Which is why I wish there was better data available about gender differences in ability and interest for programming.
However coming at it from the opposite end, remove all barriers that you can find, and work to make sure your hiring process hires women and men who perform equivalently on the job. As long as those are true, you are probably doing about as well as you can to get the best workforce you can.
Note "who perform equivalently on the job" is a subtle and important criteria. As an example, men tend to have an advantage on standardized tests. Therefore I am firmly in support of such measures as adjusting SAT scores to be better predictors of class performance, and then removing all signs of gender from applications before passing them to the hiring committee. Not with the goal of achieving any particular mix of men/women, but with the goal of getting the best performing class that you can from the available applicants.
However coming at it from the opposite end, remove all barriers that you can find, and work to make sure your hiring process hires women and men who perform equivalently on the job. As long as those are true, you are probably doing about as well as you can to get the best workforce you can.
Note "who perform equivalently on the job" is a subtle and important criteria. As an example, men tend to have an advantage on standardized tests. Therefore I am firmly in support of such measures as adjusting SAT scores to be better predictors of class performance, and then removing all signs of gender from applications before passing them to the hiring committee. Not with the goal of achieving any particular mix of men/women, but with the goal of getting the best performing class that you can from the available applicants.