An Experimental Examination of Equivalence Failures in Multi-Cultural Comparative Research

As the global economy integrates, there is an increased need to understand international business phenomena. This forces a reliance upon multi-cultural marketing research, which is evidenced by a marked increase in international studies, particularly multi-cultural comparative research (Sin, Cheung & Lee 1999). Central to the usefulness of this research is the question of the validity and comparability of results, which is greatly affected by a property known as equivalence of test instruments; or the degree to which the scales and the items in them are seen to be the same across cultures. To date, no research exists that approaches the problem of equivalence from a position of knowledge, or beginning with a known equivalence error and then tracing its psychometric effects. The dissertation fills that need by experimentally manipulating a translation error in a scale and then using conjoint analysis to decompose respondents\u27 choice patterns for items or attributes that cause equivalence failures. Results from a probability sample of American consumers indicate that: (1) current techniques to diagnose equivalence failure can adequately identify items that are inequivalent, but also (2) that items failing equivalence have a pronounced tendency to attenuate other items in the scale. Conjoint results were similarly affected by translation error. The presents serious implications for international researchers and global marketing managers, including some question as to the usefulness of existing scales in multi-cultural contexts. In addition, theoretical development regarding response behaviors in needed to explain the differences between control and experiment groups with respect to non-manipulated items. Further research is also needed to systematically examine the effect of translation error across response formats and scale types