ABSTRACT

In some fields of science, it is possible to do small and elegant laboratory studies of causal relationships, where single variables are isolated experimentally. In education, however, we can almost never hold things constant experimentally. Instead, we have to control them statistically in order to tease out the contingent relationships.