True, though when you've got a sample that large, you've got to consider the probability of making a Type I error. It would be interesting to see the study itself, especially a power analysis and the number of comparisons run on the data, etc. I would not be suprised to see a relatively...