Comment by adamzerner

The assumption is that the researchers accounted for any sort of selection bias in who participates in the study. But that's an interesting point; maybe that assumption needs to be questioned further.