When Validity Testing Lacks Validity: Comment on Levine et al.

The death of self‐construal research has been greatly exaggerated. Levine et al. (2003) highlight the alleged methodological limitations of the current body of the self‐construal model of culture. Their allegations, whether true or not, require a fair investigation. Careful examination of Levine et al. reveals pervasive logical flaws, methodological errors, and interpretation biases, which stem from ethnic stereotyping and other erroneous assumptions at various junctures. These fundamental errors, in turn, render their conclusions untenable. In testing their 1st set of hypotheses, which they proclaim to be “central to the validity” of self‐construal scales, Levine et al. bifurcate the participants into “Westerners” versus “Asians,” and then impose their a priori stereotype of how Westerners ought to be versus how Asians ought to be, which is scientifically unacceptable. This is an unscientific validation criterion (comparable to crude ethnic stereotypes such as “Asians are smarter”) and should not be used to judge the scales' validity. Next, in their discussion of priming, Levine et al. assert that the construct validity of the interdependent self‐construal requires that it must be susceptible to priming. This is an elementary logical fallacy. It simply does not follow from the definition of the interdependent self‐construal and therefore cannot be used as a criterion which reflects on construct validity. Their priming experiments are also deeply flawed by the unrepresentative sample of participants. Further, in their factor analyses, Levine et al. set up the “straw man” of a “universal a priori 2‐factor solution” although no self‐construal scholar or theory insists that there are precisely 2 universally applicable self‐construals or that the current self‐construal scales are perfect. The results of statistical analyses, such as those in Levine et al., crucially depend on the selection of presuppositions. Levine et al.'s presuppositions are untenable, yet they are passed off as self‐evident criteria for validity testing. Their article tries to create an illusion of finality, but is pervasively and fundamentally flawed.