The weak point in all of this is the definition of IQ - since there is no obvious one the motivated researchers can manipulate it to have more or less difference in both value and variation between the sexes (or other groups).
One can certainly argue that "IQ" is just whatever IQ tests measure. Despite this there are a lot of correlations involving IQ that mean that IQ tests are at least partially capturing something that's interesting. And IQ tests are fairly standard which means that it's not so easy for researchers to manipulate them.