|
The problem with being so stringent in the level of confidence is that nature itself doesn't work in that way. Scientists maybe missing discoveries simply because they don't meet the required levels of statistical analysis. Nature can be very quirky and never plays by the book. Things can crop up unexpectedly and then disappear off the charts. All the experiments in the world may not show another result.
However, in playing it safe with such stringent conditions of acceptability, it does negate to some extent the possibility of false positives and the like. But, how far do you have to go, especially when even higher levels of confidence have been rejected in the past. Where/when do you call a discovery 100% certain??. Especially in particle physics, where a spray of particles at the detector of an accelerator can number in the billions, what are your chances of finding anything even at 5 sigma??!!. Then again you may find something right off the cuff. Or repeat your experiments and not find it again. If you follow chaos theory, it would be almost impossible to repeat your experiments exactly, in any case. So how do you rate the results then??. The statistics would be meaningless.
You can see why they use such a high confidence level for the confirmation of a discovery, but it's still not the ideal way of going about it. What's needed is a symposium on this matter and a set of guidelines ruled upon that can be used across all disciplines. That way, at least we could be more confident of smoothing things out a little better than what is being experienced at present.
|