Correlations, Structured Sport, and Play

A recent study about early participation in organized sport is getting some attention (at least on my twitter feed). The study claims to find a link between participation in organized sport in kindergarten and future class room behavior. They first measured the levels of participation in structured physical activities and self-regulation of children in kindergarten (indicated by teacher/parent reported classroom behavior). They then measured these again when these children where fourth graders. What they found was a correlation between those students who participated in sport as kindergarteners and those with better reported classroom behavior in the fourth grade.

These results are quite interesting; and I’m sympathetic to them. It seems like conventional wisdom that a child who at age four participates in organized sport will develop and learn effective ways of controlling his behavior—self-regulating—in ways that will be reflected in positive and healthy behaviors as the child grows into adolescence and beyond. It is one of the common reasons one hears for the value of sport; the character development argument. I think there is a lot of truth in this—as a fan of sport I certainly want it to be true.

I have, however, two worries about accepting the study’s conclusions.

1) Overstates the results

Anyone with a passing familiarity with social science knows to always keep in mind that correlation is not causality. The fact that the Boston Celtics might seem to always lose when the Van Allen radiation belt is in flux is a correlation; there is no causal relation, so I wouldn’t bet on it. {For this admittedly obscure reference check out this Cheers episode: https://www.youtube.com/watch?v=mxodytcYD9s (scan to 3:30)}.

The authors acknowledge this: “the correlational design of this study prevents us from inferring causality.”

The concern with correlation is that something else other than the indicated variables explains the outcomes. That is, is there something else besides the participation in sport that explains the later good behavior? Now social scientists have many tools at their disposal to deal with this kind of problem and to isolate the relationship between the studied variables.  The authors did “control for child sex, body mass index, locomotion, object control, cognitive ability, family configuration, maternal education, and family functioning.” In other words, these other variables do not appear to impact the association found, providing a high confidence that the association is something real. And, unlike the Celtics and the Van Allen belt, there are good reasons to think that physical activity and behavior are connected.

Nevertheless, I do think the study is missing (or at least didn’t pay enough attention to in the published article) the possibility of some kind of self-selection bias.

It might be that the kids that are involved in team sports in kindergarten are already good self-regulators or already have predispositions to be good self-regulators. The kids that aren’t as good self-regulators or lack a strong disposition for such behaviors probably didn’t participate in sport in the first place (or end up not participating for long). In other words, they select themselves out of the sample.

If this is the case (to be clear, I am not asserting that is), then the later good self-regulation might just be following from that predisposition, not from structured sport participation per se. If we are starting with a sample that is already predisposed to be good self-regulators, then it shouldn’t be a surprise to see that they become good self-regulators. And it hasn’t been shown that structured sport participation is doing anything to create that situation.

Though the authors control for many potential confounding variables, it was not clear how they controlled for this potential self-selection. To be fair, they do say they “included baseline measures of physical activity and self-regulation.” My complaint is that they don’t explain those baselines or discuss how they measured them, making it impossible to evaluate these controls.

One factor, however, that might point the way is that the study didn’t find the same level of association between later good behavior and participation in non-physical structured activities (e.g. music lessons or chess). There isn’t an obvious reason to think that children predisposed to be good self-regulators would be more attracted to physical activities, so this suggests there is something about the structured physical activity that is, at the very least, amplifying the self-regulator dispositions.

2) Policy Implications

The authors argue that their findings support policies that encourage and increase participation in structured sport (physical activities) in kindergarten. While I think it is, all things considered, better to have more participation in sport than less, I worry about policies (especially governmental ones) that would push more children in to sport.

This is where my concern about the potential self-selection bias has a practical side. If self-selection explains a significant amount of the link here, then pushing more children into structured sport won’t result in increases in positive behavior later on. Many children would be ‘encouraged’ to participate even though they would rather do something else (like chess, art, etc.). Further, there are possible negative effects of doing this. If the new participators are not good self-regulators and the sport itself, ex hypothesis, doesn’t lead them to be, then they might end up getting into trouble, getting labeled as problem kids, or seeing themselves as failures. Whereas, it might have been the case that they would have flourished if they had engaged in the activity of their preference.

Moreover, even if the worry about self-selection here is overblown, I am still not sure it is a good idea to encourage more participation merely to produce better behaviors later on. The children ought to be encouraged to pursue activities that they find intrinsic joy in doing. I would think any positive effects would diminish if the children do not enjoy the structured physical activity.

I am also concerned that these policy recommendations run the risk of reducing—even more than our culture already has—the opportunities for unstructured play. Such activities may or may not be correlated with better self-regulation, but they are important for other developmental needs (Cognitive Benefits of Play). For example, the capacities for imagination and creativity and the character trait of independence are developed or amplified in free, unstructured play. Certainly, they need structured, organized activities (both physical and non-physical) as well. But the trend of childhood in the US seems to be towards more and more structure at the expense free play.