Mike Bailey has a good paper, "Can Behavior Genetics Contribute to Evolutionary Behavioral Science?" in Crawford and Krebs.
He makes the interesting point that shared environments INCREASE differences between twins (and perhaps other sibs), possibly due to their hierarchic competitions, each child seeking his/her most advantageous niche.
Second, twins (and probably regular sibs) become MORE similar with age, even if separated, the reverse of what might be expected if social learning were the primary variable.
Third, even if you assume a finite set of motor skills and random generation of similarities, such as both wearing a rubber band on the left wrist, the chances for shared traits seem higher -- far, far higher -- for twins and sibs than for randomly chosen comparisons. But, no body's got data -- rough definitional problem for "trait." The closer we analyze an event, the less clear the boundaries -- common problem in our business.