For one, there are observed gender disparities in certain things (you mentioned spacial reasoning) which there is no apparent or plausible political agenda in artificially creating. In fact, the predominant political agenda is largely the idea that there are no differences between men and women. For another, disparities are often observed in infants. What mechanism do you propose communicates subtle social norms to infants? You also have proposed no plausible explanation as to where these expectations came from in the first place.

Further, you are misunderstanding the idea I'm proposing. I'm sure there are individual men who have, for instance, very poor spacial reasoning and individual women who have very good spacial reasoning. One does not reason about individuals the same way they reason about populations. As a population, men, on average, might be better at spacial reasoning. As an individual man or woman, a test of spacial reasoning will yield orders of magnitude more data about their spacial reasoning abilities than simply observing their sex. I have many personality traits that are more common in the opposite sex--I'm sure most people do. It's a fundamental misunderstanding of averages that leads one to take this as evidence for your viewpoint, however.

Ultimately, measures of a population only explain populations. They may explain why, as a population, most programmers are men. They do not explain why your sister is or isn't a programmer.