As a statistical geneticist, I used to think that the heritability estimates from twin studies are broadly correct. They suggest that variance in traits like height and intelligence is mostly due to genetic variation between individuals. The limitations of twin studies have been studied extensively for decades, and no flaws that convincingly discredited them have been identified.
This is still the consensus view among geneticists, but recently new methods for estimating heritability have been developed by scientists like Sasha Gusev. Sib regression and related methods arrive at different and often lower heritability estimates for many traits, including intelligence. The balance of evidence still favors the results from twin studies, but there’s also clearly still something about heritability we don’t understand. Scott Alexander’s recent blog post provides a good overview and is balanced. It’s comprehensive and I have no insights to add. I only wish he had picked a less complicated trait than educational attainment, which is notoriously difficult to study and talk about calmly, to base his post on.
Nehaveigur 