The Older Paper: Habitus My Arse

The graph of the day comes from a paper called "What Happens When We Randomly Assign Children to Families?" by Bruce Sacerdote (.pdf), which exploits data on adoptees that were, as the title suggests, randomly assigned to families once these families were accepted, with the exception that "families with all boys or all girls are allowed to request a child of the opposite gender" (p. 7). (This is important because a popular criticism is that the nonrandom assignment of children to families might distort the results of such studies and thus lower our confidence in the results.) Via Alex Tabarrok, here's one of the paper's graphs:


To the extent that this generalizes, it leaves pointless all sociological theorizing about the social transmission of income.

The paper does find some parent-to-child social transmission for educational attainment and whether or not a person drinks, but none for height (no surprise there) or obesity and BMI (calling into question popular views on how parents influence their children's body weight later in life by shaping eating habits). The results for smoking are inconclusive.

The Marginal Revolution post which is the pointer for this illustrates the general rule that the quality of comments declines once you tackle a nature-nurture topic. With that in mind, if you wish to comment, consider whether you should have a look at the paper first.

Added: At MR, commenter Bill writes: "Parents income at the time of adoption? Hmmm. A number of college educated couples will have low income at the time of adoption and higher income during the child's gradeschool and highschool development. I wish the axis were average adult income over the child's school years." That's a fair point.