Abstract
Self-reports could be affected by 2 primary sources of distortion: content-related (CRD) and content-unrelated (CUD) distortions. CRD and CUD, however, might covary, and similar detection strategies have been used to capture both. Thus, we hypothesized that a scale developed to detect random responding—arguably, one of the most evident examples of CUD—would likely be sensitive to both CUD and, albeit to a lesser extent, CRD. Study 1 (N = 1,901) empirically tested this hypothesis by developing a random responding scale (RRS) for the recently introduced Inventory of Problems–29 (Viglione, Giromini, & Landis, Citation2017), and by testing it with both experimental feigners and honest controls. Results supported our hypothesis and offered some insight on how to pull apart CRD- from CUD-related variance. Study 2 (N = 700) then evaluated whether our RRS would perform similarly well with data from human participants instructed to respond at random versus computer-generated random data. Interestingly, the sensitivity of our RRS dropped dramatically when considering the data from human participants. Together with the results of additional analyses inspecting the patterns of responses provided by our human random responders, these findings thus posed a major question: Is humans’ random responding really random?
Acknowledgments
We thank Marianna Correnti and Edoardo Meroni for their help in the data collection.
Disclosure Statement
Luciano Giromini and Donald J. Viglione declare that they own a share in the corporation that possesses the rights to the Inventory of Problems. The other authors declare that they have no conflict of interest.
Notes
1 For one of their samples, Viglione et al. (Citation2017) described their participants’ ages in terms of classes of ages (18–24, 25–34, 35–44, 45–54, 55–67), rather than reporting M and SD values. As such, the values reported here are intended as approximate estimates.
2 IOP–29 items are copyrighted, so they cannot be reproduced without the IOP–29 authors’ permission.
3 Because homoscedasticity could not be assumed, the Welch–Satterthwaite method was used to adjust degrees of freedom.