If you have brothers or sisters, then you already know that an important piece of your childhood experience is out of your control. You cannot control whether your siblings are kind or cruel, generous or stingy, and you certainly cannot control whether your siblings will grow up needing special attention and support.
When children have special needs – ranging from common impairments like ADHD to more severe problems like mental retardation, autism, and blindness – they often take up parental time and family money. Having children with special needs is often stressful for parents, impacting their own mental health and parenting skills. Healthy siblings may also need to chip in more to care for a brother or sister with special needs, taking away time from homework and social activities. Although sometimes having a disabled sibling can promote better coping skills and self-sufficiency among non-disabled individuals, it is likely that the experience can have a negative spillover effect.
Jason Fletcher, Nicole Hair, and Barbara Wolfe try to quantify the effect of having a disabled sibling on academic performance in a new working paper. Direct comparisons between families with and without a disabled child are tricky, however, since these families are likely to have different genetic and environmental exposures that may be difficult to completely control for.
To address this challenge, the authors exploit the fact that the U.S. Panel Study of Income Dynamics follows the same extended families over time (i.e. grandparents were first surveyed in the 1960s, and then their children and grandchildren were kept in the sample). This allows them to identify not only sibling pairs but also cousin pairs. Cousins are used as “controls” who account for half of the child’s genetic background, but who are not likely to be directly affected by their disabled cousins.
The main findings from fixed effects models are displayed in the table below. Having a sibling with a developmental disability is associated with a lower score of about 4 points on a reading test (the Woodcock Johnson) and a 6 point lower score on a letter-word identification test (a reduction of about ¼ to 1/3 of a standard deviation). Smaller, but significant, effects were also seen for children who had peers with externalizing behaviors (ADHD is a classic example of this). In additional analysis, the authors look at differences by subgroups (such as girls versus boys, and older versus younger siblings), they find that the effects are much larger for females on the language skills tests, but no significant difference by sibling order.
What does this mean for policy?
The last line of the paper sums up what I think is the most important take-home finding: “These spillover effects are not currently included as a part of cost-benefit calculations when evaluating programs that aim to improve children’s health, but our results combined with others in the literature suggest current calculations of the effects of health improving programs are (potentially severely) underestimated.” In other words, interventions to improve the health and wellbeing of disabled children also benefits brothers and sisters.
I have two concluding thoughts of my own. The first is methodological. This study does an admirable job of addressing bias due to family, but I think that it’s difficult to entirely “purge” the influence of family disadvantage (not caused by the child’s disability) from estimates of the effect of the disability on siblings. In future work (probably with larger samples) it would be useful to compare children that experience random accidents that cause health impairments with children that have congenital conditions. It would also be helpful to use data that rely on doctor reports of health conditions, since bias in parental reporting of child health is difficult to control for in this kind of study.
The second thought pertains to policy. The authors posit that there are several mechanisms at play, including “behavioral spillovers” (as when a younger sibling imitates the older sibling’s behavior problems), income effects (the effect of spending money on treating the child), and substitution effects due to parents’ investing more time in one child than the other. Of these pathways, the second – income effects – is the one that most directly is affected by policy. Children’s health insurance and disability payments are obvious examples. We need to study family’s that are differentially exposed to these programs to see how much of a buffer they provide in protecting against the bad outcomes (as my post from a few week ago discussed in relation to the EITC). Not all forms of disadvantage can be combatted, but the safety net probably has a role to play here.