Jordi Duch of Northwestern University and her co-authors took a circuitous route to this conclusion. They compared the publication rates of male and female research university faculty in chemical engineering, chemistry, ecology, industrial engineering, material science, molecular biology, and psychology. These seven disciplines vary considerably in the amount of resources that scientists need to do research, as measured by what they typically spend in a year. At the low end is industrial engineering, in which much of the work is "theoretical and computational in nature" and "faculty tend to train a small number of students at a time." At the high end is molecular biology, which requires extensive labs, lots of expensive equipment, and, often, numerous grad students and postdocs to do the bench work.
Because of their relatively small requirements,industrial engineering faculty "do not need to compete against one another for limited resources," the authors state. The "institutional support" needed to do battle for funding is therefore a relatively unimportant "factor in productivity" in the field, the authors suggest. For molecular biologists, on the other hand, winning large competitive grants is crucial to supporting their labs. "Institutionally granted resources or institutional support for securing large grants" are vital to this competition and therefore become "crucial components of academic success," the authors write.
An analysis of the publications of more than 4000 faculty members "fully confirms our hypotheses," the authors state. The differences between the publication rates of male and female industrial engineers are negligible. Female molecular biologists, on the other hand, "consistently publish at a rate significantly lower than" their male colleagues. This shows, the authors conclude, that "gender differences in institutional support have had a crucial effect on the publication rates of females."
Although the authors caution that their results show only correlation, co-author Luis Amaral of Northwestern University finds them "very suggestive of causality," according to a an article in Inside Higher Ed. You can read the PLOS ONE study here.