A usual method for valuing health states is Time Trade-Off (TTO). It has been found in web surveys that TTO is affected by anchoring, as the starting point in the task affects the results, thus violating procedural invariance. The aim of this study is to explore whether anchoring is also present in face to face interviews. 147 Colombian subjects (111 male, 36 female) valued 5 better than death EQ-5D health states in a face to face interviews. Subjects were randomized to two different starting points. We tested for the presence of anchoring by means of a battery of non-parametric tests, including Kolmogorov-Smirnov and Wilcoxon-Mann-Whitney tests. Shapiro Wilk test discarded normality for all the distributions. In all tests anchoring was significant for 4 out of 5 health states, which was supported also by visual analysis of the Box and Whiskers diagrams. A higher starting point increased the elicited value by 15% - 188%. The health state with no anchoring was one with near perfect health. The size of the anchoring effect was not uniform among health states. In conclusion, anchoring effects may bias face-to-face TTO valuations. The size of the anchoring effect is small compared to other situations, such as business valuations, but it is still relevant for health policy.
