Studies across many science, technology, engineering, and math fields have documented that women author fewer papers than expected based on the percentage of women in the field. But it has been challenging to figure out why that’s the case—in part because it’s hard to identify who was potentially left off author lists. The new study overcame this hurdle by drawing on a uniquely detailed data set of nearly 10,000 U.S.-based scientific research teams, the 128,859 individuals who were employed by those teams from 2013 to 2016, and the 39,426 journal articles they produced. With these data, the authors were able to link who did the work with who ultimately got credit for it. (The study inferred gender based on names, a method that involves some error and doesn’t permit the identification of nonbinary researchers.)