So unfortunately I didn’t bookmark that particular source, but the estimates can range fairly significantly. They’re sensitive to your technique and your definition of a spoiler. For example, this article calculates both higher and lower probabilities of a spoiler. I don’t think it’s good for much more than saying that, all else being equal, RCV has fewer spoilers than FPTP (choose one). Contrast that with approval, where spoilers simply don’t exist, and approval clearly takes the cake in that category.
Where did you get the 14% figure?
So unfortunately I didn’t bookmark that particular source, but the estimates can range fairly significantly. They’re sensitive to your technique and your definition of a spoiler. For example, this article calculates both higher and lower probabilities of a spoiler. I don’t think it’s good for much more than saying that, all else being equal, RCV has fewer spoilers than FPTP (choose one). Contrast that with approval, where spoilers simply don’t exist, and approval clearly takes the cake in that category.