• Laid Off Dads, Laid Off Moms, and Child Abuse

    What happens to children when their parents lose their jobs? My intuition is that job losses stress families and put children at increased risk of child abuse. But Jason Lindo, Jessamyn Schaller, and Benjamin Hansen have surprising data saying that the answer is more interesting than that.

    Previous data suggested that my intuition was wrong and that hard economic times have little influence on child abuse. For example,

    Despite the onset of our most recent recession and the declines in family income that followed, victimization rates have actually fallen slightly from 2007 to 2010 (from 9.6 per 1,000 children to 9.2 per 1,000 children)…

    The graph below shows the association between unemployment in California (the broken lines) and four measures of officially-reported child abuse (the solid lines).


    Notice, for example, how unemployment rockets up starting in 2007, but child abuse rates do not. (This lack of association may have occurred in part because the recession reduced the ability of the state to find and report child abuse. However, I don’t think this was a threat to Lindo’s analyses.)

    The authors took a fresh look at the issue of unemployment and child abuse using California county data from 1996 to 2009. The problem with looking at the simple correlation between unemployment and child abuse is that many other factors, most of them unmeasured, affect both variables. So what Lindo and colleagues did was to look at county-specific fluctuations in unemployment and child abuse. A county-specific fluctuation is how, for example, 2005 unemployment in Los Angeles County deviated from both historical LA County unemployment levels and 2005 California-wide levels. Lindo and colleagues reasoned that local deviations for unemployment (or child abuse) in a specific county and year would be relatively uncontaminated by trends in other social factors. And therefore if they found that local upticks in unemployment corresponded to local upticks in child abuse, this would be strong evidence that increased unemployment causes child abuse.

    In the graph below, the vertical axis is the percentage increase in county-specific child abuse associated with a 0.1% increase in county-specific layoffs.


    The pink bar on the left represents a 0.68% increase in child abuse associated with layoffs of all workers. This increase was so small that it was not statistically different from zero. The blue and green bars, however, are the effects of layoffs of female and male workers respectively. They show that

    a 0.1% increase in the fraction of working-age males being laid off leads to a 3.09% increase in the number of reports of abuse. In contrast, the point estimates imply that 0.1% increase in the fraction of working-age females being laid off leads to a 3.27% reduction in the number of reports of abuse. [Emphasis added, both effects statistically significant.]

    So layoffs have large and opposite effects on child abuse, depending on whether men or women are laid off.

    Lindo and colleagues have a plausible explanation for this. Female layoffs increase the proportion of time children are with mom. Male layoffs increase the proportion they are with dad. Both genders abuse children, but a man is about three times more likely to abuse on a per hour basis than a woman.

    So much for my intuitions: I thought the story was about stress, but it is also about time and gender. This study tells us to look at how the effect of a treatment or a risk factor varies depending on the person it is applied to. This is why research study populations should be as diverse as is practicable. When it makes sense, we should include both adults and children, or men and women, and carry out the analyses that look for heterogeneity of effects. This is the motivation behind personalized medicine: finding out what works for whom.

    The policy upshot of some previous research on unemployment and child abuse was that although reducing unemployment is a very good thing, it probably won’t do much to reduce child abuse. This study doesn’t change that conclusion. But it does suggest that we need to do a better job of socializing men to be authoritative, effective, and non-abusive parents.


    • Newly at-home mothers probably attend noon nursery school pick-up and make connections, newly at-home fathers feel like fish out of water. One comes home with less stress, the other with more.

    • This is what we noted anecdotally in the late 60’s early 70’s when many were trying to claim men and women were nearly interchangeable.

    • One of the defining events of my life was realizing that my little town was an abusive hell when I was growing up only because almost everyone had lost their livelihood. I now see it as the human outcome of an economic downturn.
      Glad to see the data supports my hypothesis. Lindo might find the gender differences decrease over time. I did. A long and systemic bout of economic collapse might be different.
      Colin Turnbull’s oft criticized book, The Mountain People, struck a chord in that it described the cultural breakdown of a society driven to the brink by economic collapse. It helped me make sense of my hometown.

    • The primary story for me was a positive one after looking at the first set of graphs – what are the causes of the drastic reductions in all 4 measures of child abuse since 1995/2000?

    • Another factor that comes into play in abuse is the trade off between having more time to drink and less money to buy alcohol.

    • Authoritative?

    • It could still be about stress.

      When you mentioned stress, I thought it could go both ways: being unemployed is stressful (what am I going to do with my life? What if I can’t afford necessities?), but so is working (getting up in the morning, spending all day making someone else rich while getting barked orders at, worrying about getting out quickly enough to be home when the kids are there, etc.).

      What if men and women experience stress in different ways when it comes to unemployment? If a woman loses employment, is she more or less likely to have a partner, parents, or friends to support her financially than a man who loses employment? Women generally make less money than men, so would the smaller reduction in income reduce the stress of unemployment? Women become parents at a younger age than men do, and maybe young people are more carefree about losing a job (i.e., this is really just an age effect)?

      Still, didn’t know that the long-term trend is of a drastic reduction in child abuse, generally. Good to know.

      • Alex, I do believe that stress is an important factor. What I was trying to convey was that I had completely missed the issue of differential exposure to male and female caregivers.

    • I wonder whether the child’s sex also matters.

      • Girls are more likely to be sexually abused than boys. I am sure that is not a surprise.

        • Bill, there is a lot of undocumented sexual abuse especially among the male gender. Sexual abuse also includes non contact sexual abuse. If we were to account for undocumented sexual abuse and only include contact abuse how much of a leveling out do you presume we would see between girls and boys?