Papers from the International Association for Cross-Cultural Psychology Conferences
Publication Date
2009
Abstract
Differential item functioning or item bias is a usual threat in psychological research and many experts in the field such as Kline (1993), Nunnally and Bernstein (1994), and others have suggested various methods for its detection and removal. Item bias in terms of culture has been addressed by Poortinga, Van de Vijver, Leung, Muthén, and others, with most of the proposed methods attempting to minimize variance explained by culture itself by detecting and deleting culturally-biased items from the analysis. This is done through the detection of inequivalent –in terms of factor structure– items and their elimination before comparing for factor equivalence using “culture-free” correlation matrices. The currently proposed alternative statistical technique is to estimate indices of country eccentricity through individual differences Euclidean distance scaling models for the respective weirdness index to be computed and then proceed with variance and raw-scores adjustment, accounting for each item’s variance caused in terms of culture only. Analyzing the beforeadjustment and the after-adjustment correlation matrices through covariance structure analysis and Procrustean rotations, we might be able to evaluate the profits from the adjustment itself. The illustration data for this method were 20 measures from six countries (N=1,655) as presented in the Georgas et al. (2006) 30-nation study on family issues. Following the initial analysis of all six countries, all quintet and quartet combinations of countries were examined and evaluated for the before and after adjustment outcomes. For all 22 combinations, factor loadings after metric adjustment were slightly higher; also, for a large number of country sets, items not included in the factor structure before the adjustment were now entering it. Factor identities were not affected to a large extent, but the factor identities were somewhat altered, revealing interesting facets within each of the broader areas defined by the initial –before the raw score adjustment– factors.
ScholarWorks Citation
Mylonas, K. (2009). Reducing bias in cross-cultural factor analysis through a statistical technique for metric adjustment: Factor solutions for quintets and quartets of countries. In G. Aikaterini & K. Mylonas (Eds.), Quod Erat Demonstrandum: From Herodotus’ ethnographic journeys to cross-cultural research: Proceedings from the 18th International Congress of the International Association for Cross-Cultural Psychology. https://doi.org/10.4087/LSHZ1541
