Formal Education vs. Religious Belief: Soliciting New Evidence with Multinomial Logit Modeling

This paper exposes a serious shortcoming in the methods by which serial scientists typically analyze the relationship between formal education and religious belief. Those who think about this relationship commonly imagine its producing quite disparate patterns of effect. Two of the most prominent of these are characterized by the contrary images of erosion and fissure. The problem is that conventional statistical techniques are generally ill-suited to distinguishing between these disparate effect patterns, and so I propose a method of logit modeling that is better able to do so. I illustrate and prove the utility of this method using data drawn from the 1988 and 1993 General Social Surveys.