ABSTRACT

Interactions are fairly common in the empirical literature on delinquency. In addition to explanation and interpretation, two other types of outcome may result from the introduction of a third variable into a two-variable relation. The original relation may remain essentially unchanged in the partial tables formed according to the values of the third variable, or it may be stronger for some values and weaker for others. These outcomes are called, respectively, internal replication and interaction. McCord and McCord find many pronounced interactions, some of them excellent examples of situational intensification. Like the McCords, the Gluecks find inconsistent results in different parts of their sample. Although, like many sociologists, readers have used conditional relation, interaction, and specification more or less interchangeably, the terms do have distinct meanings. The point is that interactions such as this make statements about the relative importance of variables exceedingly hazardous.