Gender and Choice over Co-workers: Experimental Evidence

We study whether choice over co-workers matters for performance in gender-diverse teams. We carried out a lab-in-the-field experiment where students were randomly assigned co-workers meant to help them perform on tests. Co-worker allocation was randomized on two dimensions: (1) gender and (2) student preference for that co-worker at baseline. We find that randomly chosen male co-workers reduce the performance of females (12% of the average score), while preferred male co-workers have a positive yet statistically insignificant effect (6% of the average score). These effects are heterogeneous across the gender stereotype of the questions and materialize even though the two types of male co-workers have the same average ability. To investigate the mechanism behind these effects, we randomly allocated hints as an additional source of information across questions. We find that some (but not all) of these differences are driven by difficulty in accessing those hints in the presence of random male co-workers.