Comments on: Scienceblogs Dot Com Has Stats!!!

By: Btok

Btok — Sat, 12 Dec 2009 16:12:25 +0000

Check out the Copenhagen Document leaks, this knowledge may save your life! Do what you can with your Government Reps in your country to have these infringements on your life eliminated! You will literally be fighting for your and your familyâ??s existence, Click the videos below:

http://www.youtube.com/watch?v=dAqqAnUxACY

http://www.youtube.com/watch?v=94KH-WMZuw0

Request that PM Harper doesn’t sign the Copenhagen Treaty, thereby causing Canadians to lose their Sovereignty and Freedom, email the PM at: pm@pm.gc.ca

Sign the petition to protest the Inaccurate Science measurements that are being used to base the Copenhagen Treaty off of: http://www.gopetition.com/online/32485.html

By: cooper

cooper — Sun, 01 Nov 2009 15:53:26 +0000

I read his blog and saw that last week.

http://www.stat.columbia.edu/~cook/movabletype/mlm/

By: Greg Laden

Greg Laden — Sun, 01 Nov 2009 11:39:56 +0000

I think that a very large number of researchers think p-value is something other than it is, and among quasi experts you have people splitting hairs that drive any reason out of the conversation. During the exchange you link to, the journalist provided a perfectly adequate explanation that works for any member of the public and it was rejected and complexified by the scientist. I think the scientist was misinterpreting what the journalist was saying and assuming that he was conflating p-values with ‘success’ rates of a certain treatment.

Say you want to know what a compression test does on a car, because you are thinking about buying a car. It is perfectly OK to know that a compression test below a certain value means the engine has a high probablity of needing to be expensively repaired or replaced soon, and in the mean time you might not have much power and you might be burning oil. That is not, of course, what a compression test really is, but to exlain the details and have the car buyer understand them is too hard, and the car buyer does not really need to or want to know all this shit, and is not likely to end up really learning anything useful.

A simple statement that “sometimes statistics can show a successful outcome by accident, or random chance. The lower the p-value, the less likely that the particular study is wrong in this way.” is perfectly adequate. More than adequate. And more than close to reality than “the car will be blowing oil and you’ll have to shell out big bucks within months of buying it”

Statistics is not free of a cultural wrapping, and most stuff coming out of the front-hole by people who supposedly know what they are talking about is no better than what comes out of the backhole, either because they don’t really know, or because it is not really in their interest to be plainly understood by non-experts.

That applies generally to expertise, but perhaps to statistics more than a lot of other things. That whole conversation (at Science) is a great example of it being hard to distinguish front-hole from back-hole effluent.

By: Michael Spencer

Michael Spencer — Sun, 01 Nov 2009 06:16:39 +0000

oh, gosh, that’s sweet, especially after reading this ‘WTF’ piece on Science (yes, THAT Science):

http://sciencenow.sciencemag.org/cgi/content/full/2009/1030/1

Sheesh. How can any competent researcher not summarize p-value in 15 words or less??

By: clheiny

clheiny — Sun, 01 Nov 2009 02:37:55 +0000

Doh! I was trying to use the “Go To: Choose a blog” menu. Thanks for the pointer.

By: Irene

Irene — Sun, 01 Nov 2009 02:17:16 +0000

There does seem to be a link in the post in the last sentence.

By: clheiny

clheiny — Sun, 01 Nov 2009 00:37:19 +0000

So, like, is there a link for this new blog? Science Blogs doesn’t seem to know about it by that name…

By: badrescher

badrescher — Sat, 31 Oct 2009 22:41:15 +0000

Old joke, but stats blog??!! Woo Hoo!
Going there now & hoping to find it share-worthy…

By: Greg Laden

Greg Laden — Sat, 31 Oct 2009 22:24:23 +0000

In the end, the wabbit had the last laugh.

By: The Science Pundit

The Science Pundit — Sat, 31 Oct 2009 22:21:55 +0000

A statistician would momentarily grin but be highly offended by this joke. Need I remind you of all the statistical fallacies that you violated with this joke???

Okay, it was funny. 😀 And it’s nice to have somewhere in addition to 538.com to go to for my statistics fix.