When building QALYs (Quality-adjusted Life Years), the crucial element is the valuation of health state utilities. The three main valuation techniques, the rating scale, the standard gamble and the time trade-off, should theoretically provide equivalent results. However, differences in frame, as well as the failure of the normative theory
of expected utility from a descriptive point of view, give rise to significantly different utility values according to the estimation technique. We propose to analyse the problem of the best technique from an empirical perspective, comparing the three methods on the basis of the consistency of elicited responses. We use a sample of sixty individuals to analyse inconsistent responses in a qualitative fashion and identify the
potential sources of inconsistency. The methods based on a trade-off exhibit higher inconsistency rates. However, we find that the observed rate can be reduced through the introduction of visual aids during the administration of interviews.