Friday, June 27, 2008

Domestic Violence and Social Externalities

Another interesting paper by Scott Carrell (UCDavis) and Mark Hoestra (Pitt). The methodology is familiar if you've seen their other papers - peer effects with careful identification using quasi-random assignment. The title of the paper is "Externalities in the Classroom: How Domestic Violence Harms Everyone's Kids." Here's the abstract.
It is estimated that between ten and twenty percent of children in the United States are exposed to domestic violence annually. While much is known about the impact of domestic violence and other family problems on children within the home, little is known regarding the extent to which these problems spill over children outside the family. The widespread perception among parents and school officials is that these externalities are significant, though measuring them is difficult due to data and methodological limitations. We estimate the negative spillovers caused by children from troubled families by exploiting a unique data set in which children’s school records are matched to domestic violence cases filed by their parent. To overcome selection bias, we identify the effects using the idiosyncratic variation in peers from troubled families within the same school and grade over time. We find that children from troubled families significantly decrease their peers’ reading and math test scores and significantly increase misbehavior by others in the classroom. The effects are heterogeneous across income, race, and gender and appear to work primarily through troubled boys. The results are robust to within-sibling differences and we find no evidence that non-random selection is driving the results. The presence of these externalities suggests that to the extent that education policy increases a group’s exposure to children from troubled families, student performance will be affected in a negative way. Furthermore, the results are also relevant for social policy in that they provide for a more complete accounting of the social costs of family conflict.

4 comments:

J said...

Very interesting. A friend and I were just discussing something similar last night. He works at a therapeutic daycare that caters almost exclusively to low-income families trying to work their way back to stability - be that economic, emotional, medical, etc. The vast majority of the kids are from homes that have experienced or are experiencing some form of domestic abuse. He was commenting on both the cumulative effect their behavior has on each other and how the kids from the worst family situations often drive the behavior of the others and that by targeting them they can often control the whole group. However, when they do programs with kids from other institutions whose students tend to always be from more stable home situations the behavior of their own kids improves dramatically. Almost a reverse of the effect (although a limited duration one) that the authors seem to be describing, I wonder what they would say about it?

scott cunningham said...

So you're saying that your friend observes both (a) the least well off kids having behavioral spillovers on the rest of the class, but also (b) when classes are mixed, the students who aren't from broken homes appear to cause the least well-off students to behave better.

If that's what you're saying, then I can believe both of those. Clearly there's some potential positive or negative externality associated with the peer environment. It's possible that being around well-behaved children can make one behave better, but it could also be that it goes the other way and the bad behaving children cause disruptions overall in the class. Which dominates may be an empirical question, though another reader named matt on here works with one of the co-authors on this particular study I linked to, and is also working on peer effects in non-cognitive outcomes (like "leadership" aptitude), and I bet he would have an idea. Matt?

J said...

Yes, not very clearly, but I was saying both. The interactions with the more stable children are usually temporary and short lived so if they were more long term I wonder which way the behavior would tip, or another way of saying it I guess, which peer effect would prevail - a move towards disruptive behavior or a move towards stability?

scott cunningham said...

I could definitely see that, and I want to say I've seen studies that support that possibility - that good kids spill over on bad kids. This paper (which I've not yet read) is finding the bad kids' influence dominating, and that kind of fits what Ed Lazear said in his paper on classrooms, which he argued have public good aspects to them. The disruptions caused by bad kids are basically like public goods - they are nonrival and nonexclusive, which unfortunately means that we all experience them when they are produced, and no one can be excluded from having them on them. It's very intuitive to imagine that good kids are rubbing off on bad kids, too, and so one does wonder if this paper is generalizable to all situations, all ratios of abused children, etc. Also, this paper is just about domestic violence, which may mean that it is unique among negative outcomes.