I don't know, but here are some possibilities: 1) Change in competition in classes that were graded on a curve. If CS became more popular, it could have become more difficult to be an A student. 2) Change in curriculum. CS could have gone through a change in subject matter. At my alma mater in the early nineties, CS included a huge amount of math. The Math and Physics majors also had the same kind of ratio distortion as CS. There were no advertisements to boost male enrollment in Math and Physics. 3) A few new majors were created (like Operations Research and Industrial Engineering) that offered admission into lucrative jobs like banking and consulting, while being much lighter on the algorithms and math. A large number of people from CS went into those majors (both male and female). There were certainly a lot more women in Operations Research. People used to comment on how OR had all the women, and CS had all the immigrants from Asia.

Looking at the raw data, there's a boom in computer science degrees in the mid-80s, for both men and women. Then there's a significant drop-off, again for both men and women. However, in percentage terms the drop-off for women is larger.

To me, that suggests a narrative where the "bust" painted computer science as a less reliable, more risky field. As a result women were less likely to go back into it.

A similar boom/bust happens in the early 2000s. Again, the female share drops more and recovers slower.