We report the results of an original experiment that was designed to test the strength and robustness of the attraction effect. Rather than the usual simple tests for this effect, we consider a conceptually simple consumer purchasing task where alternatives are however difficult to evaluate. For the attraction effect to be observed, the consumer must go through two steps: the first is to find out that two or more options are comparable, which leads him to exclude the dominated alternatives. The second is to favor the dominant option over those that are not comparable. Our experiment allows us to determine whether and how many individuals stop before each of those two steps. The results confirm the existence of an attraction effect in our setting, but the effect is not strong. Indeed, only a minority of subjects perform the second step. The effect is not robust to introducing larger differences in prices among options and to widening the range of options to choose from. We conclude by showing that our subjects would benefit from relying more on performing asymmetric dominance editing rather than on their skills in the purchasing task.