This is a poor statement since you leave out important qualifiers that would give it some beef. you can't just auto-post "sample size too low" whenever you don't see a thousand people, you have to go deeper to justify your opinion. This sample size can be way to low or it can be very high, completely depends on the population and what you want to show and with what confidence. Summary here: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2876926/
But a study like this one here, i.e. if one were to be made, would have plenty of power with a sample size of 40: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC300808/ (also consider the opposite, how many participants it would take before you believe the result, but that kind of study would not be permitted; how many does it take to be saved before you believe it) This study - or the lack thereof, which the paper is about - nicely shows that when there is confidence about the underlying mechanism you feel much less (here: no) need for statistics and an RCT.
The problem I have is that the sample size means the there are reasonably good odds that their a significant genetic difference between the two groups.
I am making no claim as to whether it’s real - I’ve known people who say they have it, and to me seems plausible, but I’ve also known people who are just not 20 any more, and have children, who also claim to have it - it seems plausible in the latter case that people have been sold on a message of “eat healthy and you won’t feel age”.
The problem with weak sample sizes is that it has a significant negative impact whether CFS exists or not:
* if it’s a real disease it could be an incorrect marker - for example let’s say the CFS positive group were all Bostonians living in California. It’s highly likely that a random sample of California natives will have at least some uniform marker. Now people who have the disease, but aren’t from Boston may not have the marker, so are told they’re not sick.
* the disease doesn’t exist, but you’ve got a marker now. People who happen to have the bogus marker are now diagnosed with something that they don’t have. Now they have a diagnosis that prevents an actual diagnosis of real issues.
So I don’t believe it is justifiable to publish a press release on something that effects millions of people, based on a sample that is so small.
The initial small sample size might justify a larger sized sample looking at specifically that marker. Then you also have better science (you have a “this marker indicates cfs” hypothesis), and more statistical robustness.
But a study like this one here, i.e. if one were to be made, would have plenty of power with a sample size of 40: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC300808/ (also consider the opposite, how many participants it would take before you believe the result, but that kind of study would not be permitted; how many does it take to be saved before you believe it) This study - or the lack thereof, which the paper is about - nicely shows that when there is confidence about the underlying mechanism you feel much less (here: no) need for statistics and an RCT.