However, I would like to quickly contextualize the issue of causality in our working paper. Writing-up a quantitative study aimed at rendering a broad and reliable overview of the field we were not able to do full justice to the nuance of the explanatory models used in the articles we researched. In the open coding phase of our analysis we found fourteen different causal models in the articles we analyzed. In addition, about ten percent of the articles did not use causal imagery and another two or three percent presented findings affirming a null hypothesis. When we worked our codes into more abstract codes suitable for quantitative analysis we simply recoded them according to whether religious processes or social processes were the primary independent variable (or whether no primary independent variable could be isolated). This serves the purposes of our working paper well, but covers over some of the nuance in the data (i.e., the articles) we are describing. Many of these articles provided complex analyses that go beyond the straightforward causal terms we use, and pursue novel ways to conceptualize the mutual constitution of religious and social processes.