The (In)validity of Self-Construal Scales Revisited

Abstract

This paper offers a critical rejoinder in the discussion concerning the viability of self-construal scales. It is our contention that the existing data reported in our previous article (Levine et al., 2003) and elsewhere are more than sufficient to justify the conclusion that self-report scales purporting to measure interdependent and independent self-construals as 2 orthogonal constructs lack validity. The arguments to the contrary offered by Gudykunst and Lee (2003) and Kim and Raja (2003) are disputed. Additional data (N = 1,013) show that neither age, occupation, sample size, standardization, nor a 2nd-order structure can account for the problems we documented previously. Although we see potential utility in the self-construal construct, we believe that the 3 primary scales fail to meet reasonable and accepted social scientific standards. Difficult conceptual problems will need to be solved prior to the development of new and improved measures.

The Effects of Self-Construal and Religiousness on Argumentativeness: A Cross-Cultural Analysis

“…Scholars have recently debated the validity of self-construal measures, such as Singelis' 1994 scale (Bresnahan et al, 2005;Gudykunst & Lee, 2003;Kim & Raja, 2003;Levine, Bresnahan, Park, Knight Lapinski, Lee, et al, 2003;Levine, Bresnahan, Park, Knight Lapinski, Wittenbaum, et al, 2003). However, this study utilizes the measure because as Gudykunst and Lee (2003) argue, there are more than 50 studies using Singelis' and other similar self-construal scales with sound construct validity.…”

A Multimethod Multitrait Validity Assessment of Self-Construal in Japan, Korea, and the United States

“…Together, these disconfirming and contradictory results either reflect problems in the way that self-construal has been measured, deeper conceptual problems, or both (Levine et al, 2003a(Levine et al, , 2003b. Arguing on the basis of the Cronbach and Meehl (1955) nomological network approach to construct validation, Gudykunst and Lee (2003) contended that since self-construal scales have been shown to correlate with outcome measures as predicted, the scales must be construct valid.…”

“…Arguing on the basis of the Cronbach and Meehl (1955) nomological network approach to construct validation, Gudykunst and Lee (2003) contended that since self-construal scales have been shown to correlate with outcome measures as predicted, the scales must be construct valid. Levine et al (2003aLevine et al ( , 2003b, however, speculated that such correlations might be spurious and attributable to item content confounds. Gudykunst and Lee (2003) and Levine et al (2003aLevine et al ( , 2003b agreed that "multi-trait, multi-method procedures would be ideal ways to establish validity, and future attempts to establish the validity of self construals should use these procedures" (Gudykunst & Lee, 2003, p. 264).…”

“…Based on these results, Osyerman et al cautioned against the uncritical use of selfconstrual, and described the state of the literature as "tentative" (p. 87). 2 Similarly, Levine et al (2003aLevine et al ( , 2003b, using a variety of methodologies including meta-analysis, demonstrated that average effect sizes in self-construal research were often very small and widely variant across studies-and that many findings were in the wrong direction from hypotheses. Evidence from factor analyses further indicated that items comprising the various self-construal scales did not produce stable factors.…”

Evidence for the Construct Validity and Reliability of the Co-Cultural Theory Scales

“…Many studies do not provide either reliability or validity information for the measures of theoretical constructs (Hunter & Schmidt, 1990). Not uncommonly, significant bodies of literature may be based on measurement that is later found to be flawed (for a discussion of one such problem see Levine et al, 2003a;2003b). Thus, establishment of a measurement model prior to testing a theoretical model is recommended (Hunter & Gerbing, 1982).…”