This paper presents the results from an experiment investigating whether framing affects the elicitation and predictive power of preferences for cooperation, i.e., the willingness to cooper- ate with others. Cooperation preferences are elicited in three treatments using the method of Fischbacher, Gächter and Fehr (2001). The treatments vary two features of their method: the sequence and order in which the contributions of other group members are presented. The predictive power of the elicited preferences is evaluated in a one-shot and a finitely-repeated public-good game. I find that the order in which the contributions of others are presented, by and large, has no impact on the elicited preferences and their predictive power. In contrast, presenting the contributions of others in a sequence has a pronounced effect on the elicited preferences and reduces substantially their predictive power. Overall, elicited preferences are more accurate at predicting behavior when others' contributions are presented simultaneously and in ascending order, like in Fischbacher, Gächter and Fehr (2001).
