Feminists as a whole (at least in academia) do NOT maintain that people are entirely a product of society. Rather, it's more that society is responsible for *much* of a person's development and *most* of the trends associated with groups of people, and it would be to women's (and men's) benefit if we were to remove the limitations that force those arbitrary trends. Gender identity happens to be one of those things that society *cannot* manipulate, but there are things society can affect (socioeconomic status, education and test performance, the way one expresses their particular gender, sex/gender-related behavior).
Regarding math test scores, these start out relatively similar and diverge significantly during high school and college, so there seems to be a social cause for (most of) the sex/gender difference. Whether it's due to teaching style or due to other social factors, sort of the thing that has to be acknowledged is that the differences we see (and certainly the larger differences of 50+ years ago) *are* largely attributable to society rather than biology.