If the second group performed “far better” than the first group then this isn’t regression to the mean, is it? I would expect the gap to be much less or even eliminated but for the Plus C Bow group to do much better there should be something else at play, right?
Yeah, plus the selection process was weird. Any gains attributed to the plus-C bow group just throws question to the initial rankings.
They should have done a crossover trial, where group A is a random selection of archers initially getting no C bows, and then later getting them, and vice versa for group B. Paired t-tests, y’all!
Yeah, but the meme is a joke about how you can prove anything with regression to the mean. If all archers are equally good and we test them we would get varying results. If we then split them on performance and perform an intervention and test again then the “poor performing” group will do much better. Because it’s just random noise.
If the second group performed “far better” than the first group then this isn’t regression to the mean, is it? I would expect the gap to be much less or even eliminated but for the Plus C Bow group to do much better there should be something else at play, right?
Yeah, plus the selection process was weird. Any gains attributed to the plus-C bow group just throws question to the initial rankings.
They should have done a crossover trial, where group A is a random selection of archers initially getting no C bows, and then later getting them, and vice versa for group B. Paired t-tests, y’all!
Yeah, but the meme is a joke about how you can prove anything with regression to the mean. If all archers are equally good and we test them we would get varying results. If we then split them on performance and perform an intervention and test again then the “poor performing” group will do much better. Because it’s just random noise.
Oh I see, thanks