Well the idea in the earlier comment is that a 2x rep difference isn't very much to be the difference between "low" and "high". It's not disputing that you can get a difference, but saying the study didn't try very hard to probe it.