Because that would be a different study.

You could compare their behaviors to non married teens or whatever else.

Or compare the homosexual and bi teens to hetero ones.

I’m sure such studies do exist.

But you can’t just have all these wild card variables lumped together.

In research you have to try to use parameters that best represent a population if you are doing a study on general trends.

It’s very common to exclude anyone who isn’t “average” from a study on average behaviors.

Especially when you know already that there are systematic differences.

For example. Married teens, always hetero, are engaging in sex.

That would distort the actual numbers if you added them into your computations.