Abstract
Using different measures of constructs in research to develop robust evidence of relationships and effects is seen as good methodological practice. This assumes these measures possess high convergent validity. However, proxies—alternative measures of the same construct—are rarely perfectly convergent. Although some convergence is preferred to none, this study demonstrates that even modest departures from perfect convergent validity can result in substantial differences in the magnitudes of findings, creating challenges for the accumulation and interpretation of research. Using data from published research, the authors find that substantial differences in findings between studies using desired and proxy variables occur even at levels of convergent validity as high as r = .85. Implications of using measures with less-than-ideal convergent validity for the interpretability of research results are examined. Convergent validities above r = .70 are recommended, whereas those below r = .50 should be avoided. Researchers are encouraged to develop and report convergent validity data.
Keywords
Get full access to this article
View all access options for this article.
