Tuesday, 22 March 2011

Top Drawer Propaganda Commands A Significant Fee

Be very delicate when you feast your eyes on this study because, unless there's more to come, it is an expensive item. As referenced by Earl Howe in the Lords last week, this appears to be the document which government is quoting to support its implementation of the smoking ban.

It more or less covers the four areas on which the Department of Health had commissioned research as part of the promised three year review mentioned in the Health Act 2006.

List of research commissioned by the Department

Impact of smokefree legislation in England on individuals and communities: qualitative longitudinal study.
Cost: £460,000
Status: complete

Bar Workers’ Health and Environmental Tobacco Smoke Exposure (BHETSE)
Cost: £220,000
Status: complete

Evaluating the impacts of smokefree legislation using secondary data
Cost: £220,000
Status: Ongoing

Smokefree - feasibility study - secondary analysis of data relating to the hospitality sector
Cost: £47,000
Status: Ongoing
So when you read it, remember that - running to 23 sides - it works out at just over £41,000 per page. For that money, it must be the dog's cojones then, surely. Well, I suppose that depends on how you would define the word 'significant'.

Let's examine how such a word is treated in Linda Bauld's 'evidence review', shall we? On page 11, it states.

[...] new evidence has recently been published on the impact of smokefree legislation on reducing hospital admissions for myocardial infarction (MI) – heart attacks (Sims et al, 2010b).
Pause for a moment while we check the reference on page 22.

Sims M, Maxwell R, Bauld L and Gilmore A (2010b) The short-term impact of smokefree legislation in England: a retrospective analysis on hospital admissions for myocardial infarction. British Medical Journal, 340, DoI: 10.1136/bmj.c2161.
Err, is that the same 'Bauld L' who has written this review? I think you'll find that, yes. Yes, it is.

So what does the study authored by well-paid tobacco control warrior Linda Bauld, referencing a study co-authored by, err, career anti-smoking advocate Linda Bauld, say on the subject?

As part of a Department of Health funded study assessing trends in key health outcomes, an analysis of Hospital Episode Statistics was conducted to explore the impact of smokefree legislation on admissions for MI.
Very modest of Linda not to mention her involvement in it, don't you think?

The study examined emergency admissions from July 2002 to September 2008 and so included 15 months of post-legislative data. The analysis adjusted for secular and seasonal trends and also variations in population size. It found a statistically significant reduction (−2.4%) in the number of admissions for MI following the implementation of smokefree legislation.
And here, courtesy of VGIF, is what a 'statistically significant' reduction looks like.


Stark, isn't it?

Elsewhere in the document, we find something quite different.

Impact on the Hospitality Industry

Annual Business Inquiry data were only available to 2008 at the time the scoping study was conducted. These data showed that turnover fell in 2007 for bars but not for restaurants or hotels.

Data on consumer spending showed some signs of a small decline in the third quarter of 2007 in the period following the introduction of smokefree legislation, but the size of this fall did not reach statistical significance.
And what do insignificant statistics look like?

UK Pub Closures 2004-2009

After rubbing my eyes at reading this stuff, I had to check that I hadn't misunderstood the definition of significant. An unnoticeable change in the trend for heart attack admissions - for which there are hundreds of contributory factors - is described as definitely caused exclusively by the smoking ban. Whereas a dramatic and unprecedented increase in pub closures - for which there are, at most, five theoretical causes - is deemed inconsequential despite the ban being the only one which fits perfectly.

Here is what the dictionary says on the word.

significant adj \sig-ˈni-fi-kənt\

a : having or likely to have influence or effect : important; also : of a noticeably or measurably large amount

b : probably caused by something other than mere chance
Nope, I haven't got it wrong, Linda Bauld really is telling us that black is white, and vice versa.

No wonder the report cost so much - superlative bullshit from a world class leader in the field doesn't come for buttons, does it?


11 comments:

Lysistrata said...

Excellent post, Dick, and very clearly presented. Thank you.

Dick the Prick said...

I am so fucking in the wrong job - i'd have done it at £40k per page because, you know, i'm worth it!

Me & my boss used to do shed loads of analytical stuff for local gov, cops, NOMS etc and when we knew it was bollox just popped graphs on the presentation and said 'well, it says it all, really' and fucked off to the boozer. These charlatans give analysts a bad name.

handymanphil said...

As you say Dick, this is utter bollox, but at least it is significantly less repulsive that that which Labour paid £960,000 for, which I have read. That report was virtually compiled within 6 months of the implementation of the ban, using 'compliance' as the guiding light as to our happiness with the ban and ASH 'statistics' to prove that 101% of smokers really didn't want to smoke!
Oh, and the surveys all came from anti smoking delegates of various AS organisations! The Gilmore 'thing' was in that report as well!

Chalcedon said...

Dick, she should have given a P value so we can calculate the probability of X happening by chance. This is how you assess significance. a decrease of a mere 2.4% is just noise and will not be at al significant. She should have looked at the long term trends in MI and should have either sen a large decreae in MI to say this is significant, based on an apropriate P value. If there is no P value given then it is not significant.

Usually you have to sate whether you have an interest (vested) in the topic you are reporting on so the reader can judge your level of impartiality.

Anonymous said...

Just looking at the two charts side by side, without running probability statistical equations, it is self-evident that the reports are biased. Does anyone in the media, government or among readership of these reports have an open eye to the obvious, or does everyone just continue to willingly believe the unbelievable.

smokyrobinson said...

I am smoking a fag

nisakiman said...

We should be used to this kind of blatant propaganda by now. It's unrelenting.

I do wonder why no-one, but no-one in the MSM ever picks them up on this. Do they think it would be media suicide? I thought journalists were supposed to be fearless seekers after truth. So what happened to that concept then?

The only journalists who seem to notice these things are the people at 'spiked', and sadly they aren't really mainstream.

Sometimes I just get so depressed about it all. I've always despised sanctimony, and these days it's being served up by the shovelful.

Anonymous said...

As Dick The Prick commented, it's standard practice to insert lots of colourful visuals to cover for a lack of substance - as my old mentor described it, "If you can't dazzle 'em with brilliance, baffle 'em with bullshit".

His other magic trick was to close any presentation with a startlingly appropriate quote from a respected, but long dead, figure in history - except he made them all up, safe in the knowledge that no-one could ever prove Churchill or Napoleon or Brunel never actually said it. Worked every time.

Gawain Towler said...

Brilliantly done Dick,

Have you written to her, Ms Baulds, or the Minister. If not why not?

Anonymous said...

This will be long. No apologies.


Do these people know nothing about science and statistics? Do they have any common sense?

(I know. Common sense is not common but it is absent here.)

There is no need for "statistics" and "confidence limits". We are not dealing with samples - we are dealing with REAL total numbers. Assuming the data are correct:

Did the number of admissions for AMI go down?
Yes.
How much?
2.4%
(Incidentally, I won't mention the suppression of the zero to make the change look more dramatic and the use of a smooth line and block of colour when it would appear a bar graph would be better. Oops. I mentioned it - too late!)

Did the number of UK pubs closing go up between 2004 and 2009?
Yes
How much?
From just under 500 p.a. to close to 2,500 p.a.

These are the facts. Statistics and Probability Theory have no meaning here. One set of data goes down, one goes up. These are the real numbers. It happened. Get over it!

The only question is why? Why did they change? We are told that the number of entries into hospital went down because of the influence of smokefree legislation. This is a classic case of "post hoc propter hoc":

"The post hoc ergo propter hoc (after this therefore because of this) fallacy is based upon the mistaken notion that simply because one thing happens after another, the first event was a cause of the second event. Post hoc reasoning is the basis for many superstitions and erroneous beliefs."

http://www.skepdic.com/posthoc.html

Just in passing " ... superstitious and erroneous beliefs ..." seems to fit well here.

Clearly, there already was a downward trend in admissions (for reasons unknown but probably for multiple reasons). The trend continues. The smokefree attitude of government has not reversed it and if it has affected it against all the other factors the effect is not large enough to disturb the steady downward trend.

It is easier to make the case of a change on the number of pubs, although the case that it was solely due to smokefree legislation would have to be made with care. I would guess that it has had a large impact (but I cannot say how large).

Declaration of Interest

I am not a smoker. I have never smoked in my life. I dislike cigarette smoke. I did not like going into a pub and having to wash my hair and my cloths afterwards. I wish people would not smoke near me. However, I do have 2 legs and it is usually pretty easy to walk away!

Having been a scientist all my life (now 65) what I REALLY dislike is people using junk "science" and "statistics" to confuse and deceive and control others. Even if the motive is deemed to be "good" there is NO excuse for distorting the facts.

Alan Bates

Anonymous said...

Heart attack admissions could have been going down both before, during and after the ban because of the statin usage. Statin medications might have been the correlating factor, nothing to do with smoking or smoking bans.