Donald MacKinnon has a fascinating series of quantitative papers on creativity in the late 60's. He gets architecture profs to rank the creativity of pro architects (and likewise with mathematicians and research scientists), then gets them to take various psychometric tests.
Conversation
He finds poor correlation between creativity rankings of all these groups and IQ[1], but fairly strong effects associated with various personality factors—e.g. Big Five openness to experience (d=0.31) and CPI tolerance (d=0.77) and self-acceptance (d=0.69) among researchers[2].
Replying to
I was surprised also to see how many identical twins some researchers had managed to pull together to study the heritability of these psychometric scales! [e.g. 3, > 100!]
[1] andymatuschak.org/files/papers/M ch. 7
[2] andymatuschak.org/files/papers/H
[3] andymatuschak.org/files/papers/J
1
6
I found a bunch of nice meta-analyses around these topics (e.g. [4] was quite a nice bibliography), but creativity is awfully nebulous, and these methods are awfully diverse, so I'm not sure how much sense aggregation really makes.
[4] andymatuschak.org/files/papers/F
1
7