Sep 3, 2018 · 1 min read
Yup I get it. But this is just a control for subdistricts’ characteristics since my research is based on observational data, not survey. It it not the intention of this study to measure such association (the Muslim values), because my research question is more about strategic voting in a runoff-system. I control for several variables just to capture confounders that may be related to the error term of my regression. This is not a text-analysis research like you suggested (that plays with “keywords”). Thus, the output of the proportion of Muslim cannot answer your curiosity. Still, that is an interesting study to be made. But we need another research.
