The influence of relevancy and prominence of an attribute on choice probability was tested by using an example suggested by Savage. Two experiments were carried out. In the first, 65 subjects chose between two alternatives, identical except for a small bonus added to the second one. The relevancy of the bonus was manipulated. It was found that the level of relevancy did have an influence on the probability of choice of the bonus-added alternative. In the second experiment, 46 subjects participated in a decision task similar to the first experiment but here the level of prominence of the bonus was manipulated. It was found that the probability of choice of the bonus-added alternative increased as its prominence was higher. The results are considered as representing context effects and are interpreted as supporting the adequacy of dependent random utility models of choice.