It assumes that the ratio of true-to-false statements repeated is the same regardless of political correctness. (If a true PC statement is four times more likely to be repeated than a false PC statement, then a true PI statement is four times more likely to be repeated than a false PI statement.) I'm not sure that's true.

But this does give us the conditions required to assume a PI statement is more likely to be true than a PC statement, which is valuable.