A very different conclusion was reached by a careful meta-analysis of all the available twin data, recently published in a large review that Mayer and McHugh fail to even mention.
Phew.. Dense read. M&M used a dataset that did not measure the traits they're addressing. A larger review of multiple datasets showed different results.
Of the six studies using proper probability sampling methods that have been published in the peer-reviewed literature in the past 16 years, they include only one — and it just so happens to be the one with the lowest estimate of genetic influence of the entire set.
McHugh & Mayer cherry-picked their data to fit their desired results.