Vignette equivalence and response consistency; the case of job satisfaction

We compare reported job satisfaction with vignette evaluations of hypothetical jobs by using a British, Greek and Dutch data set, containing 95 randomly assigned vignettes. In order to test comparability of international data sets recently the method of anchoring vignettes has been introduced by King et al. (2004). This intuitively and attractive idea requires the properties of vignette equivalence and response consistency. In our data set both job satisfaction and vignettes are numerically evaluated on a 0-10-scale. This fact allows us to interpret the evaluations as cardinal satisfaction values and to estimate satisfaction functions for vignettes and for the own job situation. We find that both functions differ significantly: vignette evaluations appear to depend on the own job situation and other individual characteristics. Hence, without correction for those differences in background characteristics, vignette evaluations are not comparable between individuals. Similar conclusions are reached for response consistency.