Vibram settles lawsuit over FiveFingers health claims

@scott

"Marko, I think you misinterpret the pigeon experiment. "

Nope!

Sample of one, or sample of few (with few being too small) all the exactly same thing.

However I do agree with your assertion that in some circumstances the data is marginally BETTER analyzed by the pigeons than by some people. :-)

From what I remember, the stated health benefits were kind of nebulous anyway. Not to mention Vibram had a section on their website on how to slowly work into using these shoes, and I think there were even instructions in the box.

Lets face it, if some overweight, never ran before, out of shape person straps on a pair of these and goes for long runs on cement they are going to have problems.

Whatever happened to common sense and you are responsible for your own actions? Oh, wait a minute… there's that 25% for the lawyers plus costs.

I've posted plenty in the past about these and other minimalist shoes and the need to very slowing work into shape, and the need to carefully consider the running surface.

Do you think I can sue the minimalist shoe manufacturers because mine do not protect me from cactus thorns?

@Marko

To put it kindly, I'm guessing that you're not a statistician, right? There's a huge, huge gulf between a sample of 1 (where you essentially have no information about what's going on) and a larger sample size. Even 2 data points is much better than 1.

I'll stick with the pigeons. Look where they live–those guys are survivors. :)

@scott

Again scott read what you said that I was responding to. You are incorrect that there is "no" information from one data sample. There is no information about the higher moments of the probability distribution (and therefore the statistical significance), but there is information about the mean. From a Bayesian perspective you can already begin to construct predictions from one sample. For example, the best prediction in that case would be the same thing would happen again. Since this is what the straw men you are talking about are supposed to be guilty of, then from the statistical perspective they ARE making the right decision. Their guilt lies not in the difference between one sample and two, but in the small number of samples they are using to make predictions about future outcomes. There is also the issue of confusing the priors. Probably making a prediction from one sample if the prior is "me" will work fine. But it you then draw the conclusion that the prediction will work if the prior is "everyone", or even just "runners" without thoroughly sampling "runners" you would be off base.

In your earlier post you seemed be under the impression that there is some kind of major conceptual difference between one and "too few". There is no "huge gulf" from a prediction strategy perspective, for pigeons or people. Ironically, hyperbole is what the shoe company is supposed to be guilty of.

We now return you to your regularly scheduled (ever so slightly) on-point thread.