Study shows researchers ignore structural issues

[trigger warning]

The BBC has yet another badly-researched story about "obesity". I'll skip over the unmentioned assumption that "obesity" is such a bad thing it needs to be prevented at all costs, and move on to places where the article fails even in its own terms.

Firstly, the article talks about how average calorie intake has increased significantly in the US since the 1970s. Naturally, because "obesity" is a personal problem with no structural issues whatsoever, the article completely omits that since the 1970s the USA's problems with perverse food-production subsdisies, "food deserts" and so on, in many places it's difficult or impossible to get foods other than the energy-dense ones, and more expensive even when they are available.

This is also something that does not really generalise to the majority of the UK, or other Western European countries, despite the claim in the article that "many of the factors causing the obesity epidemic there are mirrored in the UK".

The recommendations of the researchers - police what people individually eat, rather than making it easier for them to get nutritious food cheaply - are equally predictable.

Secondly, the article in a box-out includes a note that:

The recommended daily calorie intake is 2,000 for women, and 2,500 for men (NHS Choices)

If we assume that there are roughly equal numbers of men and women, and gender is the only factor that affects energy needs, metabolic rates, digestion efficiency, and so on - but does so with 100% predictive power (which is ridiculous, but is what the recommendation needs to be assumed) then that gives an average recommendation of 2,250 calories daily for each person.

Compare that with the figures from the article on US average - the article doesn't specify if it's mean or median - calorie intake:

• 1977-8: 1,803 calories
• 1994ish (implied): 2,145 calories
• 2003-6: 2,374 calories

So, comparing with the "recommended" average, we can see that it's only in the last decade or so that US residents have stopped being (on average) underfed.

The average hides a lot of rich/poor disparity, of course, and a lot of people in the US are still malnourished - my point is that even in terms of pretending these averages mean anything at all, the conclusion isn't the one that the rest of the article implies.

Of course, this is fairly typical of "anti-obesity" research and campaigning: set an unjustified target for calorie intake - and then claim it's bad when people get anywhere near it. "Pro-malnutrition" might be a better term for them: perhaps if they didn't have the privilege not to remember the effects of widespread malnutrition themselves, they'd think more carefully about what they were saying.

Crime statistics? Now why would journalists understand those?

[trigger warning]

So the BBC has an article on the number of women who have been convicted for domestic violence. It's as bad as you'd expect a BBC article to be.

It notes a rise in convictions of women for domestic violence between 2005 and 2010 from 1,500 to 4,000.

The article then goes on for several paragraphs speculating about why this might be, cultures of violence in women, etc.

In the final paragraph, after most readers will have got bored and wandered off, they note that in the same time period convictions of men committing domestic violence rose from 28,000 to 55,000 - not quite as large a proportionate rise, but still big.

Further penalty points, too, for their heteronormative assumption that all domestic violence is committed against someone of the 'opposite' gender to the attacker.

Domestic violence reduction is actually a relative success for our justice system and government - with the rates of assault significantly falling, and the rate of convictions significantly increasing. It remains a serious problem - around 300,000 incidents a year - but this is a third of what it was 15 years ago, and a vastly-increased proportion (now around 1 in 6) of those perpetrators are convicted (even if not all of them get appropriate sentences).

It's still really bad and there's much still to do - but there has been significant progress made. Not that you'd know that from this article.

Lots of quotes from un-named "some experts", too - presumably because no real expert would actually put their name to such uninformed speculation.

Here's the complaint that I sent them.

Regarding your article "Women's convictions for domestic violence 'double'", I found it to be poorly researched and sensationalist, leaving a key piece of context for the final paragraph, and ignoring obvious research that could easily answer some of the "experts say this" / "other experts say that" pseudo-debate in the article. By doing so, it gives a completely misleading impression, both about the extent of domestic violence generally, and the number of women committing it.

Firstly, convictions are not at a measure of the prevalence of crime. A doubling of convictions could mean twice as many crimes were being committed, or it could mean that the number of crimes was constant but the reporting rate had doubled, or it could mean that the reporting rate was constant but improvements in police and CPS procedure meant that the chances of a conviction had doubled. Or - more likely - some combination of the three.

Fortunately, the British Crime Survey is a long-standing statistical measure of crime, including domestic violence, and - together with other CPS and Police/Home Office statistics - can be used to answer this. A quick summary of the figures can easily be found by searching for "BCS Domestic Violence" - http://www.dewar4research.org/DOCS/DVGovtStatsAug09.pdf for instance

It shows that - far from increasing, the rates of domestic violence against both men and women have been generally decreasing. The proportion of victims who were male is also roughly the same as it has been over the last decade.

Secondly, the key piece of context - that there had also been a massive increase in convictions for domestic violence in general and by men specifically - is left for the last paragraph. This is a key piece of context, which strongly suggests that much of the increase is due to improvements in prosecution and investigation - contrary to the suggestions in the early paragraphs - and yet it is buried at the bottom of the page where relatively few readers will see it.

This news of a massive increase in convictions for men is not accompanied by "some experts" wondering if it is part of a "growing culture of violence" among men - despite the 2005 number for men being seven times the 2010 number for women.

Thirdly, the article assumes throughout that all male victims of domestic violence were attacked by women, and vice versa. This is obviously not the case, but the article implies it anyway.

Please let me know what steps will be taken to ensure that such basic mistakes in reporting on crime statistics do not recur in future.

Journalists downplay rape prevalence in DR Congo

[trigger warning]

Several news sources are reporting on this study on rape prevalence in the Democratic Republic of Congo. The headlines are all along the lines of "48 rapes per hour" or "1100 rapes a day". Here's the BBC article.

My first thought on reading the headline? "That's really low." Somewhere between study, press release and news article, they've lost some really important details.

The "shocking" figure they quote is - while absolutely terrible - only a fraction of the real problem.

The UK is a similar size to the DRC, in terms of population (62 million and 67 million respectively). Going from British Crime Survey data, my low bound for the number of rapes annually in the UK is around 200,000. It's a low estimate because the BCS data doesn't cover children or people aged 60 or over, but it gives a rough estimate.

The reporting claims the figure for the DRC to be around 400,000. Worse, even accounting for the slightly higher population, but not much worse.

"UK: 20 rapes every hour, study finds" would be an equally accurate headline (and perhaps expressing it like that might shock a few people into action).

The figure for the DRC is considerably higher - not just twice as bad as the UK

1. The study, according to the article, only covers women and girls aged 15-49. Like the BCS, that misses out a lot of people (and, indeed, the authors of the study acknowledge this)
2. The figure of just over 400,000 is not how many rapes occurred, but how many women were raped in the last 12 months. Like the BCS, it doesn't measure multiple victimisation. Given that the lifetime estimate is only four times higher, there must be a lot of multiple victimisation going on.
3. They also note a figure of over 3 million women experiencing "intimate partner sexual violence", though the abstract doesn't say whether this is lifetime or annual. It's not quite clear how this figure relates to the initial count, but from the Guardian reporting it seems to be a separate figure, and includes yet more rapes. Not at all surprising - but imagine how much better the UK's figures would look if they only counted stranger rapes; now think about what that implies for the DRC figures.
4. Unlike the BCS, the figures this study were based on don't appear to have had the same effort put into avoiding reporting stigma (Dr Peterman explicitly mentions this as a cause of underestimation).

Add all that together and this study is still a massive underestimate.

It's also a big increase on previous estimates - which says a lot about those estimates.

That rate is significantly higher than the previous estimate of 16,000 rapes reported in one year by the UN.

If anyone at the UN actually believed that figure - again, only about twice the number reported in the UK - to be remotely accurate, they would have been congratulating the DRC on its successful strategies against sexual violence. Clearly, no-one actually did - the UN staff are not naive! But again, it's measuring something completely different - and the DRC government points out that the difference between the two figures is due to reporting and collection, not a change in the number of actual rapes.

But I don't think the journalists get just how high rape prevalence is in countries like the UK, or they wouldn't print things like (BBC):

The highest numbers of rapes were found in war-ravaged North Kivu, where an average of 67 women out of 1,000 have been raped at least once.

...which, back of envelope, makes North Kivu noticeably safer than the UK. Or it could be - as the authors of the study state - a severe underestimate.

If UK journalists - and the public as a whole - were actually aware how many rapes took place in their own country, then they might not have to have the obvious underestimates here pointed out to them - underestimates that make the DRC's problems with rape look nowhere near as bad as they actually are.

In an effort to get the reporting to be more accurate, here's the message I sent to BBC Complaints - I also sent a similar one to the Guardian regarding their article (to reader@guardian.co.uk). I didn't even get into the extra 3 million, since I wanted to keep the initial complaint short. I'll mention it if they get back to me.

Dear Editors,

Your article reported that "A study by US scientists has concluded that an average of 48 women and girls are raped every hour in the Democratic Republic of Congo.". The study actually concludes something far worse than this - the figure of 400,000 a year from which the hourly rate is derived is the number of women and girls raped at least once in the last twelve months. Dividing this into hours forgets that many will be raped far more than just once. Furthermore it ignores the statements made - and reported on - by Dr Amber Peterman that the estimate from their study is likely to be a significant underestimate.

This becomes obvious when you consider the comparable rate for the UK. The British Crime Survey finds that around 70,000 people between the ages of 16 and 59 have been raped at least once in the last twelve months. Studies on multiple victimisation suggest that many rape victims are raped more than once, for an estimate of around 200,000 rapes each year in the UK - or around 20 every hour.

Similarly, your article gives a lifetime victimisation rate for the worst area - North Kivu - of 67 in 1,000 - a rate that is actually somewhat less bad than the equivalent rate for women in the UK.

I'm not suggesting that rape in the UK is as prevalent as it is in the DRC - though "UK: 20 rapes every hour, studies show" is a horrifying statistic in its own right - but by doing a simple division your headline and article actually hides the true scale of the problem in the DRC.

Please correct the article - especially the headline and lede paragraph - to better reflect the original study.

Yours faithfully

[me]

Or it might be that your major underlying assumption is rubbish

[trigger warning]

Another piece of terrible BBC reporting (complete with stock "headless fatty" photo), this time about life expectancy and the "obesity epidemic".

The lead paragraph is:

Life expectancy in the UK is on the rise, along with the rest of Europe, despite fears over the impact of obesity, a population expert has said.

In summary: life expectancy has been rising in the UK, mainly because of better treatment for heart conditions. This is a surprise, because as reading the news makes clear, everyone is becoming "obese" and should be dead already.

It can't at all be because - as major population studies in Canada, Germany and the USA have consistently shown - there is no statistically significant link between BMI (the measure of "obesity") and life expectancy.

We are yet to see the impact of a generation of people who have been obese from childhood through to adulthood. We can't predict how that will affect life expectancy figures in the future.

While more individuals nowadays are likely to be sufficiently far above the government-mandated weight that they count as "obese" for most of their lives (and especially now that the government - through the time-honoured technique of "making it up" - is defining more and more children as "overweight"), there are plenty of people who have been "obese" for all their lives, who were in those major studies, and who didn't die at any greater rates than anyone else.

But this argument¹ will keep them going on the "fat will kill you" track for a few decades.

At some point historians of medicine are going to look back and show how researchers managed to completely ignore major population studies saying "no effect" in favour of studies that don't get beyond correlation is causation, because they were so convinced of the truth of their main hypothesis that they just ignored evidence against it.

Until then, more terrible public policy and reporting, and more encouraging of hatred against fat people.

Footnote

¹ It occurs to me that this resembles the backstop argument for people when presented with unbeatable evidence of (e.g.) gender or racial inequality: "that is due to discrimination 30 years ago the after-effects of which haven't worked their way out of the system yet; in another 30 years it'll have gone". It's an argument that never gets old.

Reverse causality, age, and scientists

According to the BBC, "Fear of falling 'boosts elderly's fall risk'" - that is, the more worried an old person is about falling over, the more likely they are to actually do so. The full text of the study is freely available, for once.

The reporting is actually a reasonable summary of the paper's conclusions, for once - though it as usual misses out the numerous caveats written into the paper and presents the conclusion as established unambiguous fact.

Of course, the paper's conclusions are somewhat dubious.

The research assessed 500 people aged between 70 and 90 and without a fairly long list of disabilities and health conditions to calculate an estimated fall risk based on their physiology (various assessments of proprioception, balance, visual senses, muscle strength, and so on). It then asked them about their perceived fall risk.

The study then compared what the researchers' thought their chance of falling over was with what the individuals themselves thought, and divided them into four groups for "researchers think" high or low versus "individuals think" high or low. The study talks as if these were very distinct groups, but if you look at Figure 2 in the paper, there is a lot of clustering around the centre, and you could shift the boundaries quite significantly with relatively small changes in where you defined the centre point.

Over a year of follow-up study they discovered that of those people who believed they had a low risk of falling, but the researchers' assessment categorised them as high risk, only 30% had actually fallen. Meanwhile, of those people who believed they had a high risk of falling, but who were assessed by the researchers as having a low risk, 40% had fallen.

The obvious explanation, to me, is that the physiological assessment for risk of falling is missing at least one major factor that makes it inaccurate in a number of cases. This is mentioned as a possibility in the paper:

It is possible that the anxious group had a residual physiological fall risk not encompassed by the physiological profile assessment. Thus, although the group’s low physiological profile assessment score indicated that the integrity of individual sensorimotor components was adequate, their relatively poor performance on the coordinated stability test suggests that the integration of each of these physiological systems in the maintenance of dynamic balance control was impaired.22 However, the disparity between physiological and perceived fall risk in the anxious group seemed to be strongly related to psychological factors.

...but dismissed without much of an explanation as to why.

As far as I can tell, they're getting cause and effect completely backwards.

They're assessing people who have been living for some time, and presumably have fallen over during that time. In the 69-89 years prior to this study, they've presumably each built up a good picture of how often they individually fall over. Those who fall over often presumably are more worried about doing so again (or at least, what's actually being measured, think they are more likely to do so again whether or not this worries them).

When you study them for a year - no surprises: past experience is a good predictor of future experience in this area.

The results they've got are exactly the results you'd expect for any event the frequency of which can be predicted reasonably well by the people it happens to, but not as well by a statistical assessment over the population. You could conclude for any of those that "expecting X causes X" but it wouldn't actually be justifiable.

Dismissing this possibility with a vague statement that "the disparity between physiological and perceived fall risk in the anxious group seemed to be strongly related to psychological factors" doesn't actually disprove it, especially when actual fall risk matches perceived fall risk quite well.

But I guess it's easier to disbelieve old people than to believe your "physiological assessment" is flawed.

Gender pay gap: survey says 57 years before reporting is accurate

The Chartered Management Institute have released the results of a survey of the gender pay gap among managers.

It looks as if the £10,000 gap reported is the difference between the means, which doesn't make it particularly useful. While it's both unsurprising and wrong that top-level male managers in big companies make several hundred thousand more on average than their female counterparts (and there are significantly more men than women at this level), most women (and indeed almost as many men) are never going to get anywhere near that sort of job anyway. At the junior management level, there was an average £1,000 annual gap between men and women, which is still fairly substantial.

The study has been reported (or at least, the press release reprinted) quite widely, and all of the major news organisations I've found (Telegraph, Guardian, BBC, and Daily Mail) have generally failed to do any sort of useful journalism:

• They all report the raw figures from the press release (I've rounded them off above because the extra precision is meaningless)
• They all report the '57 year' figure as if a constant rate averaging means anything at all. It makes sense to note it because the fact that the year-on-year improvement is tiny compared with the size of the gap is important - but it should be made clearer that it's an illustration of the slow rate of progress, not a prediction.
• A lot of the headlines and sub-headings talk about this being the pay gap as a whole. The article text makes clear it's only talking about managers ... and none of them mention that the pay gap in general, when they go on to look at the UK's generally large pay gap, is not just composed of pay differentials between equivalent jobs ("horizontal") but also due to women not being employed in the higher paid jobs as much ("vertical"), and doesn't just include managers. There are a lot of class issues and intersections with other discriminations being ignored here.
• Other than the Daily Mail¹, none of them note that part-time salaries were converted to a full-time equivalent. The Daily Mail forgets to note that since women are far more likely to work part-time than men, then the conversion may well hide further problems.
• It's not clear what proportion of the gap is "women being paid less than men for the same job" (which the reporting is assuming and focusing on) and which is "women not being in the high paid jobs" (which seems more likely to be causing most of the gap). The distinction is very important - the Telegraph mentions that the Equality Act will outlaw "pay secrecy" clauses in contracts from October, which will certainly help with the first cause but not with the second - and not actually mentioned. The Guardian mentions Sweden's better record on pay gaps, but it's worth noting that Sweden enforces gender quotas on senior management boards of large companies, so the "vertical" gap is at least slightly reduced there.
• Some of the articles have reprinted the press release's figures of the number of employees (around 43,000) and organisations (around 200) that were surveyed, but none of them mention whether there was any attempt to get a representative sample within the organisations (i.e. do the proportions of male and female managers surveyed at each level represent the proportions within those organisations).

As is typical, the survey raises far more questions than it answers, and the journalists don't ask any of them.

¹ And what have we come to when the - narrowly - best reporting of an equality issue is carried out by the Mail?

Class and court reporting

The BBC reports that a man has been jailed for a year for grabbing a helicopter as it took off. This particular bit of my reporting caught my eye:

Sentencing Jafari, Judge Michael Roach said it was a "deliberate and much more a reckless and dangerous act" and, despite the fact he was a business and family man, he had "no choice" other than to send him to prison.

From that reporting, it sounds like, had Jafari been working-class from an inner-city area with a bad reputation (let's ignore the implausibility of anyone with that background being able to have a helicopter land close enough to grab even if they wanted to) rather than a millionaire in an extremely expensive part of Bristol, a custodial sentence would have been much more obvious. Looking at some of the other reporting of the case, though - Metro, Bristol Evening Post, Daily Mail - that's not actually what was said.

But sentencing yesterday, at Bristol Crown Court, Judge Michael Roach told Jafari: ‘Grabbing the right-hand skid on take off was very dangerous and was liable to destabilise the aircraft which had only risen 6ft off the ground. In my judgement your behaviour was deliberate and reckless.

‘You are an intelligent, resourceful man who on this particular occasion let your temper get the better of you and you acted in a dangerous manner. In my judgement the case is too serious to justify a suspended sentence.’

He ordered Jafari to pay £2,800 in costs. ‘There is no reason, it seems to me, why the public should bear the expense.’

No reference to him being a business or family man at all in the judge's words.

So where is the - and it's not in quotes, so it's not directly from his words - BBC's assertion that the sentencing was "despite [him being] a business and family man", rather than in the judge's actual words "[because] the case is too serious to justify a suspended sentence" coming from?.

The judge does refer to Jafari as "intelligent, resourceful" but that's not at all the same thing; furthermore the judge does not put these characteristics forward as a reason for a more lenient sentence (if anything, they're part of the reason to require costs be paid, which the BBC doesn't mention).

So the judge doesn't appear to believe that sentencing should depend on social class, but some of the BBC's writers do.

No, really, this helps no-one.

[trigger warning]

Hannah Mudge has searched through all reports of false accusations in the Daily Mail in the last couple of years. They're getting more commonplace.

It also gives a large sample of recorded false accusations, that can be checked through to see if any of the cases mentioned were previously reported in the press.

(As you'd expect, almost all of them were not)

Some of the articles don't say who was accused. I've tried to look for reporting of the cases in other papers, and if that didn't work, looked for reporting of cases matching what details are available. I've excluded the entries from Hannah's list where the Daily Mail claims that the claim was false on the basis of an acquittal (which I don't believe is always sufficient evidence that the claim is false, even if it is sufficient to prove that the defendant did not commit rape) - this does not remove many cases. I have included cases where the false allegation trial was ongoing even when I was unable to find confirmation that the defendant in the false allegation trial was convicted.

In only one of these cases - including some where the defendant was convicted and later released after an appeal - was the original rape case reported on in the press, so it doesn't significantly change the results to exclude them.

Reported date (from Hannah's list)	Date of verdict different?	Rape case reported before?	Notes
25 June 2010	No	No
18 June 2010	No	No
18 June 2010	Yes - 1999/2006	No
11 June 2010	No	No
8 June 2010	No	No suspects
2 June 2010 case 1	No	No
2 June 2010 case 2	Yes - 2006/8	No
25 May 2010	-	-	Duplicate of 2 June 2010 report
10 May 2010	No	No
5 March 2010	No	No	Gail Sherwood
16 February 2010	No	No
19 January 2010	No	No
5 December 2009	No	No
7 November 2009	No	No
6 November 2009	No	No
1 November 2009	No	Once in a newswire, but no suspects named
14 August 2009	-	-	Duplicate of 16 February 2010 report
22 July 2009	No	No
3 July 2009	No	No
2 July 2009	No	Once, but before any suspects found
30 June 2009	No	No
21 May 2009	*	No	Conviction of accuser not reported previously
23 December 2008	No	No
17 September 2008	No	No
29 August 2008	No	No
25 August 2008	No	No
18 August 2008	No	No
29 July 2008	No	No
8 April 2008	No	Yes, post arrest, one "in brief" entry in local paper
7 March 2008	No	No
12 February 2008	No	No
13 November 2007	No	No
14 June 2007	No	No
12 May 2007	No	No
23 April 2007	No	No
24 March 2007	No	No
16 March 2007	No	No

So of the 37 cases where someone was charged with making a false allegation, only 1 was reported in a way that named the suspected rapist, and that briefly in a local paper (which didn't get a mention when the suspect was later talking about the effect on his life). This is not that surprising - in most of the cases no-one was charged (though several were arrested), which significantly reduces the chances of cases reaching the attention of the media.

The estimate of slightly more than one a year, which I thought was using generous assumptions to begin with, now looks even less likely - I was assuming that only about 4% of rape cases would be reported, almost all post-arrest. It appears from this that of the sub-sample of false allegations, the figure may be even lower than that - around 2.5%, and almost all post-charge. That reduces the number of people likely to be helped after a false allegation by the government proposals to one every three or four years (and again, these stories all show that it's not the non-existent press attention that causes the harm to people who are falsely accused).

So, as expected, it's a proposal that helps no-one but could harm thousands.

Of course, when a rape defendant is granted anonymity, the Daily Mail takes a very different line.

A distorted image of science

[trigger warning for eating disorders]

The BBC, in its usual uncritical way, reports on a paper claiming that people's sense of proprioception is not great. Proprioception, if you haven't heard the term before, is the internal sense of position - you know where any limbs you have are in relation to the rest of your body, at what angle your head is to your shoulders, and so on, without having to look.

The paper itself (payment/subscription needed for full text) is really quite interesting. They tested how exact this sense of proprioception is by asking people to point to the fingertips and knuckles of one hand with the other hand while not being able to see it, and found that people tended to miss by a fair bit, estimating their hands to be shorter and wider than they really were.

I tested this briefly in an uncontrolled fashion with my own hands, and yes, this does seem to happen to at least some people.

So, the next time someone tells you that they know X like the back of their hand? Worry.

This is not an area I have much knowledge about, so if there are giant flaws in the research methodology, I don't stand much chance of finding them. There are none stunningly obvious to me as a scientific generalist, though some things that do stand out are:

• All of their experimental samples (for four variations of the study) have more women than men, some by a significant margin. If gender was significantly affecting the results they didn't mention it, but this suggests some other bias in how they recruited their samples which might be significant.
• They don't mention, but it seems likely by implication, that all of the sample had relatively average vision. Whether this effect also occurs in people who aren't able to use vision to supplement proprioception in most cases seems an important question.

The flaws, as usual, come in the interpretation for the press release and news articles, where, according to the BBC:

The brain naturally distorts body image - a finding which could explain eating disorders like anorexia, say experts.

[...]

Distorted perception may dominate in some people, leading to body image problems, a US journal reports.

Lead researcher Dr Matthew Longo said: "These findings may well be relevant to psychiatric conditions involving body image such as anorexia nervosa, as there may be a general bias towards perceiving the body to be wider than it is.

"Our results show dramatic distortions of hand shape, which were highly consistent across participants."

...and in come the wild extrapolations.

It is, I suppose, possible that the body image distortions of anorexia nervosa have the same basic cause as this one, and it would be entirely reasonable to do some follow-up studies in this direction.

But there's no evidence yet to justify Dr Longo's speculation that there's even a tendency to misperceive your own body width in the general¹ case, or that people with anorexia nervosa misguess where the edges of their body are more than people without.

The proprioception errors they report on are fairly clearly not "proprioception plus vision" errors, but anorexia nervosa is not generally a condition that only takes effect when not observing oneself visually, which is some evidence that they aren't the same effects.

The hand distortion seems to be fairly consistent across the sample and not related to any obvious social effects. Anorexia nervosa has been fairly strongly linked to social effects, and varies widely in severity across the population.

It's an interesting study with potentially interesting follow-ups, but both the news article and Dr Longo's quotes are assuming prior to doing the needed follow-up studies that the distorted body image of proprioception and the distorted body image of anorexia nervosa are distorted in the same way, or even involve the same meaning of "body image". As usual, the BBC seems to have largely uncritically reported and extended a press release.

Here's the original press release - there's one relevant quote at the end, in a press release that otherwise actually summarises the research very well:

“These findings may well be relevant to psychiatric conditions involving body image such as anorexia nervosa, as there may be a general bias towards perceiving the body to be wider than it is. Our healthy participants had a basically accurate visual image of their own body, but the brain’s model of the hand underlying position sense was highly distorted. This distorted perception could come to dominate in some people, leading to distortions of body image as well, such as in eating disorders,” said Dr Longo.

It's not as bad as some of the BBC's science reporting, but it's got all the same types of error - unjustified extrapolation, uncritical reprinting of press releases, not reading the actual paper, etc.

¹ They might also need to sharpen up their recruitment strategies to get a control group from a society where "thin is good / fat is bad" isn't the overwhelming social message.

Parents have more sense than BBC journalists, then

From the BBC, another reprint-the-press-release article, this time with the headline Parents 'more worried about murder than obesity' threat.

On the whole, this seems like a sensible balance. Murder may be rarer than "obesity" but the consequences are significantly more severe.

The findings, in a YouGov poll of 1,244 parents, contrast with data showing the risk of a child being killed by a stranger is a million to one.

The risk of severe health problems for children due to lack of exercise is one in three, figures have suggested.

"Figures have suggested". No mention of what figures, or how reliable, or anything. Also note the addition of the qualifier "by a stranger", which is necessary given that the risk of being killed - at any age - by someone one knows is considerably higher.

As mentioned before, research shows that - for Western industrialised populations at least - the death rate is not significantly correlated with weight (and in so far as it is, the government-mandated weight is correlated with a higher risk of death within a particular time period than heavier weights).

The article does talk more about "long-term health problems" than death as a consequence of obesity, but what's not mentioned is that the evidence that the correlation is causative is sparse. Indeed, for many "obesity-related" conditions, there is more evidence of the reverse - that weight gain is either a symptom of an underlying condition or caused by an underlying factor that also causes susceptibility to that condition.

That lack of evidence of causation is of course a serious problem for the government policies developed around this. It's much like the government, upon discovering that people with blonde hair were more likely to have a particular health condition, recommending hair dye - except that in this case some of the suggestions for hiding the symptom are actively harmful to health.

The article doesn't, at least, mention restrictive diets for children as a possible "solution", instead advocating more exercise and walking to school. The problems with "more exercise" and "walking to school" as a solution have been repeatedly said elsewhere, so to summarise:

• Not every child can exercise that much, because of disabilities, and/or a lack of suitable or sufficiently cheap places to exercise near where they live, and/or a lack of free time.
• Similarly, it's not practical for every child to walk to their primary school, due to the distance, or lack of time, or inaccessible routes.
• The effects of exercise on weight are not particularly clear in adults. I don't know if there have been useful studies specifically for children.

So that's the conclusions, but what about the data used to conclude them? Going back to the article, it's fairly clear that - despite YouGov being a reputable survey company - the questions that they were given to ask were clearly set with the aim of producing the press release in mind. From the BBC article.

But only one in 20 picked concerns about poor health in later life due to the child's levels of physical activity.

Here's the options for the question, which was "which of these do you fear most for your child?"

• They are injured or killed in a road traffic accident - 30%
• They are abducted or murdered - 30%
• Poor health in later life due to your child's current levels of physical activity - 5%
• None of these - 21%
• Don't know - 7%
• Prefer not to say - 7%

This is not a good question.

Note that the survey - unlike the press release and the article - doesn't qualify "abducted or murdered" with "by a stranger". Given how prevalent abusive relationships are, that makes the "million to one" statistic pretty much irrelevant. It's like asking someone whether they're scared of being bitten, and then laughing at them if they say "yes" on the grounds that hardly anyone is attacked by sharks.

The important thing to note is that the survey only gives those three options, plus a few standard "none" options. If the survey was repeated, but with a more substantial list of plausible fears (and perhaps some that the surveyors believed less plausible), then this would give a very different result. People tend to go more for options that they're prompted for, rather than "Other". Of the three options, especially as worded, the third is - regardless of relative risk - considerably less scary.

It wouldn't surprise me if they wrote most of this press release before the survey results were back, and just added the exact numbers later.

So, bad data collection, and then conclusions that aren't really supported by the data anyway, and then reprinted pretty much from the press release without any challenge. Exactly the "standards" I've come to expect from science reporting, in fact.

In 1950, crisp packets contained no crisps

According to the BBC, some researchers have analysed the size of the plates in depictions of the Last Supper between 1000AD and either the 1700s or 1900AD (the article gives both as dates for the most recent painting analysed).

This research is not, as might be expected, found in an interesting journal of art history, but in the International Journal of Obesity. I don't have access to the full text of the research, so I can't tell how good it is. However, the news article is not hopeful.

The Guardian article gives slightly more details, including the source they used to find the paintings (which are mostly Western European and Russian in origin, judging by the Amazon's index).

So, flaws in at least the reporting, and possibly the research too:

Firstly, the differences in style between eras of European painting are ignored. Pre-renaissance painting was largely symbolic, for instance, and the pursuit of a "realistic" style has varied considerably over the centuries. The Last Supper, as one of the major events of Christianity, has a lot of associated symbolism. That starts to make comparisons based on a simple head:plate ratio less valid (though not as invalid as they would be from a larger selection).

Neither of the researchers is an art historian by speciality (Reuters describes what they do specialise in), but without the full text I can't tell if they asked someone who was or not.

Secondly, the use of plates may have changed over that time. Was every plate depicted a plate of comparable function? Without the source material and a better knowledge of the technology I can't tell. There is a risk, though, of comparing side plates to main plates. Similarly, changes in numbers of courses may have had an effect - if the number of courses being eaten increases, the average plate size for a course will probably decrease, and vice versa.

Thirdly, the reporting seems to be linking this to "portion sizes are getting bigger so there's an obesity crisis!!!". Since the phrase "super-sized" appears in all three news articles, despite them otherwise being very different in wording, I'm guessing this was introduced either by the press release or by Brian Wansink.

The major problem with this is that the range 1000AD to 1900AD was not characterised by any sort of obesity crisis whatsoever. It was characterised by major famines, general malnutrition, and other problems related to lack of food (especially for people outside the nobility, and especially towards the start of the period).

As farming and storage technology improved, and as European nations invaded large parts of the rest of the world to bring increased prosperity at their expense, it probably is true that over those 900 years the amount of food available to Europeans generally increased. This may have had a general effect on the size of plates and the amount of food on them in artwork - that's entirely plausible.

The conclusion, however, that some people seem to be drawing, is entirely unfounded. From the BBC article:

Charlene Shoneye, an obesity dietician for the charity Weight Concern, said: "I'm really not surprised by these findings because the size of our plates and food portions has increased.

"Twenty years ago, for example, most crisps used to come in packs that were 20g. Now they are 30g, 50g or even 60g, and we are still eating the whole pack.

"This super-sizing has changed our perception of normal."

But she said it was not too late to reverse the trend and that individuals, society and the food industry should look to smaller portions.

So, this person is not surprised at a general increasing trend between 1000AD and 1900AD because of a separate trend between 1990AD and 2010AD¹? This is fairly typical of some of the "science" behind the "obesity epidemic", but not being able to tell the difference between a trend based on the slow eradication of malnutrition and a trend based on marketing decisions is not a good sign.

Widespread malnutrition probably would stop most people from being above the government-mandated weights, but it's not a great policy idea. The whole "obesity epidemic" panic, though, seems to be predicated on a "wasn't everyone healthier when they were borderline-malnourished" idea - with "so let's put them on diets to induce this state" the obvious conclusion.

It's a conclusion that can only be drawn when you have the privilege not to be able to personally or culturally² remember widespread and unavoidable malnutrition.

¹ Further on the crisps point - yes, the packs do vary in size (you can even get 100g and 150g packs of some brands), but the cost varies with the size. A 50g bag of crisps is more expensive than a 30g bag of crisps. The 30g bag of crisps is probably more expensive (after correcting for general crisp-related inflation) than the 20g bags were.

² There are people, and groups of people, of course, even within the Western European and Northern American countries currently worrying about the "obesity epidemic", who don't have enough food. Society and privilege make them sufficiently invisible to many "obesity researchers", though.

Can the BBC be trusted with a web page?

No.

[trigger warning]

(via Liberal Conspiracy) The BBC Have Your Say forums currently have as a question:

Should Uganda debate gay execution?

Should homosexuals face execution? Yes, we accept it is a stark and disturbing question. But this is the reality behind an Anti-Homosexuality Bill being debated on Friday by the Ugandan parliament which would see some homosexual offences punishable by death.

There are many legitimate questions to ask about the current situation in Uganda. This is not, no matter how many disclaimers are added, one of them.

You can complain to the BBC about this. My complaint letter is:

I can not understand how you can think this is a legitimate question to be asking. There are plenty of reasonable questions that could be asked about the proposed legislation in Uganda, but "should homosexuals face execution?" is not one of them.

There is no way to ask that question that does not imply that "yes" is a legitimate answer. Your "house rules" for Have Your Say contain "Do not post messages that are unlawful, harassing, defamatory, abusive, threatening, harmful, obscene, profane, sexually oriented, homophobic or racially offensive." - when the question itself meets several of those, it sets an unacceptable example.

I expect a full and sincere apology from the BBC for this, and details of what steps are being taken to ensure that similar questions are not asked in future.

Update, 18 December

The BBC claims it has done nothing wrong, and then after that fails to stop complaints, issues a non-apology. I'm aware that the BBC considers non-apologies completely adequate. The National Union of Journalists has also come in on the "mass murder is unacceptable" side of the debate.

I got a reply to my complaint pointing me to those posts, claiming to "understand the concerns I have raised". My reply to that was:

[Neither post] acknowledges at all that:

• it was not an appropriate question to ask
• the problem was not merely with the original headline but with the entire concept for the question

Nor does it provide an apology beyond the "we're sorry if anyone might possibly have been offended, but we're still right", or describe what steps will be taken to prevent a recurrence of this sort of question.

I find it an utterly unsatisfactory response to the complaint that does not make me believe that you "understand the concerns [I] have raised".

Lost in translation

So, the BBC report on research about what skin tones are considered healthiest (for young white people, that is). It must be a time for another instalment of "bad reporting of science". Here's the article's lede.

A golden glow is the healthiest and most attractive look for Caucasian skin, researchers have claimed.

This isn't (of course) actually what the researchers claim in their article (full text, DOI 10.1007/s10764-009-9380-z). For once, it's open-access publishing, so you can read the paper yourself. The distinction between what the researchers say and what the BBC thinks they say is more subtle than usual, but there's a lot to unpack in that difference.

The paper examines what the effects of changing redness, yellowness and lightness on perceptions of health are. (Note that strongly increasing lightness makes the red and yellow changes hard to tell as the whole image becomes oversaturated). There are pictures of this process in the paper, which show the extreme endpoints of the range they considered.

The volunteers providing the face pictures used for the sample, and the volunteers assessing the apparent health after colour transformation, were all white and young. Looking at the ages and colours, I'd guess that they just did the typical academic shortcut of grabbing a bunch of students (and St Andrews, where the research was carried out, is one of the whitest universities in the UK).

The study showed that increasing redness to an amount associated with the reddest end of the sample, increasing yellowness beyond the amount any of the sample actually had in their faces, and increasing lightness (whiteness, in many ways) to approximately the top end of the sample, were all associated with increased perceived health.

So this is where the BBC article first falls down: failing to clearly draw a difference between health and perceived health. The article repeatedly confuses the two (in a way that the paper itself largely avoids) - saying that redder/yellower skin is considered healthier-looking, but then going on to look at reasons why things considered healthy might make skin that colour, which misses out the important step - which the paper doesn't do, but also doesn't claim to do - where a correlation between "skin tone assessed as healthy" and "actually healthy" is looked for.

The second failing is not long after that: the researchers say that health perception is linked to attractiveness (and reference a lot of previous research there). They don't say that the colour range assessed as most healthy is also assessed as most attractive. It's an obvious jump, but it's not one supported by this experiment because it's not what they looked for.

The researchers noticed, unreported on, a significant gender-based difference: female skin was made lighter; male skin was made redder and yellower. This apparently is a pre-existing (statistical) difference in (possibly only white, the paper doesn't say) male and female skin, but the effect of the changes to make the skin appear healthier expanded this difference above its real extent.

There's an unexamined possibility here, too. Perceived health is considered attractive. But there's a lot of socialisation in other ways about what's considered attractive - for women, very to impossibly thin, very white (again, impossibly so for most people) - and a lot of explicit connection between this beauty ideal and health from the diet industry and "beauty" magazines. So, were the sample just pushing the faces towards what they thought of as attractive, because they assumed that correlated more strongly with health? Certainly plenty of things that go with perceived health don't go with actual health (weight, for instance)

There's a collision of a lot of stereotypes around health and disability, sexism and racism here (with a bit of heteronormativity hanging around too, as every study of "mate choice" seemingly has), that I think the study accidentally exposes, and it would be interesting to see if the study was duplicated with different skin colours, or in societies with different cultural views about what "looks" healthy, or even with a wider age range of both faces and assessors.

You shouldn't be able to make a study finding how lighter (whiter) skin is considered healthier-looking and so more attractive without looking at how society considers whiteness more valuable and more attractive and examining how this might be a different cause than simple perceptions of healthiness. But they do, and the BBC doesn't question it either.

Advice men never get

The Times Education Supplement reports on comments made by Jill Berry, president of the Girls' Schools Association, which represents the self-proclaimed leading independent girls' schools.

The advice is the usual "don't expect to have it all", with a secondary strand of keeping ambitions high anyway. It would probably do her pupils good if boys were also taught that "having it all" was not realistic without putting a lot of effort in to balance work and family life, too, but this doesn't get a mention in the article. The independence and ambitions she wishes her pupils to achieve would come more easily if the part of men in making it difficult wasn't ignored.

So that's the gendered differentiation in expectations, which was obvious just from the headlines this story got in the news. There's also a bit of heteronormativity, too. Less obvious just from the headlines is the classism.

A few examples:

[...] said it was "healthy" for girls to aspire to a "flash sports car with a baby seat in the back" [...]

I don't think it's unreasonable to have as an aspiration "have enough money that you can afford the things you need and still have some spare". A "flash sports car" - not even an ordinary everyday sports car - requires a lot more money than that.

They will need to realise that there may be times when they might not want to work, or they might want to take a lesser job because their priorities have changed. [...]

Or, most girls, when they grow up to be women, won't have the luxury of "not working because they don't want to" for much of their lives (and neither will most men, for that matter). They might end up in that "lesser job" (by which I assume she means "less well-paid" as opposed to "less satisfying", though both are true) because it was what was available, rather than because they wanted to do.

"Most women cannot keep all the plates spinning," she said; "sometimes the plates crash."

The dirty secret she refuses to reveal and so implies through omission isn't the case, is that most men can't, either. The difference is that women are expected not only to spin their own plates, but to catch any that fall off the men's poles.

When men are told not to expect it "all" either, then I'll be less critical of this sort of thing.

It doesn't surprise me that the head of an expensive private school doesn't consider the class privilege inherent in their statements. It does show how little critical journalism there is, however, that all the major news sources I've found reporting on this - TES, BBC, Reuters, Independent, Daily Mail so far - have presented her advice as advice applicable to girls in general, rather than the upper-class and possibly upper-middle class girls that Mrs Berry considers representative (conversely, of course, the advice is more applicable to boys than either Mrs Berry or the news organisations think)

Is there any major news organisation out there that doesn't just spend most of the day rewriting and reprinting press releases for the vast majority of its articles?

Bad science, lazy reporting, and a side order of heteronormativity

Some articles, you can read the first paragraph, or maybe the headline, and get an instant sense that it bears no relation to the real story.

So, the BBC is reporting on a recent paper in the European Journal of Operational Research. First mistake: the paper is described as being by a Bath University team. Of the five authors, only one (Fragnière, the second author), is from Bath, so this seems to be stretching the definition a bit. The others are from various Swiss institutions. So, no prizes for guessing the source of the press release.

Second mistake: the BBC says:

The researchers studied interviews of more than 1,500 couples who were married or in a serious relationship.

Five years later, they followed up 1,000 of the couples to see which had lasted.

The paper itself says:

We use data from the 1999 Swiss contemporary family survey by Widmer et al., 2003. This is a representative random sample of 1534 couples, married or unmarried, ages 18–75, residing in the three linguistic areas of Switzerland. In 2004, the authors did a follow-up study on 1074 of those couples to determine how many had separated (Widmer et al., 2006 E.D. Widmer, J. Kellerhals and R. Levy, Types of conjugal interactions and conjugal conflict: A Longitudinal Assessment, European Sociological Review 20 (2006), pp. 63–77.Widmer et al., 2006).

Someone else did that research. They used that research as a basis for their own work. There's a huge difference there.

Both the paper and the BBC, unnaturally, say "couples" when they mean "male-female couples".

The BBC then says:

They found that if the wife was five or more years older than her husband, they were more than three times as likely to divorce than if they were the same age.

If the age gap is reversed, and the man is older than the woman, the odds of marital bliss are higher.

The latter "finding" isn't statistically significant, however.

Add in a better education for the woman - Beyonce has her high school diploma, unlike husband Jay-Z - and the chances of lasting happiness improve further.

The paper really doesn't imply this at all. It finds statistically a significant difference where marriages between people at different educational levels are more likely to last the full five years than marriages between people at the same educational level. The paper doesn't give the 95% significance range, only the value and whether this is significant at the 95% level, so while the five-year stability is increased more for marriages where the female partner is more educated than for those where the male partner is more educated, both are increased above the baseline.

Back to the BBC

Those who have never divorced fare better too. But couples in which one member has been through a divorce in the past are less stable than those in which both members are divorcees.

The only case that is statistically significantly different to the baseline case they chose (neither divorced) is where the male partner is formerly divorced but the female partner is not.

That's three of the four factors discussed. Here's the fourth, which is larger than the age one and doesn't get a mention. From the paper:

Wife and husband are both Swiss (REF)    -

Wife and husband are both from Western countries    3.451*

So, yes, if you want a stable marriage, the best approach is to be Swiss and marry someone else who is Swiss. British readers of BBC news are somewhat out of luck. (Note: this result may not generalise outside Switzerland, but then, the others might not either)

This, incidentally, has all been drawn from the first three chapters of the paper. The press release, and the BBC, seem to have stayed away from the fourth chapter onwards, probably because it then starts to get very strange. There's a slight hint of it in Fragnière's quote in the BBC article.

The researchers constructed a model where they'd take the demographics from the study, and their found weightings for the data, and then swap people around between couples to produce the set of couples with the lowest predicted divorce rate. That didn't make it into the news article, did it. I bet it didn't get as far as the press release before someone in Bath Uni's marketing department thought "this sounds a bit creepy".

To create our optimization model, we assumed a central matrimonial agency with knowledge of all the characteristics of all potential mates, whose mission is to optimize romantic matches.

Okay, maybe more than a bit.

The researchers acknowledge that this swap (99.81% of the couples were found to be "incorrect") is based on a whole bunch of limitations, and might not actually give the same result in reality (they don't suggest trying it, but they do suggest using these factors to give hints in matchmaking services)

There are several practical issues related to our model that we would like to address at this point. This study refers to married, or formerly married couples, and this does not take into account the odds of initially liking each other. In a real case of “marriage agency” the quality of the allocation must also include high probability of liking each other from the beginning. Moreover, in the real world the size of the men and women sets is different, which obviously is not represented in the model since our sample solely contained information about couples. These certainly represent limitations of our model

(emphasis mine)

A major limitation that they've not addressed is the sexuality of these couples. There are probably enough bisexual people within these couples that a straightforward swap of one man for another won't always be "optimal. Of course, that would complicate the mathematical model significantly, so it's not surprising that they didn't try it, but it is disappointing (and also not surprising) that they didn't notice and/or acknowledge this as a limitation.

It's not the first painfully heteronormative study around relationships, and it won't be the last.

So yes, a study of already dubious value, combined with some even worse reporting. Specifically, I think, "copying the press release without verifying it", since the New Zealand Herald, the Daily Mail, and the Telegraph all make similar mistakes (and all use one or both of the same two celebrity couples as examples). I wonder how many of the journalists bothered to use their news organisations subscriptions to check out the original paper (since none of them mentioned the last three chapters, I guess none)

Racism and language

So, Mark Easton at the BBC (warning: the comments are generally worse than the article) complains about the difficulties in talking about race. Now, I admit, that it's a complex subject, and it's one that I know far less about than I should. On the other hand, he seems far too willing to give up on it.

Discussing race in this country is to walk on egg-shells. When even an experienced signed-up multiculturalist like Mr Straw gets caught out, it becomes obvious how difficult it is even to find the language in which to conduct a grown-up debate about it.

I have talked to ordinary voters about the subject, there is often discomfort over terminology. Some fear that using the word "black" might be construed as evidence of racism, so they opt for an even more contentious term - "coloured" - in the belief that this will soften their argument.

A few months ago, I received an e-mail from a Chinese viewer who told me he had been offended by my use of the word "Asian" when what I really meant was people who hailed from the Indian sub-continent. On another occasion I was taken to task for the phrase "non-white" - a shorthand for all ethnic minorities which was deemed insulting.

People generally don't want to offend and the shifting sands of acceptable racial vocabulary mean that many dare not even step into the territory. It is a dangerous domain - one false move and you are branded a bigot.

That really depends what you've just said. There are two types of wrong descriptions of races, ethnicities and nationalities.

• Those which people are just expected to know are insulting. (Non-bigoted equivalent example: "arsehole")
• Those which are archaic or inaccurate. In general, the first time you get this wrong, you'll get a polite correction. The bigot brand comes when you refuse the correction and loudly insist on using it anyway. Conversely, if you have a reputation for politely accepting and implementing corrections, you're more likely to get the benefit of the doubt for different mistakes in future

For either individuals or groups there's a responsibility to make an effort to find out when it becomes relevant what they identify themselves as. In other contexts, people have no problem with this at all - I identify, in terms of nationality, as British, English, Northern, and Yorkshire.

• I don't identify as European, despite this being geographically true, and despite being on balance pro-EU, but I don't mind being described as such.
• I don't identify as from County Durham, despite having lived there for several years, but I might say I'm "from Durham" if asked where I was from.
• Someone from Scotland might want to be pedantic about my identification as Northern, based on the fact that there's about as much of Britain north of here as there is south, but if they called me Southern based on that I would not be happy.

Similarly, someone from Scotland wouldn't identify as English (or even necessarily British), and the archaic term "Scotch" would also be insulting because of its history of use.

Identifying people as they wish to be identified (and noting that this might change over time as previously innocuous terms get used as insults) really isn't that difficult a principle to stick to.

As regards his examples:

• Afro-Caribbean vs African-Caribbean: the woman in the audience said explicitly which one she preferred, including pronunciation. You don't even need to research any of the history behind the two terms at that point.
• Black vs Coloured: a quick glance at the names of various organisations makes this really obvious.
• Asian/Chinese/Indian sub-continent: in the UK, "Asian", as a description of people, tends to mean "South Asian". In the US, it tends to mean "East Asian". So, using a term to mean something other than the obvious meaning of "from the continent of Asia" is going to be insulting to the people you didn't refer to: in the same way that saying "American" to mean "from the USA" gives some Canadians offence (and the entire South American continent even more), and using "European" to refer only to the Iberian peninsula would be considered strange. Even without the insult and disappearance of entire countries, it's an imprecise term that a journalist has no business using to mean "Indian sub-continent".
• non-white: describing people, extremely broadly, as "not in the privileged group" makes the various differences between those people disappear, and problematically sets "white" as a default. There are relatively few cases where this sort of term is needed: in the UK, BAME (Black, Asian, and Minority Ethnic[ities]) can be used instead as a more positive identifier (like LGBTQIA, variants with fewer letters also appear).

I can't help feeling that Mr Easton using the corrections as a learning experience, and doing some research, might have been more useful than complaining about how difficult it was to talk about race nowadays.

The press, and the accountability of the press

So, large parts of the UK internet have been rightly angry at Jan Moir's Daily Mail article. The Press Complaints Commission received over 21,000 complaints about the article, which apparently is more than they normally get in several years - the objections to the article have been widespread enough to get their own stories.

Meanwhile, the Daily Express publishes an attack on people with mental ill-health, and the Daily Mail and the BBC publish some more transphobic reports. Neither appears to have been noticed, much. Nor do many of the other pieces of bigoted reporting - whether transphobic, sexist, ablist, racist, homophobic, fat-hating, or otherwise - that can be found published by most news organisations on most days of the week.

Moir - horrific though their opinions and subsequent attempted defence were - seems to have become the target of so many complaints by attacking a recently deceased popular celebrity (and their family), rather than because of their bigotry. What gets noticed on the internet is often largely a matter of luck - good or bad - which of course means that the majority of these attitudes never receive much mainstream challenge.

Prejudice as accepted fact

Sunny at Liberal Conspiracy comments on the recent controversy over Andrew Marr's questioning of Gordon Brown. [don't read the comments]

Andrew Marr, for the BBC, asked:

A lot of people in this country use prescription painkillers and pills to help them get through; are you one of those people?

The Prime Minister, unsurprisingly, refused to answer the question.

Sunny criticises the way that this "legitimised a smear" that's been going around for ages, but I think this misses the point.

The most irritating thing is that the entire controversy over it is based on a very dangerous chain of illogic. There have been plenty of people leaping in to deny that Gordon Brown takes painkillers, and to criticise the people who've suggested it. This is entirely expected, all part of politics, etc. What's getting lost in this is that it shouldn't be an issue in the first place.

In the 2008 US Presidential campaign, Barack Obama was accused by his opponents of being "a Muslim". This was denied strongly and repeatedly refuted by numerous people - but very few people thought to add - or start with - "and so what if he is?". Colin Powell is the only major US political figure I can remember raising that point, though there may have been others.

It's the same issue here - Andrew Marr was undoubtedly entirely in the wrong to ask a question about Brown's medical history: there are very few questions a journalist can ask about anyone's medical history, and none of those possibilities applied here.

The rumours that have been started about Brown's medical history and mental health are undoubtedly intended by their initiators as smears. They really shouldn't be - it should be as pointless a question as "So, I hear you ate a carrot yesterday, Prime Minister. Could you confirm or deny this rumour?"

It works as a 'smear' because enough people vaguely believe a variation on one of these two clearly false statements:

• People with chronic pain should not be Prime Minister. Our Prime Minister needs to be an idealised specimen of humanity (and, you know, straight, white, male, and other defaults would be good, while you're at it)
• People with chronic pain should not take medication for it, and doing so makes them unsuitable to be Prime Minister

You can substitute for chronic pain any of the other numerous mental health problems people have suggested Brown might have (most commonly depression, something which the generally popular and successful Prime Minister Churchill also lived with). I'd certainly rather any politician with health problems, mental or physical, was receiving appropriate treatment for those problems. Laurie at Penny Red covered this well the last time it came up (which, note the post dates, was 3 weeks ago...). Going back a bit further, Charles Kennedy was forced to resign as leader of the Liberal Democrats over his alcholism, and has since been replaced by various people who are neither as competent nor as popular with the public. Good going there, Lib Dems.

The people responding to this and criticising the "legitimisation of a smear" need to be more careful not to legitimise the idea that it would be bad if it was true while they're at it. Breaking that illogic is the only long-term way out of this.

Permanently marked

The news articles on this story are generally so full of poor assumptions and ablism that I don't have a good one to link to. The BBC one isn't quite as bad as the others, so: a man has been removed from the London taxi training course. The facts of the case, as far as I can tell through the reporting, are this:

[trigger warning]

In 2000, while living with paranoid schizophrenia, he strangled his wife. He pleaded guilty on grounds of diminished responsibility, and was sentenced for manslaughter to indefinite imprisonment in 2001.

Some time after this, the parole board assessed that he was no longer a danger to society and he was released. In early 2009 he worked for a private hire company in London, and later on began training for the London black cabs (which have a higher standard of training required).

Transport for London's guidelines, at the time, allowed people with 'spent' convictions to train with them. When the press discovered, via a leak, that this man was taking this training, they raised a giant fuss. TfL then disallowed the man to continue with the training, and announced that it would change its licensing guidelines to exclude people with spent convictions for violent offences except in exceptional mitigating circumstances.

What no-one seems to be generally pointing out in the mainstream press, though, is that this is surely precisely one of those circumstances. He was under the effects of a severe mental illness at the time he killed his wife. The court at the time accepted that this was the only reason for his actions in allowing the diminished responsibility plea (which is really quite rare). He has since been assessed as healthy by the parole board.

Let's take an example with a physical illness. Someone gets a highly contagious, highly dangerous physical illness. A family member catches it and dies, but they survive after extensive treatment in hospital. Having recovered, they apply for a job as a taxi driver, but are turned down because a previous health problem once led to someone's death. The logic being used, of course, is "what if they're wrong about being healthy?" to which the answer is "and where did you get your license to practice medicine again?".

Of course, it's in all the press, and the existing taxi drivers were up in arms about it, so that sort of thinking is never going to be possible in this case. TfL had been trying to be reasonable about it...

[a few days ago] Jeroen Weimar, chief operating officer at TfL, said: [...] "We have got five to six years of independent medical advice saying this man is no danger to the community and fit and proper to be a taxi driver or a minicab driver."

...but have since presumably calculated that it's better to stop taking hits to their reputation than to support people with previous mental health problems. It's appalling that the two things are contradictory.

Terrible article on weight

I read BBC News a lot, and in general, my thoughts about its stories are more about how bad the events described are, not how bad the reporting is. This is not one of those times.

BBC News's science (and related fields) reporting is one of the weakest aspects of the operation, and this article about perceptions of obesity is a good example of this.

Let's start at the top

The poll, carried out by YouGov for Slimming World, found just over a quarter of 2,000 people questioned had measurements which would place them squarely in the obese camp.

So, no reason that a survey about weight for a magazine that exists to promote diet plans might in any way be slightly biased. This potential for bias never even gets a mention in the rest of the article.

Over half of those deemed morbidly obese believed they ate a healthy diet, while more than a third of the overweight said they had never tried to shed the pounds.

Given that it's extremely possible to be fat and eat a healthy diet (or indeed, as I was as a teenager, very underweight despite eating numerous large meals), that's not as unlikely as it's being framed as. I also wonder what proportion of that third are women, since "overweight" BMI ranges are more socially acceptable in men (and arguably less noticeable at all, given that the cultural images of women in the UK are generally extremely thin).

It also doesn't say, though I expect the survey collected it, how many of the "underweight"¹ people have also dieted in an attempt to get thinner still.

While our life expectancies have increased at the same time as our weight, the consensus now is that cases of diseases such as diabetes and even cancer could be reduced if everybody strove to be within the "normal" Body Mass Index (BMI) range

Okay, so life expectancies have increased as weight increases (and bear in mind that the increase in weight averaged across the whole population has been small). Therefore, we should reduce the weights of people to increase life expectancies even more.

That's not particularly convincing. In the US, there's a study called NHANES, which is a huge survey of health and diet. It's been carried out three times, and in each case, the people studied have been the subject of a later follow-up study. One of the basic checks in the follow-up study is whether or not the person is still alive.

There's a 2005 paper in the Journal of the American Medical Association called "Excess Deaths Associated With Underweight, Overweight, and Obesity" (copy via "Alas, a blog"). It takes the 18-25 BMI category as a baseline, and compares that (controlling for age and other obvious factors, such as smoking) with four other categories:

• Less than 18 ("underweight")
• 25-30 ("overweight")
• 30-35 ("obese")
• More than 35 ("also obese, but more so")

It's a simple test: check what proportion of the 18-25 BMI range have died since the original study, and then see what proportion of the other ranges have died. If the arguments about fat being dangerous are true, then the heavier categories should have more deaths - the "Excess deaths" referred to in the title of the paper. For shorthand, I'll refer to categories as being "more dangerous" or "less dangerous", if they have excess deaths or fewer than expected deaths respectively.

There are some useful graphs at the bottom of the second PDF, but in summary:

• In adults between 25 and 59 years old, there is no statistically significant difference between the 18-25 range, the 25-30 range, and the 30-35 range. The 35+ range is associated with an increase in the number of deaths (though not for people who have never smoked). The 18- range is also not statistically significant.
• Between 60 and 69 years old, the 18- range becomes statistically significantly more dangerous. The 35+ range remains statistically significantly more dangerous than the 18-35 range, but by a lower margin.
• At 70 years old and over, all ranges except 18- are statistically the same as the baseline. 18- is again more dangerous

It's worth noting further that even the statistically significant changes are relatively small - even the most dangerous category (BMI 18-, age 60-69), is only between 2 and 3 times more dangerous. On a similar measure, smoking is around 20 times more dangerous. Conversely, probably the safest category (BMI 25-30, age 25-59) is between baseline and 33% safer.

The study also notes that the older data from NHANES 1 has a higher death rate. The more recent data from NHANES 2 and NHANES 3 shows no statistically significant increase in death rates for any category except 18- (in NHANES 3 only. Conversely, NHANES 2 shows a statistically significant decrease in death rates for the 25-30 category.

So, essentially, differences in BMI are unlikely to change your chance of dying in a given time period much at all, especially if you don't smoke (unless you get old, in which case it starts to get somewhat dangerous if it drops too low, but even then not by much).

This bit of the last excerpt needs more comment.

[...] if everybody strove to be within the "normal" Body Mass Index (BMI) range

There are multiple major problems with this sub-sentence alone:

1. BMI is a horrible measure of healthiness-of-weight. It was designed as a statistical measure of weight within populations, not as something that could be applied to individuals. The "normal" category comes from the late 19th century, which had far more problems with malnutrition than the present day. It doesn't distinguish at all between fat and muscle mass (most athletes are "obese" according to a BMI scale). It doesn't do anything different for body type - it's designed for white male Europeans of average build.
2. No-one has managed to find a way to change the long-term weight of an individual to a different selected weight. Short-term, starvation will reduce it and overeating (and we're talking 10,000 calories a day, not just an extra portion of chips) will increase it. Long-term, this changed weight will revert back to the previous norm for that person in almost all cases. This is like saying "if everyone strove to be taller, we wouldn't need to buy so many stepladders". It's not true, and even if it was, it wouldn't mean people could get taller.
3. Diets and other weight-loss interventions can be very bad for health in and of themselves. "Yo-yo" dieting (a natural consequence of telling people that they have to lose weight or else
4. "normal" has been picked arbitrarily based on what was average over a century ago. You'd expect that with malnutrition being reduced in European countries since then that the mean and median weights would rise... in fact, it would be worrying if they didn't.
5. (I've probably missed at least one, but this is enough...)

The focus on the extreme in television documentaries about the very large but also in the pictures that are chosen to illustrate articles about obesity have also been held up as another potential culprit.

A few points to the BBC for identifying the "headless fatty" pictures as not really depicting what the 'average' "obese" person looks like. Minus several million points for going ahead and using one at the top of the article anyway (it cuts out quite a lot that's not the head, too).

The article goes on to criticise the negativity around fat people, and mentions the abuse that fat people (including children) receive, but only identifies this as a problem because:

"All the discussion around overweight children is so negative that it is not surprising parents find it difficult to acknowledge there is a problem. It's a defence mechanism," says Dr Susan Jebb of the Human Nutrition Research Laboratory of the Medical Research Council.

So, it's only a problem that fat people are subjected to abuse for being fat, and told that being fat is bad, because it means that we can't tell them that being fat is bad in a detached, emotionless, scientific manner. Glad that's sorted, then.

It terrifies me that this is the quality of the debate that's going on about the (probably non-existent) "obesity crisis". So far, the only actual outcome all this worry has appears to be a reduction in the quality (but not the length) of life of fat people from all the abuse. Oh, and a great profit for the dieting industry. Keep commissioning those surveys, then.

Finally, missing from the article, any opposing view at all. The BBC science reporting does have a tendency to reprint press releases uncritically. Not every article needs to have an opposing view - and the media more generally certainly has a problem with looking too hard for them - but this is a news organisation that's run positive reports on research like the research I've quoted.

¹ "underweight" and "overweight" are terrible terms in themselves, since they imply an "ideal" or "normal" weight and are generally used to imply that this "ideal weight" is the same for everyone (after 'correcting' for height). More on "underweight" and "overweight" at Alas, though there doesn't seem to be any consensus on what the "unmarked" category should be called. "medium-weight" or some variation sort-of works, though the variations on "government-mandated standard weight" make the point that normal/unmarked/medium has already been set arbitrarily.