Validating the use of anchoring vignettes for the correction of response scale differences in subjective questions