Children of divorce are more likely to eventually get divorced themselves. But why? Earlier behavioral genetic work concluded that, contrary to popular platitudes, the transmission mechanism is heredity, not environment. As Judith Harris put it:

A twin study of 1500 pairs of adult identical and fraternal twins analyzed their and their parent’s marital histories… The analysis churned out that: about ½ of the variation could be attributed to genes shared with twins or parents. The other ½ was due to environmental causes. But none of the variation could be blamed on the home the twins grew up in. Any similarities could be fully accounted by the genes they share.

A new paper by a bunch of researchers (including Eric Turkheimer) challenges this result using a “children of twins” design. Key idea: You gather data on twins and their children. Then you can see, for example, what happens to their offspring if one twin gets divorced and the other doesn’t.

The main results from this paper are very different from Harris’. While genetics still matters…

The use of the Children of Twins Design revealed that parental divorce (or risk factors specifically associated with parental marital instability within twin families) accounted for approximately 66% of the initial, unadjusted estimate of the intergenerational transmission. Because our design allowed us to control for genetic and unmeasured environmental selection effects (for the twin parents) as well as measured characteristics of both parents, this is perhaps the strongest evidence to date that parental divorce directly causes an increase in offspring divorce.

Four caveats:

1. The effect size is pretty small. When you look identical twin pairs discordant for divorce (i.e., one divorced, one didn’t):

Children of identical cotwins whose parents remained married had a 17.1% risk for divorce, whereas children of identical cotwins whose parents divorced had a 22.0% risk for divorce.

2. The children of twins design can’t really distinguish between pure environmental effects and (unmeasured) genetic effects of the non-twin parent:

As discussed above, the Children of Twins Design does not control for genetic risk from the spouse of twins (Eaves et al., 2005).

In other words, if twins are discordant for divorce, the reason could easily be that the divorced twin married someone genetically predisposed to divorce. This in turn would make their children more likely to get divorced as a result of heredity alone.

3. The approach ignores child-to-parent causation. A difficult kid might cause his parents’ divorce, then destroy his own marriage when he grows up:

Furthermore, characteristics of the offspring that increase the likelihood of divorce in both generations could also explain the relation (reciprocal influences).

4. Finally, this paper did a standard analysis of variance for the original group of twins, and reached the standard conclusion that shared environment didn’t matter. In fact, the point estimate of for shared environmental variance was 0%:

Twin analyses indicated that additive genetic factors accounted for 15% (95% CI [confidence interval] = 5%–19%) of the variation in marital instability… Nonshared environmental factors, environmental influences that make siblings and twins different, accounted for most of the variance (85%, 81%–90%), with a minimal role of environmental factors that equally influenced both twins (0%, 0%–7%).

If you read this paper, you’ll probably be astounded by the level of care and craftsmanship. Still, it seems pretty confusing to call something a “genetically informed” study when it can’t distinguish between true environmental effects and unmeasured genetic effects of non-twin parents. If anyone’s got some additional insight, I’d like to hear it.