The specific concern I was addressing in the first part of my post was the drop in female CS majors that occurred in the mid 1980s. I was only addressing the difference in graduates between then and now (not all female CS graduates by any means) and only what I believe to be the largest factor (as there were certainly other economic and social factors that affected the situation).

Considering 1985 and 2007 (1985 having the largest amount of female graduates in math and CS), we see women going from 38.8% to 25.4% of the labor force. All we need to show is that the difference in female math and CS graduates in 1985 were primarily mainframe operators and/or keypunch operators (although the number of graduates in math could affect the CS numbers). In addition, I nowhere stated that women of that period were ignorant of the requirements of keypunching.

My mother's earnings as a mainframe operator and keypunch operator with a degree in CS far outstripped the earning potential of someone with just a keypunching certificate. If others were aware of this difference, then getting the degree would be worth the extra work (keeping in mind that a degree also carries the possibility for promotion). This is not much different from women getting bachelors degrees only to work at secretarial positions today or pursuing a masters in Nursing when an associates degree suffices for all non-management jobs.

The second part of my post was focused on the overall numbers (that remaining 25% of women who have always worked in CS) and a potential reason for the relative stability of that number over time (presuming the peak and decline in the 1970-80s can be accounted for independently).