Moving Beyond Traditional Methods of Survey Validation

Click to increase image sizeClick to decrease image size

Notes

1. Kane seems to have missed this point when he expresses resistance to the proposal of formulating causal theories about the relationship between attributes and their putative measures on the grounds that causal inference is indeed not a straightforward matter. This seems to repeat behaviorism’s mistake of allowing perceived epistemic limitations to unnecessarily constrain the formulation of theory. This said, inspection of the fit of an appropriately matched data set and latent variable model can provide a corroborative test of such causal hypotheses (e.g., Borsboom, Mellenbergh, & van Heerden, 2003), as could methods for investigating response processes such as cognitive interviews.

2. In the metrological literature, the concept of definitional uncertainty (JCGM, 2012) provides an upper bound on the total uncertainty and, therefore, the quality of measurement results.