Oh, that's fairly easy to answer: the field changed to be more like other engineering fields, and lo-and-behold, the percentage of women in the field gradually approached what it had always been in engineering.

The graph[1] that is usually shown is highly misleading, because it compares CS with "physical sciences", and law, but does not include engineering. If you add engineering, it all makes a bit more sense[2].

Also, you seem to be approaching this from the angle of "can we prove that it isn't sexism" (which is the common approach). It seems that the correct approach is for those who claim that it is sexism to make the case, which after looking at the evidence I think they absolutely have not. (I used to think it was, but that was because I bought the narrative without looking at the evidence).

And the narrative is strong. For example, the link to the percentage of women decreasing with the discipline becoming more engineering like is known. However, the explanation then shifts to "the men in CS got together and decided to make the discipline more engineering-like expressly in order to keep the women out". Seriously??

[1] https://media.npr.org/assets/img/2014/10/21/womencoding_wide...

[2] https://media.licdn.com/mpr/mpr/p/7/005/06b/02d/33fe648.jpg