Do you have any reason to think brand of deodorant and diet are strongly correlated with Parkinson's? (Especially pre-diagnosis.)
Ideal would be to check large numbers of undiagnosed people, and then see how many of those she "alerted" on developed the disease, but given the generally-low incidence of Parkinson's I suspect this approach would be impractical. Larger sample sizes than 12 would always be nice, of course.
I always wonder if this is a good idea. While getting a false positive is not really a problem, because you're going to do a follow up experiment, what happens to the things we miss? If you do an experiment that doesn't really have a large enough sample size, or comes from a biased sample (because it's really an offshoot of a different experiment) and you decided that there is no effect, does it stop others from researching that effect? I suppose since we don't tend to publish negative results maybe it doesn't matter, but it's always something that has niggled at me.
The trade-off between Type I and Type II error is an inherent problem in research. But false positives are most certainly a problem, too. Just look at the issues psychology and biomedicine have been grappling with in terms of replication. Whole careers were wasted based on what seem now like false positives.
I can't tell if you're being sarcastic or not... But the "research" is literally just concluding "Hey, there may be a simple way to test for this incredibly hard-to-diagnose disease".
Well, if diet or deodorant causes Parkinson's, that's absolutely meaningful. ;)
The methodology should have been mentioned more in the article, and should be scrutinized, but that doesn't mean it's worthless if she truly diagnosed these people after a (single?) blind experiment.
While it's true this is mostly justification for further investigation, correctly categorizing 12/12 people into 2 categories actually has a p-val of .000244141 = (1/(2^12)), which would easily allow you to reject the null hypothesis of random categorization. The stronger the effect, the fewer samples you need.
We consider n=12 generally underpowered only because many real-world effects are way weaker than the ability this woman demonstrated.
What makes this result meaningless? The probability of her guessing all 12 correctly at random is: 0.0002 (i.e., approximately 0.5^12). So, it is far more statistically significant than many published results.
Meaningless to draw large scale conclusions on. It's a "This is something we should look more closely at" not a "Send this person around the country STAT"
Ideal would be to check large numbers of undiagnosed people, and then see how many of those she "alerted" on developed the disease, but given the generally-low incidence of Parkinson's I suspect this approach would be impractical. Larger sample sizes than 12 would always be nice, of course.