?

Log in

No account? Create an account
Giving evidence, Sir Roy defended the calculations he used to arrive… - Off in the distance
browse
my journal
links
May 2016
 

The Bellinghman
Date: 2005-07-15 16:14
Subject: (no subject)
Security: Public
Giving evidence, Sir Roy defended the calculations he used to arrive at the 73 million to one figure. (BBC report here)

No, he still doesn't get it. He gave evidence saying that if two or more children died in the same family at different times, the probability should be calculated as if they were totally independent.

And therefore, since the chance for the death of one child is in the thousands, the chances for two is in the millions. Ergo, if it did happen, the cause could only be murder.

Ah, hello? Anyone there? You start from an assumption that the causes of two deaths are unconnected. You then get odds saying that two deaths happening is exceedingly improbable. You then get the two deaths, proving that it can happen. So you then say that the causes of the two deaths are connected (reasonable) -- and that the cause must therefore be murder.

Duh! There is no connection, and therefore the connection must be murder? Could he really not see the logical fallacy in that?

Was it really not possible that other connections might exist? Like shared genetics, giving shared predispositions in a shared environment?

And yet he still defends his evidence.
Post A Comment | 3 Comments | | Link






The Uitlander
User: uitlander
Date: 2005-07-15 15:41 (UTC)
Subject: (no subject)
He didn't pass O-Level maths then?
Reply | Thread | Link



The Bellinghman
User: bellinghman
Date: 2005-07-15 15:47 (UTC)
Subject: (no subject)
I am under the belief that at one stage, he did indeed say that he was not a statistician.

Which makes his continuing defense of his use of statistics breathtaking.
Reply | Parent | Thread | Link



The Uitlander
User: uitlander
Date: 2005-07-15 16:00 (UTC)
Subject: (no subject)
Hm. This reminds me of my archaeology lecture on how to spot dodgy data. Specifically anything where they show neat patterns of data diving into 25%s, 50%s and 33.3%s. Otherwise known as sample sizes of under 10. Its amazing how many allegedly 'good' PhDs descended to this tactic whilst carefully obfuscating the actual sample counts. The Prof at NCL looked suitably shocked when I told him his favourite seasonality study was a pile of poop, and proceeded to explain why .... but then this was the man who said on one occassion 'You know, I've been talking to the creationist student and he's right - we have no scientific explanation for the actual mechanics of evolution' (he was teaching the human evolution course at the time), and then accused me of watching too much science fiction when I said 'mutation'.

Mind you, I also had to teach his kids how to use venn diagrams (he couldn;t understand them and they needed some coaching for a school entrance exam) apparently it was too 'mathematical' for him. Ye gods, he claimed to be an 'archaeological scientist'.
Reply | Parent | Thread | Link