Spent some time in college studying the use of this to detect election fraud. A really intriguing concept, but then the question becomes: what next? Is the statistical data enough to trigger a deeper investigation?
You are phrasing your question like it might be controversial. Why wouldn't 'statistical data' be enough to trigger a deeper investigation?
Everything in politics is controversial. "Let's feed kids more veggies in school lunches" somehow became socialism.
Where does benfords law apply in elections? Not in vote counts (unless the set of precinct sizes is largely fabricated) or insignificant digits of vote counts.
It is not straight forward but look into the Second-Digit Benford's Law (2BL-test).