"Don't worry, honey, the other children won't judge you on your looks."

Joseph recently brought up the subject of separated-twin studies. Let's take things to the next level with a thought experiment using clones. We'll have to go full-bore mad scientist mode here and assume we have exclusive access not only to an unlimited supply of clones but to undetectable plastic surgery techniques that can be performed without the subject's knowledge.

From early childhood, we assign different treatment combinations to different groups of our genetically identical subjects. We adjust features to match those that tend to elicit strong responses in terms of attractiveness, intelligence, likeability, and other traits. We modify height and body type, again making sure to cover the extremes. We even determine vocal characteristics.

After about thirty years of this, we assess the subjects using a variety of personal, professional, and academic metrics, the same sort of metrics often used in twin studies.

Now we come to the big question, are you willing bet a sizeable amount of money that none of these treatments are significant? If the answer is no, you shouldn't put a lot of faith in twin studies that measure these same factors because these studies never control for appearance.

How we look does not determine who we are but it does have a big impact on the life that we lead. It makes people more or less likely to want to hang out with us (particularly if those people see us as potential mates). Through the halo effect it affects people's opinion of us (including opinions in the form of grades and professional evaluations). Perhaps most importantly, it changes the way we see ourselves.

Separated twin studies have often been held up as a gold standard. Pyrite might be a more apt description. The data are invariably confounded in numerous ways that are difficult to correct for. In addition to the factors mentioned above, twins (barring surrogacy) share a prenatal background, the same relative age compared to their classmates and the same absolute age. You could easily find yourself comparing a pair of twins who suffered from mild FAS, were the youngest students in their classes, and were born in 1960 against a pair of twins who had excellent prenatal care, were the oldest in their classes and who were born in 1950.

Twins can be incredibly useful as test subjects, but the popular notion that these separated twin studies provide clean, unambiguous findings is simply wrong. Every one of them has serious problems with confounded data and these problems hold even for perfect studies with twins separated at birth, randomly assigned to adoptive parents and kept isolated from each other (conditions that aren't met that often). Personally, I have trouble overlooking all these concerns and putting a lot of weight into these findings, but, of course, I didn't inherit a trusting nature.