Step-Families v Single Parent Families
I mentioned this before concerning studies with single parents and gay parents in my post In Life, All Other Child Rearing Factors are Not Equal. Let me reframe this in terms of step parents vs single parents and see if this helps.
When we study outcomes that depend on many variables and wish to look at only one variable such as family structure, we isolate that variable by artificially making "all other factors equal." These factors, like income, education, etc. are those that I have mentioned before. Each study lists the variables they chose to "control" for.
So if the only variable we wish to study is the difference it makes for children to be raise by step families vs single parent families, researchers attempt to make all other factors equal.
Say we choose from the pool of available subjects a random sub-set of forty children who are being raised in step families. Then we would need to find a sub-set of forty children being raised by single parents who matched the step-families in every other important way. They must have the same average household income, the same average education, the same average age and the same per cent who own a home in the suburbs, etc.
For example, if we have a step family where both parents earn $30,000/yr, can we match this family with one where the single parent makes $30,000/yr also? No! We have to find a family where the single parent makes $60,000/yr so the total household income in these matching families will be the same.
When we step back afterward and look at the group of single parents we have chosen, we find that in order to match the step-families, we had to chose a sub-set of single parents whose qualities are well above the average for all single parents in general.
When these studies find no significant difference between these carefully matched groups, does that mean that a child in an average step-family is no better off than a child in an average single parent family? No, of course not. Only when the researchers selected out a group of higher income, better educated, house in the suburbs single parents to make "all other factors equal" did they find no differences between the two groups. Do you think the typical single parent makes $60,000/yr?
If instead, we select a completely random set of children in step-families and compare these with a completely random set of children in single parent families, no matching at all (or look at the census data), the children in two parent families are overwhelmingly better off than the children in one parent families because, in the real world, all other things are not equal.
Is it better for children to have two parents instead of just one as the AAP statement contends? You betcha!