Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
Although discrimination remains prevalent, the reasons for its occurrence remain hotly debated. To disentangle vying explanations, researchers have begun using laboratory experiments. However, this research has not allowed, or studied, the effects of selection. In this paper, we examine discrimination in a variant of Berg et al.'s (1995) investment game where subjects can and cannot select partners. We find little evidence of discrimination without selection but significant discrimination with selection: subjects select and send more to partners of the opposite gender. The discrimination cannot be explained by performance (amount returned), but can be explained by tastes and beliefs towards each gender's trustworthiness.