[LINK] IPA, astroturfing and fantsy themes

[Richard Chirgwin]

On 21/02/12 2:05 PM, Jan Whitaker wrote:
> At 01:45 PM 21/02/2012, Paul Bolger wrote:
>> Interestingly Dr Winston hasn't been totally saintly in the scientific
>> method area either.
>>
>> http://www.abc.net.au/radionational/programs/scienceshow/bad-science---ben-goldacre-part-2/3179824
>
> Thanks. I think he's just a big name that is on such a high demand
> that he may not pay attention as he should. Was the study by Oxford
> University or wasn't it? Was it a student project or a real research
> study with proper controls? Goldacre also says this:
>
> "Ben Goldacre:  And then on top of that obviously every year there's
> something called regression to the mean, just by the natural play of
> chance...if you've just rolled a double six, it's actually very
> likely that your next roll will be lower than a double six. If you've
> just rolled a double one, it's actually very likely that your next
> roll will be higher than a double one, just from the play of chance.
> So if you go in and do an intervention like this, handing out the
> pills in an area which has just had particularly poor exam results,
> then you might expect that they'd do better anyway. "
>
> That's just wrong. There is always an equal chance of any dice
> combination. Always. He would be terrible at coin toss guessing.
> And I'm not sure regression to the mean is anything about chance of
> rolling any particular number. But I'm not going to waste time doing
> the research on what it is.
Oops. Yes, there is an equal chance of any dice combination - Ben has 
expressed himself badly here - because there is only one double-six 
combination (ie combination adding up to 12), and only one combination 
adding up to 2 (double one).

2 - 1-1
3 - 1-2 and 2-1
4 - 1-3, 2-2 and 3-1
5 - 1-4, 2-3, 3-2 and 4-1
6 - 1-5, 2-4, 3-3, 4-2 and 5-1
7 - 1-6, 2-5, 3-4, 4-3, 5-2 and 6-1
8 - 2-6, 3-5, 4-4, 5-3 and 6-2
9 - 3-6, 4-5, 5-4 and 6-3
10 - 4-6, 5-5 and 6-4
11 - 5-6 and 6-5
12 - 6-6

For any roll of two dice, a total of 7 is the most probable and totals 
of 2 or 12 the least probable. Goldacre is right statistically, but like 
I said, he wrote it badly.
>
>> So then you get into this childish unbearable circular stuff
>> where...in their own press release they called it a trial, in all of
>> their interviews they called it a trial, in the company's press
>> release they called it a trial.
> And that is what happens in Australian govt programs All. The. Time.
> Just look at the current crop of 'ehealth record trials'. No they're
> not. They are pilot roll-outs, with very little 'pilot' either.
>
> Maybe we are all a bit too loose with the terminology, but if major
> decisions are going to be based on them, then at least get the
> representation right.
>
> That was 2008. I wonder what's become of the whole fiasco.
Still happening and still something of a fiasco.

RC
>
> Jan
> disclaimer: I take 1000mg of fish oil daily, for joint health, not my
> brain, which seems to be losing its functions, at least a little
>
>
>
>
>
>
>
> Melbourne, Victoria, Australia
> jwhit at janwhitaker.com
> blog: http://janwhitaker.com/jansblog/
> business: http://www.janwhitaker.com
>
> Our truest response to the irrationality of the world is to paint or
> sing or write, for only in such response do we find truth.
> ~Madeline L'Engle, writer
>
> _ __________________ _
> _______________________________________________
> Link mailing list
> Link at mailman.anu.edu.au
> http://mailman.anu.edu.au/mailman/listinfo/link
>