this brief chapter describes the methodology used to locate and summarize studies of the effectiveness of attempts to prevent or reduce problem behavior. It provides an overview of the studies included in the next two chapters. The chapter begins with a discussion of the importance of experimental and quasi-experimental studies both for refining theories about the causes of problem behavior and for learning how to prevent it.

Correlation versus Causality

Although the risk factors summarized in earlier chapters are correlated with current and future problem behavior, most have not been demonstrated to cause problem behavior. Even longitudinal studies fall short of the standards necessary to establish causal relations: covariation, temporal priority, and nonspuriousness. Studies summarized in the two preceding chapters always satisfy the first, often the second, but almost never the third requirement.

Correlation does not imply causation. A correlation between an individual risk factor (e.g., academic performance, beliefs in the validity of rules) or a characteristic of the school environment (e.g., collegial relations, fairness, and clarity of rules) and problem behavior cannot be interpreted causally because the problem behavior may have preceded the risk factor in time or because the correlation may arise simply because both the risk factor and the problem behavior are caused by a third factor. Even if the correlation arises from a causal association, the direction of the causation may not be as expected. The direction of associations established in correlational research is often ambiguous.