However, women's work is only more highly-approved if their gender is not mentioned, the study suggested.
The paper, authored by a group of six students from California Polytechnic State University and North Carolina State University, has been published online, but is not yet peer-reviewed.
During their resarch, the team analysed of the behaviour of over a million users of Github, a code-sharing website where users can collaborate, post their work, and have others give them feedback.
The students paid particular attention to 'pull requests' - code submissions contributed to a project by a developer. After receiving a pull request, the owner of the project can choose whether or not to accept it.
The researchers found that pull requests from women to projects they were not involved with were accepted more often than those from men.
However, when the women's gender was identifiable, through information listed on their Github profile or elsewhere, their pull request acceptance rate dropped by 9.3 per cent, taking it below the rate for men.
Looking for possible explanations for the disparity, they examined whether fixes made by women were more needed in certain projects, or whether they only had an advantage in certain programming languages. However, no correlations were found.
The team concluded: "Our results suggest that although women on Github may be more competent overall, bias against them exists nontheless."
If accurate, the study would provide extra ammunition to those who believe the world of tech has a problem with women.
This is evidenced by the gender disparity at leading Silicon Valley companies - only 18 per cent of Google's global technical staff are women. At Facebook, the figure is 16 per cent.