1/ In terms of macro-level prediction or assessment - correlations made between IQ and demographics (sex, race, etc.) have almost no redeeming value - and deserve about as much time and academic attention as phrenology or any other historical pseudoscience. 
2/ In terms of pseudoscientific qualities, both IQ demographics “research” and phrenology “science” share similarities in that:
a. data is derived from an inquiry format that is both arbitrary (with respect to intelligence) and biased (with respect to the designer of the test)
a. data is derived from an inquiry format that is both arbitrary (with respect to intelligence) and biased (with respect to the designer of the test)
b. data is bucketed into “demographic characteristics” that are often nothing more than arbitrary and/or superficial considerations
c. even if the methodology were somehow valid for this use, important individual variations are suppressed in the quest to define a "quantifiable" group similarity
3/ Given these limitations and others, IQ data has almost no meaningful use for interpreting a group trend. It's important to note that its creator, Alfred Binet, never intended the test to be used as such.
4/ He developed the first practical IQ test as a method to screen for childhood learning disabilities - to identify children who could possibly benefit from further assessment and engagement.
5/ Binet was very forthright about the limitations and prescribed uses of his test, and was very much against the “promotion of intelligence as a single, unitary construct.”
