Lies, damn lies and statistics

[cojak]

http://www.theregister.co.uk/2011/04...causes_cancer/

Loved the last paragraph the best

[DaveB]

GIVING UP BOOZE CAUSES CANCER - shock study ? The Register

Loved the last paragraph the best

From the Daily mash article linked...

'The tulipfaced King of the Monkeymen'.

[NickFitz]

Unfortunately, El Reg has completely misinterpreted the figures they quote: they paraphrase "If we assume causality, among men and women, 10% (95% confidence interval 7 to 13%) and 3% (1 to 5%) of the incidence of total cancer was attributable to former and current alcohol consumption in the selected European countries" which is in the abstract. If you read the full paper, you find that this is summarising the findings whose presentation begins: "If we assume causality, these estimates would translate into 10% (95% confidence interval 7% to 13%) of total cancer in men (table 3)⇓ and 3.0% (1% to 5%) of total cancer in women (table 4)⇓ being attributable to alcohol consumption in these selected European countries" which makes it clear that the figures of 10% and 3% refer to the distinction between men and women, not the distinction between former and current alcohol use.

[gingerjedi]

From the Daily mash article linked...

And this...

Meanwhile, Brubaker admitted that bringing up the subject of drinking and cancer on a Friday, during a spell of warm, sunny weather, does make him seem like the sort of utterly miserable ****er who deserves to be eaten by a crocodile.

[minestrone]

Unfortunately, El Reg has completely misinterpreted the figures they quote: they paraphrase "If we assume causality, among men and women, 10% (95% confidence interval 7 to 13%) and 3% (1 to 5%) of the incidence of total cancer was attributable to former and current alcohol consumption in the selected European countries" which is in the abstract. If you read the full paper, you find that this is summarising the findings whose presentation begins: "If we assume causality, these estimates would translate into 10% (95% confidence interval 7% to 13%) of total cancer in men (table 3)⇓ and 3.0% (1% to 5%) of total cancer in women (table 4)⇓ being attributable to alcohol consumption in these selected European countries" which makes it clear that the figures of 10% and 3% refer to the distinction between men and women, not the distinction between former and current alcohol use.

[sasguru]

Unfortunately, El Reg has completely misinterpreted the figures they quote: they paraphrase "If we assume causality, among men and women, 10% (95% confidence interval 7 to 13%) and 3% (1 to 5%) of the incidence of total cancer was attributable to former and current alcohol consumption in the selected European countries" which is in the abstract. If you read the full paper, you find that this is summarising the findings whose presentation begins: "If we assume causality, these estimates would translate into 10% (95% confidence interval 7% to 13%) of total cancer in men (table 3)⇓ and 3.0% (1% to 5%) of total cancer in women (table 4)⇓ being attributable to alcohol consumption in these selected European countries" which makes it clear that the figures of 10% and 3% refer to the distinction between men and women, not the distinction between former and current alcohol use.

Indeed.

I love it when journalists and banjo-strummers assume (in true CUK fashion) that they can comment on a discipline on which they have not spent the requisite number of years to master.
The study is pretty thorough and a brief perusal shows no obvious fault with the methodology.
It is pretty clear that there is a strong link between some cancers and excessive alcohol consumption.

[cojak]

Commenting?

I thought that presenting was merely that - oh, I might have made an observation regarding that last paragraph.

I'll leave commenting to the more mouthy members of the congregation

[minestrone]

The Lancet said following the judgment of the General Medical Council (GMC) fitness to practise panel last Thursday it had become clear that several elements of the 1998 paper by Dr Andrew Wakefield and others were incorrect.

The panel made a number of criticisms of Dr Wakefield, including that he was misleading and irresponsible in the way he described the study.

The research sparked a massive drop in the number of children given the triple jab for measles, mumps and rubella.

The editors of the Lancet said it had become clear that several elements of the paper were incorrect, contrary to the findings of an earlier investigation.

''In particular, the claims in the original paper that children were 'consecutively referred' and that investigations were 'approved' by the local ethics committee have been proven to be false. Therefore we fully retract this paper from the published record.''

Don't assume medical journals are always correct. As Milgram shown people's susceptibility increased when they were told (incorrectly) the tester was a doctor.

[sasguru]

Don't assume medical journals are always correct. As Milgram shown people's susceptibility increased when they were told (incorrectly) the tester was a doctor.

I don't.
Doctors can be notoriously innumerate and unscientific, since their mathematical (and hence logical) training is trivial, as opposed to the ability to master loads of facts.
Luckily since stats is really applied maths, it is easy to check if a methodology is sound or not, since, unlike what most people believe, there are sound foundations for it.
The "Lies... blah... blah .and Statistics" quote is just a cliche, originated by Mark Twain, I believe.

[shaunbhoy]

I love it when journalists and banjo-strummers assume (in true CUK fashion) that they can comment on a discipline on which they have not spent the requisite number of years to master.

I love it when dimwitted laborious plodders that take years to master relatively basic concepts, get on their soapbox and lecture others that are clearly light years ahead of them intellectually.