[My apologies for formatting problems. Let me know if this still doesn’t look right. – B.C.]

Since there was some disagreement about whether I was correctly interpreting Judy Harris‘  The Nurture Assumption, I decided to go straight to the source.  Here’s our full exchange, posted with her permission.

Me:

I know that your work focuses mostly on personality traits. But economists are a lot more interested in what we call “outcomes” – income, educational attainment, health, etc. Do you think that shared family environment matters substantially more for more these “outcomes” than for personality traits? Or are you equally skeptical of the nurture assumption in these other areas, too?

Harris:

Hi Bryan,

You asked,

I know that your work focuses mostly on personality traits. But economists are a lot more interested in what we call “outcomes” – income, educational attainment, health, etc. Do you think that shared family environment matters substantially more for more these “outcomes” than for personality traits? Or are you equally skeptical of the nurture assumption in these other areas, too?

First of all, “Shared environment,” as defined by behavioral
geneticists, is not the same as “shared family environment” and
shouldn’t be interpreted that way. Shared environment, by definition,
is anything that makes siblings who grew up in the same family more
alike than a similarly related pair of siblings who grew up in different
families.

I don’t believe that shared FAMILY environment matters substantially
for income, educational attainment, health, etc. But I do believe
that shared ENVIRONMENT plays a role in those things, because siblings
who grow up together share a subculture: neighborhood, schools,
ethnic group, and so on. Different subcultures foster different
attitudes toward things like smoking, doing well in school, etc.
It is this environment – the environment siblings share outside the
home – that affects these outcomes.

But, as I said in the The Nurture Assumption (in a chapter titled
“What Parents Can Do”), there are some ways that the parents
themselves can have an influence. Some example are cooking styles,
religion (WHICH religion the offspring practices in adulthood, but
not how faithfully or fervently he practices it), leisure-time
activities, and of course, occupation. There is no question that,
more often than chance, offspring follow their parents’ occupation.

However, this might not be due to “parental influence” in the
usual sense. Here is an endnote from the 2nd edition of The Nurture
Assumption, which came out in February:

Denrell & Le Mens, 2007, offer a novel theory of social influence that explains how parents influence their children’s choice of a profession or their leisure-time activities. People who are in close contact often have similar attitudes toward such things, not because they influence each others’ attitudes directly, but because their close association has a direct effect on “the activities and objects an individual gets exposed to” (p. 398). For example, let’s say that B is the parent of A. “If B influences the activities A will sample, it is not necessary that A identifies with B, that A wants to comply with B, or that A regards the opinions of B as informative for social influence to occur” (p. 399). Thus, if B is a physician, A will “sample” more activities or objects associated with the medical profession.

Here is the reference:

Denrell, J., & Le Mens, G. (2007). Interdependent sampling and social influence.
Psychological Review, 114, 398-422.

Best,
Judy

Me:

I agree that “shared environment” and “shared family environment” are not conceptually the same. But in practice, almost all variance decompositions treat them as the same, don’t they? When a study reports a c^2 of .05, that includes anything that makes siblings in the same household similar, right?

This echoes an earlier conversation we had about “the effects of the effects of genes” versus “the effects of the effects of family environment.” Standard estimates of h^2 include indirect effects of genes; why shouldn’t standard estimates of c^2 include indirect effects of parenting?

Harris:

I agree that “shared environment” and “shared family environment” are not conceptually the same. But in practice, almost all variance decompositions treat them as the same, don’t they? When a study reports a c^2 of .05, that includes *anything* that makes siblings in the same household similar, right?

Right, it includes anything, either within the home or outside of it,
that siblings who grow up together share and that siblings who grow up
in different households don’t share. Any similarities between reared-
together siblings that aren’t due to genes are attributed to these shared
features of the environment. But most people, when they hear “shared
environment,” think of the home environment. I attribute that .05 to
the subculture (neighborhood, school, SES, and/or ethnic group) shared
by reared-together siblings.

This echoes an earlier conversation we had about “the effects of the effects of genes” versus “the effects of the effects of family environment.” Standard estimates of h^2 include indirect effects of genes; why shouldn’t standard estimates of c^2 include indirect effects of parenting?

Right, they do, if by “indirect effects of parenting” you mean the
parents’ ability to choose their children’s neighborhood, school, etc.

I discussed this in The Nurture Assumption, though I did not call it
an “indirect effect of parenting.”

Best,
Judy