Most political data in the US relates to the correlations between voting behavior and variables like ethnicity, socioeconomic status, parental politics, age and location. The same data exists substituting religious behavior for voting behavior, and doesn't generate any significant controversy.
The example you give would be analogous in polsci to a study that correlates members of a political party to the likelihood to engage in protests or violent crime based on tenets of a party platform. A study like that would meet with just as much challenge.
Religious sociology is a separate discipline and I actually think the forum does okay in discussing that.
Regarding the thesis, you have a point, and it may be a bad example, but the fundamental texts of a religion have a different character and importance than something as temporary and often mostly symbolic as a party program. Textual analysis is at the core of religious studies and the data points you can acquire from a textual and linguistic analysis are far easier to challenge than the sort of empirical data polsci typically uses.