Framing Matters in Gender-Paired Dictator Games

We show that social context matters in gender-paired dictator decisions. Our experiment investigates the influence of gender-pairing and framing on monetary transfers in a 2x2x2 design where sender gender, recipient gender, and frame, i.e. give or take, are varied. We are the first to combine all three variables and uncover that giving information about the gender of the recipient accommodates framing effects. If each of the three manipulated variables were to be analyzed independently, our data would confirm previous findings where females transfer more than males and framing has no effect (Eckel and Grossman, 1998; Dreber et al., 2013). However, we investigate the manipulated variables in interaction and find that framing matters when information about recipient gender is salient. For both genders, transfers in opposite-sex pairs are always higher than in same-sex pairs, but signicantly higher in the take frame. We thus suggest that the gender composition of the sample, gender-pairing, or beliefs about the counterpart's gender should be controlled for in experiments testing gender differences in social interaction.