFA is much more widely applicable than previously thought and is advised, even when unbalanced item sets are used to measure constructs.