Construct labeling: public trust and scientific credibility

Analyses should logically ensue from the data on which they are based (Morely et al., 2001). When such data are mislabeled (whether knowingly or not) and, in turn, research findings are misconstrued, the integrity of a discipline’s knowledge base is compromised and scientific progress may be delayed, as future investigations in a topic area are misdirected and resources are wasted pursuing false leads (Stroebe et al., 2012). Moreover, as has become increasingly evident, publicity surrounding such mislabeling, and the ensuing findings, has invariably eroded public trust in science (Ioannidis, 2012). Construct labeling within the scientific community is, thus, a serious matter that requires action. One among many examples of construct mislabeling is a recent paper, “How Much (More) Should CEOsMake? AUniversal Desire forMore Equal Pay,” Kiatpongsan andNorton (2014) claim to have compared people’s estimates of chief executive officer and unskilled worker actual wages to their ideal for what those wages should be1. Kiatpongsan and Norton base their actual and ideal wage comparisons on four items taken from the 2009 International Social Survey Programme: Social Inequality IV (ISSP Research Group, 2012). The ISSP Social Inequality module was administered to 55,659 respondents from 41 countries between February 2009 and November 2010. A review focusing specifically on the data collected in the United States, however, reveals that Kiatpongrsan and Norton’s wage-comparison analysis rests on a mislabeling of its inherent core constructs: viz., people’s estimates of the wage disparities between what CEOs and unskilled workers are actually paid vis-à-vis people’s ideal wage differential.