Quote:
|
Originally Posted by aceventura3
Is that level of error acceptable in all forms of science? Of course not, there is not always doubt.
In your field how did they determine that <5% statistically significant rather than 0? Probably because there are unknown variables that can not be controlled for in every circumstance. As this concept applies to climate models, there are also variables that cannot be controlled for and 90% certainty may be as good as it can get, I understand that. But, the earth may be like one of those patients that doesn't respond the way normal modeling would predict. And we may still be in the relative dark-ages when it comes to understanding climate change on this planet.
|
There is always doubt in all science, or more generally, in the universe. It is
mathematically impossible to have an alpha level equal to zero. The level of alpha you consider depends on what you are looking for and the costs an benefits of false-negatives and false-positives (among other issues). Wikipedia provides a reasonable summary of the issues surrounding statistic significance.
http://en.wikipedia.org/wiki/Statistical_significance
Again, I suspect that the primary sources have found effects with alphas <10%. I also expect that the sizes of the effects they have found are large enough to consider important. I am interested in the primary sources, but I don't have the time to go searching for them. It seems unreasonable to me to suggest that because there is always doubt and we don't know everything, we should ignore what we do know.