Beam me to Planet Gore !
Posted on 03/21/2010 9:37:39 AM PDT by Ernest_at_the_Beach
Statistics is strange....mathematics....usually.
Lies, damned lies and statistics...
I have to dig up an old text I used to teach a small section of statistics.
Was always unconfortable with it.
If a large number of people have come to intuitively understand that after reading decades of hysterical “science” news stories, then this isn’t really that much of a revelation. There are a lot of fields of study that are treated as a science when they’re not.
Brings to memory Bayes Theorem.
Statistics is a well understood mathematic discipline. It’s weaknesses and strengths are well documented. The problem is not the math behind science, it is the persons that are intentionally trying to sell a falsehood.
There are entire books written on how to deal with alpha and beta errors which this article insinuates are not understood.
Easy way to frame this argument. If you were a high school science student and had to use statistical “smoothing” to produce the “right” data, what kind of grade would you get?
If you do 20 tests to the 95% significance, one would be expected to report as true (null hypothesis rejected) even when the null hypothesis is true.
That is why a single test is meaningless, and can be manipulated. When different experimenters repeat the test, and find the same result, you then can have confidence.
Nothing is settled, ever. Even if the value of Pi started reporting differently (due to a change in the nature of the cosmos) then the current value of Pi would be studied, and the change reported.
Science doesn’t tell you what is true. It is a good way of finding out what is not true.
It seems to me that it is less over-reliance on statistics, than misapplication of statistics that is at fault in the state of “climate science”. And this misapplication has two aspects: the much-discussed one involving dishonesty—selecting weather stations about which statistical inferences must be drawn (due to gaps or the need to estimate an urban heat-island effect) in preference for ones with long continuous track records in rural areas, omitting weather stations in colder regions (e.g. the Andes) and making inferences based on “nearby” stations in areas with radically different climate—and one involving an honest conceptual error.
The conceptual error lies behind the “weather is not climate” mantra, that hides the fact that climate IS weather, averaged over longish-time intervals, but I think also hides the mistaken assumption that the variability of weather is random noise of the sort statistical methods are useful for dealing with. In fact, the unpredictability of weather is due to the underlying non-linear dynamics that does not go away when you take time averages.
Truth!
The article is, ironically, misleading. The problem is the misuse or misunderstanding of statistics, not that the scientific method is incorrect or that statistics is incorrect.
Another issue (that I will soon write about in my blog http://libertyphysics.wordpress.com/) is that statistical correlations are very often confused with cause and effect. For example, say a statistically correct study is done showing people in countries who eat more yogurt live longer than people in countries who eat less yogurt. I'm being simplistic, of course.
That doesn't mean that eating yogurt will make you live longer, no matter how correct the statistics. The scientists who do these studies usually know better but the fawning news media does not and reports such findings as if they were a call to action.
Worse, even if the statistics is done correctly, say we discover that, on average, eating salt raises a population's blood pressure, that conclusion doesn't necessarily apply to any given individual member of the population. For example, statistically, men are stronger than women. But I can easily find a woman stronger than the average man just as I can easilly find someone who can eat salt without an increase in blood pressure.
Misuse and misunderstanding is the issue.
We live in a very strange world in so many ways...
|
|
|||
Gods |
Thanks Ernest_at_the_Beach. |
||
|
· Discover · Nat Geographic · Texas AM Anthro News · Yahoo Anthro & Archaeo · Google · · The Archaeology Channel · Excerpt, or Link only? · cgk's list of ping lists · |
|||
AGW is extremely robust with respect to data, all observations confirm it at the 100% confidence level.
Well, if you can pick and choose data, literally anything’s possible. Also see Ancel Keys.
The current trend is to model a dynamic system, and then calibrating the model to historic statistics by using hidden assumptions, (which can be really wrong.) Then the modelers feel confident in using the model to predict the future and to regulate the heck out of you.
Sciencianism. That’s what it’s become. State funded Sciencianism.
Just like the arts. Throw too much money at anything, make the money too available, and you get what you pay for.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.