No, twin studies are not about twins separated at birth (usually)

Since I was young, I heard the phrase “twin studies” and naively assumed that these studies were conducted something like this:

1. Take two identical twins

2. Raise the twins in two different environments

3. Look at differences in life outcome and traits, and presume that any differences must be due to environment (not genetic differences).

But this is not at all how “twin studies” actually work!

It is extremely rare for twins to be raised apart (and it’s unlikely that an ethical review board would approve intentionally separating two twins at birth). Adoption agencies and child welfare systems generally try to keep twins together.

(And, contrary to what certain classic Disney films led me to believe in childhood, divorced parents of identical twins usually don’t split custody by keeping each twin ignorant of their sibling’s existence until they meet at summer camp.)

While there is a famous study that included examinations of identical twins who were reared apart, this is not what the modal twin study looks like. Most twins studies focus on twins that were raised in the same environment.

At first blush, shared environment might seem like it would be an obvious confounder. If all we had were a bunch of studies looking at identical twins, and observing that “yup, these two identical twins raised in the same household sure do have similar life outcomes,” it would be difficult to determine how much of that was due to shared environment versus shared genetics.

But here’s the interesting part: “outcome similarity that results from shared environment” is something that we can account for, because we have plenty of examples of kids who are raised in a shared environment without sharing 100% of their DNA. Remember, not all twins are identical!

Identical vs fraternal twins

Identical twins are monozygotic: they’re formed from a single zygote (from one egg and one sperm cell) which split and formed two embryos. Because both twins were formed from the same combination of egg and sperm cell, they have the same genetic material from each parent, making them 100% genetically similar.¹

However, fraternal twins are dizygotic, formed from two zygotes. If two separate eggs are ovulated in the same cycle and each is fertilized by a different sperm, two embryos can develop and implant. Both fraternal twins have the same same mother and (usually) the same father, but they’re two different egg cells and two different sperm cells, so they’re about as similar as you’d expect any two siblings to be. (On average works out to around 50% genetic similarity, as opposed to the 100% genetic similarity you’d expect from identical twins.)

Now, we look at the delta

If we have data for “fraternal twins raised in a shared environment” and also have data for “identical twins raised in a shared environment,” then we can account for the shared environment! By comparing the degree of similarity for each group of twins, we can determine the extent to which genetics contribute to a particular traits.

For example, if we find that identical twins are highly likely to share a particular trait, but fraternal twins are only moderately likely to share a particular trait, then we can presume that the delta between “highly likely” and “moderately likely” is due to the fact that the identical twins are 100% genetically similar while fraternal twins are only 50% genetically similar.

If the trait were simply the result of environment and didn’t have a genetic component, then when examining that trait we would expect fraternal twins to be just as similar to each other as identical twins.

Back to the original question

There are no big thorny ethical problem involved with carrying out a twin study, because conducting this type of twin study doesn’t have to involve separating twins at birth. And you can get pretty high enrollment, because study participation is often as simple as going in for a physical examination and blood test, and/or a questionnaire or survey. Participation doesn’t require any unusual life circumstance beyond “having an identical or fraternal twin who is willing to undergo the same examination.”

It’s totally feasible to have twin studies with N > 1,000. The TwinsUK registry, for example, has over 15,000 twins aged 18-100.

A tale of two twins

A couple has identical twins and puts the two boys up for adoption. One boy is adopted by a Spanish family who names him Juan, while the other is adopted by an Egyptian family who names him Amal.

Years later, the birth parents receive a letter from the family who adopted Juan, along with a photo of him. The wife is happy to see how Juan has grown up, but then suddenly looks sad. The husband notices and asks if she's disappointed that they haven't received a photo of Amal. The wife responds, “No, they’re identical twins: if you've seen Juan, you've seen Amal.” A beat later, she adds, “I was just thinking that if both families had sent us data about the boys’ personalities and health outcomes, we could compare the two and know that the delta was due to environmental differences and not genetic factors.”

The husband nods. “I would be interested in seeing that study, but then again, I’m an interested party. A study involving our genetic offspring would be interesting to me, but to an outsider, it would just be a study of N=1. However, if we want to encourage our sons to contribute to scientific research, we could send them information about how to sign up for a twin registry like the Stanford Twin registry, the TwinsUK registry, or the Swedish Twin Registry, among others.”

“That’s a wonderful idea,” said the wife. “But how do you suppose that we could go about informing the people we know about this opportunity?”

The husband considers the question. “I suppose that we could post about it on the internet. And if people find that post interesting, or want more people to know about how they can enroll in twin studies, they can share the post, too.”

1. For modeling, it’s useful to treat monozygotic twins as sharing 100% of segregating alleles; in reality, tiny post-zygotic mutations can make them not perfectly identical.