Sorry; initially, I didn’t notice that this post was linked to a specific question.
This question is getting at something more commonly seen in Flaw Questions – The biased sample. Imagine you were conducting a scientific experiment to determine whether or not stretching prevented injuries. What would you do? You’d probably take a bunch of joggers and split them into two groups, and you’d want those groups to be as equal as possible in terms of age, sex, physical condition, etc. then you’d have one group stretch, and you’d tell the other group not to stretch, and you’d see if there was a difference in the number of injuries you saw in the groups.
That’s not what happened here; instead, we had a self-selecting sample. Runners decide on their own whether or not they want to stretch.me that sort of sample is always prone to some sort of bias. So now, you don’t have two equal groups, one of which stretches and the other doesn’t; you have two groups:
1) Stretches *and is prone to injury*
2) Does not stretch *and is not prone to injury*
If stretching did not prevent injury (the conclusion of the passage), wouldn’t you expect more members of group 1 to get injured? They’re the injury-prone group. But that’s not what happened. Even though they’re more injury-prone, they had the exact same rate of injuries as group 2. That suggests that in fact, stretching DOES help prevent injury – by stretching, the injury prone group keeps their rate of injury to the same rate as a group who are NOT injury-prone.
(Above analysis is on the assumption that (D) is true).