Can an Enemy Be a Child's Friend?

New York Times article "Can an Enemy Be a Child's Friend?" (May 17th, 2010) states that teenagers who face their classmate's enmity and reciprocate it are better socially adjusted. Throughout, the article's context suggests that such behavior could cause individuals to become better adjusted. Corrections sees heterogeneity rather than causality.

In a series of recent experiments, a group of psychologists at the University of California, Los Angeles, recorded mutual dislike among 2,003 middle school students. Unlike previous studies on the same topic, these researchers also compared children who reciprocated a fellow classmate’s dislike with those who did not. Students who were not named at all on anyone’s blacklist were excluded from this analysis.

This comparison found that the girls who returned classmates’ hostility scored significantly higher on peers’ and teachers’ ratings of social competence. They were more popular and widely admired. The boys who did the same scored highly on teachers’ ratings of classroom behavior.

Corrections is perpetually pointing out the difference between correlation and causation. In a parallel argument, we note that these correlations are scarcely even suggestive of causation when we are dealing with optimizing agents. Imagine that an individual faces enmity from a classmate. Their ability to respond successfully by reflecting enmity back is heterogeneous in the population--they vary in their effectiveness. Furthermore, socially competent people have a greater degree of efficacy in their ability to reflect enmity. Were we running the analysis the Times runs, we would display the following relationship (click to enlarge), which clearly displays a positive relationship between an individual's enmity reflection decision and their social competence:



However, let us imagine that social competence and effectiveness at reflecting enmity have the positive relationship that we graphically display below (click to enlarge):



Then we might say that all individuals who are effective at reflecting enmity do so, and all those that do not, don't. We have in truth the relationship (click to enlarge):



Ordinarily, when we point out that correlation is not causation, we note that a third factor is likely causing both. Here, we posit causality: enmity doesn't cause social competence--social competence causes enmity.  Further, in this causal relationship, boosting enmity among those not displaying it would clearly hurt them (we display this in the last diagram). Students who don't display enmity don't do so for a reason.