Worth noting that the correct response for Starmer is likely not statistically distinguishable from what we'd expect were respondents to answer at random (11% vs 1/11 ≈ 9%).

Proves that something only really cuts through once you've repeated it until you're red in the face.