The p values don't necessarily imply what you think they do. The p value you end up with is a function of the modeling assumptions you make. For example, if you

The p values don't necessarily imply what you think they do. The p value you end up with is a function of the modeling assumptions you make. For example, if your errors are correlated (which I'm sure they are in these swim meets since many data points can correspond to the same swimmer and many are from the same race), then you need to account for that. It's not uncommon to have a p value like that, then you account for something you forgot and it goes away.