Too many people in this thread using stats, to paraphrase Andrew Lang, as a drunk uses a lamp-post, for support rather than illumination.
Once again, they're are little more than a guide even in Tests so in ODI's they're even more prone to bias. Use accordingly or you're wasting your time.
EDIT: For those who poo-poo using individual judgement, you really couldn't be more wrong. Aside from the fact that personal judgement does count as 'data' (qualitative), in doing serious statistical analysis, there are personal judgements which abound before you even perform analyses anyway which will affect the results. Whether the data are normal, which test to use, whether the output passes the smell test, etc. There's an entire branch of statistics (Bayesian) in which, far from being discouraged, its part of the process before every test you run to exercise personal judgement about what the distribution will likely look like or how probabilities will change when extra information is added. The rationale goes that all supposedly objective statistical tests which have been used/are still being used will, by design, produce incorrect results because they don't use personal judgement and should be replaced by their Bayesian equivalents.
It's far from being universally accepted but that's inertia, really. Can tell you now, as someone who's in the job market, the ability to do Bayesian statistics is popping up in just about every job outside of Australia and more within Australia every year. It's the wave of the future, man.
.
