Academic Achievement: Temporal Order and Behavioral Propensity
To establish a risk marker, temporal order is not an important consideration. But if we are interested in understanding causal direction, temporal order is important to establish due to the possibility that deviant children are likely to get bad grades, and the causal direction might be from propensity for violence to grades rather than the other direction. Even in a longitudinal study, it is tricky to truly eliminate the possibility that antisocial propensity influenced grades, rather than the reverse, without careful modeling.
Plomin (1990) commented some time ago that academic achievement is about half genetic. For those who accept this estimate, at least half the temporal order problem is solved. A few studies have used techniques to help elevate confidence in temporal order. Some have used longitudinal data to address the problem (e.g., Herrenkohl et al., 2001). Others have controlled for earlier antisocial behavior. Ellickson and McGuigan (2000) controlled for earlier problem behavior in the form of deviance and drug use. Hyeonsook (2010) controlled for earlier wave aggression (only 6 months earlier). McNulty and Bellair (2003) also controlled for prior wave violence (one year earlier). In all of these, cases the association between academic achievement and violence is still statistically significant and in the predicted direction. Only findings by Kosterman et al. (2001) were inconsistent with the predicted pattern.