Social Icons

Saturday 19 May 2012

Journalism And Fun With Stats

Few journalists independently verify "facts" and figures these days, and many fail to seek out contrary viewpoints . Perhaps they never have done so, because honestly who has the time?  Journalists often work under extremely tight deadlines, they are usually working on several dozen pieces simultaneously, and they rarely have support staff to assist them in finding out information. Out of necessity, journalists rely on trusted contacts, insiders, university and college published studies, and publicly-available government statistics to present the "facts" of any given story.  In other words, anything you can find on the Internet these days.

Moreover, to write a good article for public consumption, it needs an air of sensationalism to it to grab the reader's attention.  It does seem to me that people prefer to be entertained by their news much like they want to be wowed by their fiction, so editors and journalists are eager to find some calamity or Armageddon angle to oblige their readers.  The news has always been sensationalised, and it is inaccurate to suggest otherwise.  "If it bleeds, it leads." 

It will be no surprise to any regular reader here that I despise the mainstream media. I rarely read the news unless I'm searching for something specific, because honestly I do not trust the media.  It's not the sensationalism.  I expect that, although it does annoy.  No, it's the presentation of opinion as if it were fact.  It's presenting bollocks estimates of figures as though they are entirely accurate without seeking independent verification. Hey, I know you media guys don't have much time, but come on, couldn't you at least try to be fair and honest for the sake of your readers and viewers? Could you at least include a line that says you didn't bother to check these things out fully and completely for the sake of your own integrity? I could be happy with that.

All of this brings me to a recent plain packs article by Michael Skapinger, columnist for FT.com.  Now, I am vaguely familiar with Mr Skapinger. Three years ago he wrote a pretty fair opinion piece about smokers in the workplace from his viewpoint as a non-smoker.  The gist of that article was that smokers are more social at work, are better at networking with colleagues, and therefore benefit from their smoke breaks despite any of the harm smoking may cause. I think he even felt bad for smokers being forced to stand outside once the ban came in.  So, I have to admit that I when read this latest article by Skapinger, titled "Tobacco companies versus the plain truth," (note: registration may be required) I was a bit stunned to read this (emphasis added):

"So when Alison Cooper, chief executive of Imperial Tobacco, says that forcing tobacco companies to sell cigarettes in plain packets is not about health but is just anti-business, every other business person should tell her not to taint them with her death and disease-ridden trade."

This sounds a lot like Glyn Moody's hack piece on trade marks, doesn't it?

Even so, I will concede that Skapinger has a right to an opinion like anyone else. What I take issue with is his presentation of "facts."  He cites a UK government report (DH report) as the basis for saying that smoking kills more people than "drink, drugs, road crashes, all other accidents, suicide and preventable diabetes combined."

Well, this is quite selective in itself by limiting it to only these other causes of death, but it doesn't actually represent the full picture of UK mortality rates and leading causes of death .  In the DH report, it claims that 81,400 people die from smoking-related disease.  The report derived that number from an NHS document, "Statistics on Smoking: England, 2010; NHS Information Centre for Health and Social Care" (NHS report).  The DH report even gives you the handy graph below to work it all out how they came to the 81,400 figure.  Simple, right?


But, no, it's not so simple at all.  Because "attributable to smoking" is not the same thing as "definitely caused by smoking."  Now we need to go to the NHS report and see how they came to those figures, where we learn in the introduction to section 4 pg 76, that "These figures are estimates of the numbers of admissions and deaths in England which were caused by smoking."  OK, we have estimates only.  If you were to peruse the NHS report, you would see a whole lot of "can be caused" but nowhere do you see any real data on the actual cause of death by smoking.  Why?  Because the NHS doesn't keep track of that information.  So a statistician estimated it based on -- "In 2009, there were a total of 448,230 deaths of adults aged 35 and over in England, 81,400 (18%) of which were estimated to be attributable to smoking."  There you go.

And how did the statistician derive the estimate for 18%?  I'll tell you.  It is currently estimated that between 18 to 23% of the UK population are smokers. This estimate includes occasional smokers, you know that very friendly girl who bums fag after fag off you at the pub after she's had three vodka Red Bulls. Of course, it's all estimates. Nobody really knows the actual number, so it's an assumption. It is, quite possibly, unknowable.  So you take the low end of the estimate (although using a mean may have been preferred from a tobacco control perspective) and you tally up the deaths, split it among women and men smokers, and away you go to your nicely round figure of 81,400 deaths per year attributable to smoking.

Bollocks?  You bet, yet it is good enough for tobacco control's lies, the NHS, and for journalists apparently.  But let's have some fun with bullshit and completely inaccurate representation of facts, because I'm not going to drill down to precise causes of death -- indeed, who has time for that?  If they can do it, so can I.  For this exercise, I will use the 81,400 figure (for illustrative purposes, because it's handy) along with the mortality figures from 2010 (MS Excel).  I recommend downloading the Excel spreadsheet, because it's quite fascinating, particularly the number of "unknown causes of death."  Anyway...

We will focus on only two tables in the spreadsheet, table 5.2 - Neoplasms (all cancer types) and table 5.9 - Diseases of the circulatory system; both of these are the primary sources for determining smoking deaths and I've rounded the percentages up a bit. 

In 2010, there were 493,242 total deaths of all ages, not just 35 and over, but all ages.


29% of all deaths are due to neoplasms (all types of cancer). Total neoplasms / cancer deaths = 141,116.

32% of all deaths are due to circulatory diseases.  Total circulatory disease deaths = 158,084

So we know that 61% of all deaths are due to cancers or circulatory diseases.  Total deaths = 299,232

Obviously, not all of these deaths could be attributed to tobacco use, but again, time is a factor so I haven't isolated particular causes of death.  Anal cancer is included in these figures, but do not be alarmed, smoking causes ALL CANCERS!  That's what they tell you.  Moving on...

Using the DH's 2009 figure of 81,400 (again it is an estimate only and not accurate), only 27% of the 61% of cancer and circulatory deaths are "smoking-related."  So, roughly 1 in 4. Right?  That means about 73% of non-smokers, or approx 217,832 people died in 2010 from these two types of "smoking-related" diseases somehow.  So you really cannot say that "smoking is the leading cause of preventable death" because it just isn't (and there is no such thing as premature death or perhaps even preventable death. It is just death).  Still, how do you explain the other three quarters? Why are they dying from cancer or heart diseases?  Which makes you wonder what is really causing all these deaths... it couldn't be getting old could it?

Furthermore, if we use the 81,400 figure against the overall death, we see that "smoking deaths" hover around 16.5% of all deaths annually (less than the estimated percentage of smokers), compared to the 61% of all cancer and heart disease deaths combined.  So what is going on here?  Shouldn't non-smokers be living forever?  Why are more non-smokers dying of smoking-related deaths than smokers?  This just doesn't make sense to me.  But, hey, like anyone, I'm just having fun with statistics, picking and choosing what I like, mixing up the years, and presenting all of it in a particular way to support my viewpoint.  If they can do it, so can I.  So who is more accurate?  Them? Me?  Does it really fucking matter?  We are all going to die, some of us quite horribly, it is very sad to say.  Nothing can stop death. It comes for all of us.

But I would expect journalists to exercise a little due diligence when presenting figures as facts.  They should fact check and look to see how the data was derived all the way to the original source, and not just be witting or unwitting mouthpieces for tobacco control and hateful government agendas. They should provide independent verification of figures, even when figures are derived from reports supplied by the government, or at least admit they haven't done any of this.  But they don't. Maybe they cannot due to time, or maybe they simply cannot be bothered.  After all, why let facts get in the way of a good story.  Hmmm...?

(For further study you may also wish to view the Leading Causes of Death by age group in England and Wales (MS Excel), which is incredibly illuminating.  Also, a good reference starting point for future research is here.)