The Storm Warning That Wasn't
from the whoops dept
The folks over at the National Weather Service were a bit surprised yesterday when they saw that they had put out an “impending blizzard warning” when they were all sitting around marveling out how nice the weather actually was. Turns out that someone in another office had sent out a test message that accidentally went live. Then, thanks to our automated world, the winter storm warning started appearing all over the internet on weather pages that simply take information from the National Weather Service and repost it. The only thing I can think of is that this is the first time I’ve actually seen the weather people admit that they were wrong.
Comments on “The Storm Warning That Wasn't”
Storm Warning...
Cheap shot!! Accurate, but none the less cheap ;->
A math problem most doctors got wrong:
from an article in the Financial Times:
About 1 per cent of women have breast cancer and a cancer screening method can detect 80 per cent of genuine cancers but also has a false-alarm rate of 10 per cent. What is the probability that women producing a positive result really have breast cancer? Most doctors reckoned it was at least 70 per cent.
The correct answer, found by only two doctors, is just 8 per cent.
Re: A math problem most doctors got wrong:
7.5%, I think.
Picture an initial sample of 1000 women. 10 have breast cancer. Out of the 990 who do not, the 10% false alarm rate means the test will show 99 positives. Out of the 10 women who have breast cancer, the 80% detection rate means the test will show 8 positives. So, after testing 1000 women the test shows 107 positives. Only 10 women have breast cancer, but only 8 of the positive test patients have breast cancer. 8 out of 107 is 7.5% (or, more precisely, but still not exact: 7.48%).
Nit-picking, I know. But as long as you’re talking about getting a math problem wrong…
Re: Re: A math problem most doctors got wrong:
I have no idea what this has to do with the particular post, but it looks like a standard Simpson’s paradox situation, which are fairly common, but always seem to amaze those who never took a statistics class.
Classic errors and fallacies
They have no excuse for getting that problem wrong, it is a fairly high profile piece of mandatory teaching in medical school. If you get it wrong in a oral exam, you can expect to repeat the the course.
That said, a senior professor very publicly got a related problem wrong as an expert witness in the UK, and more than a dozen women were wrongly jailed for their cot death babies…..
but then the same problem is taught as the cause of a classic fallacy in law school, so the defence lawyers and presiding judges have something to explain too…
The false storm warning is a different classic error : not checking your sources. Something which matters ever more when sites crib from each other, the net can become an engine of misinformation