Indeed, but the author only suggests it is not exclusively a pipeline problem. Having looked into these figures in the past in some previous research, there is certainly an element of pipeline problem, but it certainly isn’t exclusively that. That pipeline problem doesn’t start at birth (so the social conditioning and self-exclusion happen after that). It’s also not just one pipeline, like any flow of ‘things’ through other ‘things’ it’s fractal. Also, let’s remember that pipeline problems, like any queue, can occur just as much because of the faster exit rate from the profession, than entry rate into it (akin to the classic ‘buffer underruns’ that generated so many CD coasters in the 90s). That is rightly touched upon in the article as retention and that retention, or lack thereof, like any viable risk, has an impact. The cost of recruiting to fill the role and the loss of competitive advantage when the knowledge walks out of the door.
In our experience from attendance at each @CodeClub around the UK, children aged 9–11 who take up coding are split about 47:53 as girls:boys with many hitting or exceeding 50:50. So at that age, there is no reason for the pipeline under-run for women and I’d argue that’s where the normal should be. It is only when we get to college and university ages where we start to see the huge disparity in the figures your graph rightly shows.
A major caveat too. These studies are not naturally longitudinal. That means they take snapshots in time and don’t follow the individual through their life choices or ‘forced’ decisions. They have some value, but in terms of causal links, this sort of research provides next to no contribution to a long-term body of knowledge of the phenomenon, unless you keep snap-shotting to watch the macroscopic change happen and you’re only interested in reporting the figures, not fixing the problem or determining why its happening. Snapshotting data like that is akin to a really basic monte-carlo method, and as the saying goes, it’s fast but any conclusions are probably wrong :)