Researchers in these journals have not impressed me with their rigor (coming from another discipline I was appalled at what got through the referring process, although I am sure it has gotten better over time). Maybe those with running problems run on softer surface because of those problems and those who don't run on the hard surfaces. Same statistics, different inferences (e.g., despite having a higher injury propensity, those using softer surfaces do not have more injuries than their counterparts less prone that use hard surfaces....). Here the number of respondents is high but they are pretty heterogeneous in ways we do not observe as readers.
My case is such that if I do too much of my running on hard surfaces I have problems especially with my feet and that makes it difficult to run too much in the winter because everything is hard here for several months unless you are running in the snow. 'But anecdotes do not data make.'