How to read a paper

January 29th, 2011 by Ben Goldacre in bad science, mail, statistics, sun | 49 Comments »

Ben Goldacre, The Guardian, Saturday 29 January 2011

If science has any authority, it derives from transparency: you can check the claims against the working. Sometimes you hit a brick wall. Sometimes you might consider a shortcut. Let’s look at 3 types of checking.

First up, in the Sun, a child has been born at 7:43, just like their two siblings (though one was in the evening). The Sun says the odds on this are 300 million to one. This is an easy thing to check, because the information is all there in the paper.

The Sun are wrong. There are 60 minutes in each hour, 12 hours on a clock, that’s 720 minutes. The first child can be born in any minute – we’re not interested in the chance of 3 children being born at 7:43 more than any other minute, just the chances of 3 being born in the same minute. So once the first child is born, there’s a one in 720 chance of the next child matching on birth time, and if that happens, then a one in 720 chance of the next one matching too. 720 x 720 makes the overall odds of 3 matching birth minutes 518,400 to one, not 300 million.

Since there are 167,000 third or more-th children born in England and Wales each year, you’ll see this coincidence once every three years, more frequently if you include the rest of the kingdom, and even more frequently if the midwife squints at the clock and says: “Oh, was the last one born at 7:43? Well…”

Our next case takes more elaborate checking, since it involves an experiment and its interpretation. Scientists at Lancaster University, say the Daily Mail and the BBC, have devised an amazing piece of paedophile identification software. It reads your messages and decides if the person you’re chatting to on the internet is another young person, or an adult who is pretending to be young.

This is a tricky problem to solve on a handheld device, or indeed anywhere. There is a press release on the Lancaster University website explaining that this device has been studied and found to work. I asked for details. The methods and results of this study are secret. No paper has been submitted for publication.

So actually there’s no complicated interpretation problem here: nobody can know what these scientists measured, how they measured it, what the numbers were like, how closely the experiment mirrored a real world situation, or anything at all. When the Raelian cult said they’d cloned a baby, but we weren’t allowed to see it, nobody took them seriously. Until someone’s willing to tell me what they measured and how they measured it, they might as well be Raelians.

Is this flippant? We live in a big world, filled with amazing scientific work to read. It can also be overwhelming, and you need someone to walk you through the forest. This brings us on to our last form of checking, the trickiest of all: how do you know if someone has fairly represented the findings of an entire field, or simply cherry picked the results that suit them, to build a story?

Zoe Harcombe sells diet books. This week in the Daily Mail she was explaining that fruit and veg are actually no good for you. There’s a fascinating conversation to be had about the evidence base on the relationship between diet and health: would you start with Zoe’s work?

We all rely on heuristics, or shortcuts. Trusting an authority is one. Zoe boasts in the Mail that she is “studying for a PhD in nutrition” but she admitted to me, tediously, inevitably, that she’s not registered for a PhD anywhere (although she is thinking about doing one in the future).

Does it matter? We read a precis of research as a shortcut, but once you lose trust, to double check whether someone has fairly represented an entire field, you’d have to read that field’s entire canon, and after many years of  work, whatever your other conclusions were, the strongest would be that any timesaving benefit from reading a precis has plainly been annihilated. Given that this is the case, I know it’s harsh, and you may disagree, but in a busy world, I’m not sure I see the point of a Zoe Harcombe.

Ben Goldacre, The Guardian, Saturday 29 January 2011


++++++++++++++++++++++++++++++++++++++++++
If you like what I do, and you want me to do more, you can: buy my books Bad Science and Bad Pharma, give them to your friends, put them on your reading list, employ me to do a talk, or tweet this article to your friends. Thanks! ++++++++++++++++++++++++++++++++++++++++++

49 Responses



  1. Rhys Morgan said,

    January 29, 2011 at 1:23 am

    Ah!
    Zoe Harcombe. I knew I recognised her name and ideas from somewhere. Turns out they were from her – bit.ly/hQEAZZ
    As you can see, this isn’t the first time she’s been in the Daily Fail, and I doubt it’ll be the last…

  2. Statto said,

    January 29, 2011 at 2:37 am

    The Sun is descending into self-parody: from Friday’s paper, In the family weigh begins ‘MUM Nerys Evans scaled amazing odds when her three children were born at exactly the same weight. Son Rhys, now ten, daughter Seryn, two next month, and four-month-old Nya were each precisely 8 lb 6 oz when they came into the world.’

    Assuming that the mean birthweight is 3.6 kg with about a 0.5 kg standard deviation, I get odds of about 2% for a child weighing 8 lb 6±0.5 oz. This makes the odds of her second and third children weighing the same as her first a WHOPPING 1 in 2,300, probably minus a bit because birthweight to an individual mother may be more tightly distributed.

    This is definitely news.

  3. JonathanE said,

    January 29, 2011 at 7:30 am

    As Universities do more commercially orientated research there is every chance that secrecy will surround findings.
    Of course for most universities the bulk of their funding will still come from government via undergraduate fees and for some via research funding, so there is an argument to be had about the amount of disclosure we should expect. This might be very important when they trade on the fact that they are a university to imply that people should take a cognitive shortcut and trust their findings.

  4. misterjohn said,

    January 29, 2011 at 10:40 am

    The link in para 4 isn’t working.

  5. TwentyMuleTeam said,

    January 29, 2011 at 11:49 am

    Oh, I get it. She’s selling product. ‘Nuff said. theharcombediet.com.

  6. mesmer said,

    January 29, 2011 at 1:10 pm

    I’m not a dietician, and my attempts to understand the complexity of the subject seem to turn my brain to the consistency of the mushy peas that I should probably be avoiding. I have not read Zoe Harcombe, but did find her (DM article)interesting, especially in relation to the evidence to support 5 F&V per day. It is unusual to read criticism of this, and almost heretical to question the mantra.

    See also this link from ‘spiked’
    www.spiked-online.com/index.php/site/article/8869/

    Apologies for the slight digression; I am fascinated with friends who seem to be on the verge of a nervous breakdown (leading to risk imminent death within the year) because they have only managed 4 by midnight. Is anyone able to point me in the direction of other studies on this (5 a day) subject?

    Thanks

  7. tug said,

    January 29, 2011 at 1:28 pm

    The Lancaster people have a list of “publications” www.comp.lancs.ac.uk/isis/index_files/Publications.htm

    They seem to be all conference proceedings rather than peer reviewed articles.

    Seeing as the research behind this “product” was paid for by you and me (via the EPSRC / ESRC) is seems unreasonable to withhold the details from the public.

  8. Chris Bailey said,

    January 29, 2011 at 1:45 pm

    Even if the Sun were quoting the odds of three children being born at the same time, AND that time being 7:43 then their numbers are still off by 20%: 720*720*720 = 373.25 million

  9. Doormat said,

    January 29, 2011 at 2:17 pm

    @tug: You should be a little careful– my understanding is that in computer science, it is very common to publish in conference proceedings (unlike my field, mathematics, where if you had a good result, you probably wouldn’t choose a conference).

    Actually, amusingly, a workshop I went to, run by the EPSRC, had this project as an example application. I remember thinking that it seemed very unlikely that this was technically feasible– guessing the age of someone from how they wrote. I seem to recall that most of the other, young researchers were equally skeptical– but the referee reports were good, and the people from EPSRC were keen, and in real-life (we were doing a mock-up of how decisions were made, with real, past applications) it was funded.

    As you say, given this is taxpayer funded, it really should all be openly available. (And I think I asked the question– if it’s allegedly soooo useful to, say, the police force, why couldn’t they pay?)

  10. daven said,

    January 29, 2011 at 5:19 pm

    Since I have seen young people being able to detect when an older person is pretending to be their age, it would not be impossible to get a machine learning algorithm to detect the same difference.

    I would expect it to be worse at this task than a teenager, but better than the average 8-year old: based on what the young people in the HUWY project have told us (huwy.eu/ – still in progress, hence no papers yet).

    As for the conference papers, it looks like they have written up the early stages of the project, but haven’t yet written up their final outcomes. That is typical in computer science, where a lot of work gets done in the time between the close of a call for papers and the conference itself.

    There is a question on whether this is a suitable topic for a spin-off company, rather than releasing the algorithms for everyone to use in Free and Open Source Software. Is it ethical to delay reducing somewhat a risk to children in order to make money? Of course, they may release the details after a delay to get the business started, which is a common compromise.

  11. leeb said,

    January 30, 2011 at 10:06 am

    Disappointing then to see the esteemed Gaurdian repeating the same Sun story on the back of the family section

  12. linger said,

    January 30, 2011 at 10:29 am

    I think it would be possible to achieve quite high levels of accuracy in estimating writer age (at least to the resolution of “adult/non-adult”) from a trained neural net model. However, that approach can tend to result in a black box mechanism, which does not lend itself easily to explicit description of the method used by the final (trained) model. Given likely applications of the software in tracking pedophiles, there may also be valid reasons for not publishing the specifics of the features used by the model. But even so, some skepticism is warranted unless we are given a more detailed account of:
    (i) what type of result was counted as “successful identification” (e.g. were only true positives counted, or were false positives also taken into account in assessing accuracy? Does the application report a binary decision, or instead a probability that the writer is adult? If the latter, what probability level is used as a measure of “success” when the writer is in fact adult?);
    (ii) the actual quantitative results of testing the model on new data not in the training set;
    (iii) the limitations imposed by the nature of the input used to train and test the model (e.g. is it valid only for British English?).

  13. Paul said,

    January 30, 2011 at 11:02 am

    I see that, at the back of the Guardian Family section, the odds of three babies each born at 7:43 were reported as 300 BILLION to one.
    Nice one, Guardian

  14. Paul said,

    January 30, 2011 at 11:03 am

    Oops – sorry leeb – you’d already spotted it

  15. 5ynic said,

    January 30, 2011 at 11:09 am

    In fairness, Zoe makes reasonable points about fructose and fats. Totally wacky to claim that there are no worthwhile nutrients in fruit though. Studies contradicting that are two-apenny (search any pubs DB for [fruit] + cancer”).

  16. Joe R said,

    January 30, 2011 at 11:14 am

    Quite agree with Doormat first point – thought there are some non peer-reviewed conferences in computer science they are very rare – it’s come as quite the surprise to me to find that there is another way of doing things…

    @Linger – they are not estimating writer age from a large corpus of normal writing, they are detecting if someone is pretending to be a different age from a tiny selection of words, which is vastly harder. If this work was actually accurately doing what the press release said then it would also be able to make a decent stab at the problems of spam email, identity theft, and plagiarism as a side effect

    I’m not convinced that all details should automatically be freely available in general – there are lots of issues there – but certainly I can see no reason to keep the experimental setup and the results secret.

  17. Andrew Clegg said,

    January 30, 2011 at 12:10 pm

    @tug (comment 6):

    In the computer science world, conference proceedings are the primary form of publication (in certain fields at least) and are frequently peer-reviewed. Some even have more stringent peer review procedures than many journals in other sciences (e.g. double-blinding).

    Having said that, I don’t see the most prominent conference in natural language processing (Association for Computational Linguistics) in the list.

    I fully agree that it’s unethical to not make the results of public-funded work freely available though.

  18. chris lawson said,

    January 30, 2011 at 12:53 pm

    @5cynic — I know you’re not defending her, but I’m afraid I don’t even agree that Harcombe makes good points about fructose and fat. She has grossly simplified the biochemistry of fructose then calls it “biochemical fact.” She claims that sugars are oxidants when fructose is a reducing sugar, that is, it is an anti-oxidant. She claims that we don’t need to worry about potassium intake as “it is found in water.” She grotesquely distorts dietary guidelines by pretending they say “fat is bad for you” when all the guidelines I have seen recommend a good intake of mono- and poly-unsaturated fats and a modest (but not zero) intake of animal fats.

    More? She twice refers to studies without naming them. One of them was JNCI J Natl Cancer Inst (2010) 102 (8): 529-537. doi: 10.1093/jnci/djq072 and does not exactly agree with her description (they conclude, “our study supports the notion of a modest cancer preventive effect of high intake of fruits and vegetables and we can exclude chance as a likely factor” — and the fact that the recruited cohort could be as young as 25 would tend to mask any protective effect). The other study (Eur Heart J (2011)
    doi: 10.1093/eurheartj/ehq465) showed a 22% reduction in fatal heart disease — and they excluded anyone with a history of heart disease and did not count non-fatal MIs and strokes. For these (and other technical reasons), the preventive effect is probably higher in the general population than in this study cohort. What Harcombe has done is to take the caveats expressed by researchers as an excuse to dismiss their findings entirely (while mentioning no caveats for any of the “facts” she likes to report).

    She invokes ridiculous conspiracy theories (apparently the NCI and every dietitian in the world were bought out by fruit companies at a meeting in 1991!). Her argument about dietary fibre is too painfully idiotic to recount.

    The clincher is her ultimate Stupid Dietary Advice claim that she does not want to “put chemicals into her body.” I can’t see how she’ll live more than a few days if she’s true to her word.

  19. its2am said,

    January 30, 2011 at 7:35 pm

    Local Phoenix AZ Prime Time news: “Green Lantern wins box office receipts”.

    Hum.

    Research (2 min): Only two movies opened that weekend. One scored 28% and Green Lantern scored 41%. Although the statement was factual the intent is obvious. And just being “news” is statement enough.

    Also, bookending stories can be used to inflate or deflate the middle story.

    Same Night:

    1. “The car JFK’s corpse was transfered in after his assassination is up for auction and the car may not be the real one used”.

    2. “Gas prices are rising at an alarming rate”.

    3. “Foods that give you a tan naturally without the sun”!

    See? These story line up’s are not chance or accidental. Staff gather in a room and specifically put these stories in order for a reason. I believe while people are laughing and responding to each other about the bizarre JFK story the middle story turns into background noise and goes unnoticed and while the mind readjusts it is thrown into another comical spin from the tanning story.

    Omission is the worst kind of travesty perpetrated onto American Society. Did any of you know that in Spokane Washington a bomb with the potential to kill hundreds was found on a M.L.K parade route? This should have been huge, a blockbuster. Spokane is not a small town, it is the second largest city in Washington State.

    www.examiner.com/political-buzz-in-national/mlk-parade-bomb-spokane-called-worst-device-ever-seen-this-country

    Contrition by Omission.

  20. lex-man said,

    January 30, 2011 at 7:53 pm

    @Linger: Even if a neural net was employed details such as the data used as inputs (typing speed, words used, etc), how many outputs the net had how the output was interrupted could be given. Also the way the net was trained could be given, GA training or backwards propagation for example

    Also how the system was tested could be explained easily and the way the level of accuracy was determined could be explained.

  21. bagpuss7 said,

    January 31, 2011 at 8:56 am

    What concerns me, is that in the same way trumpeting “passive smoking” deaths diverts attention away from the worrying increase in childhood Asthema, this kind of stuff diverts attaention from the real causes of the increase in weight in our society.

    While I am fairly sure that too much food and not enough exercise are a big part of the problem, if you were to show with solid research that, for example, incessant food advertising was part of the problem this would upset a lot of major political donors, sorry, food producers.

    From about 8pm onwards you get at least one food advert every twenty minutes on most channels. It doesn’t help.

    OK so I’m a cynic.

  22. heavens said,

    January 31, 2011 at 4:33 pm

    If the “paedophile detector” actually works, then I’m not sure that publishing its inner workings is such a good idea. Methods of avoiding detection is (I’m told, by people whose jobs are to find them) a significant discussion topic within that group. If you published a paper saying “Real children have more typos [one point per transposition, two points for commonly misspelled words], forget to use commas [one point per skipped comma], and never use a semicolon [minus twenty for every semicolon not part of a smiley face]“, then you would doubtless see paedos mimicking this the next day.

    Really, it’s no different than being thoughtful about how you discuss exploitable holes in operating system security—where the model that does the least harm to innocent victims is to patch first, and talk about the details later.

  23. Tim OLeary said,

    January 31, 2011 at 4:35 pm

    That software would be really handy if you were a paedophile.

    It would enable you to find out if you were being entrapped by a police person posing as an underage target. Which in the majority of prosecutions has been the main way charges and convictions have occurred.

  24. Ben Goldacre said,

    January 31, 2011 at 6:24 pm

    heavens: nobody has asked the Lancaster paedophile detector people to publish their inner workings, merely for their evidence THAT it works, not how it works.

    Tim O’Leary: that is an extremely, extremely, extremely good point.

  25. magicdairyfairy said,

    January 31, 2011 at 7:50 pm

    Because I am a maths nerd, I complained to the press complaints commission about the very similar story that was in the Express (I think) fairly recently about some mother beating odds of “50 million to one” to have three babies born on the same day. They agreed with me that this was nonsense and made them take the story off the website. But still, why is this news in the first place?

  26. Andy.R said,

    January 31, 2011 at 7:57 pm

    I also watched the Lancaster Paedophile Machine’s claims of “90% success rate” on BBC news and heard little alarm bells going off. I emailed, asking for published data. Obviously they didn’t reply. But what always cheers me up is watching “Internet Safety Consultant” Jon Taylor saying that within 6–7 seconds after introducing yourself as a 12 year old girl, you would be groomed “without fail”.

    With claims like that, you gotta laugh.

    Link is here: www.bbc.co.uk/news/uk-england-lancashire-12265244

  27. Ben Goldacre said,

    January 31, 2011 at 10:03 pm

    hi andy

    thanks for that jon taylor link

    he really is an expert on internet safety

    apparently there are hundreds,

    “if not thousands”

    of chatrooms in the uk.

  28. jenniferbridge said,

    February 1, 2011 at 8:38 am

    Slightly random question – but sort of related to the diet above. I’ve got a a friend who is a former anorexic / bulimic and she is evagelising about the Allelujah diet (front cover “Eat God’s Way”) it’s all about eating no meat, law fruit and veg. Firstly Jesus ate fish (feeding the five thousand) and secondly he ate lamb at the last support. My plea is “CAN ANYONE SEND ME RANDOMISED CONTROL TRIALS OF THE ALLELUJAH DIET?” Please?

    Thanks

    Jenn

  29. jenniferbridge said,

    February 1, 2011 at 8:38 am

    * ignore typos pls

  30. gestr said,

    February 1, 2011 at 1:55 pm

    Further to what Tim O’Leary said, a far more disturbing application of such a programme – providing it works – which again would require no knowledge of its inner workings, just knowledge of the police´s faith in its ability, is the possibility of using it to hone one´s ability to pose as a child.

    All a paedophile would need to do is download the app, use it on themselves, and see if they could bypass the software. In that case, any child using the software would be an even easier target.

    Very worrying.

  31. AJH said,

    February 1, 2011 at 11:17 pm

    I was horrified by the TV coverage I saw of the “paedo-detector” which was entirely uncritical and claimed 100% success rate. Lots of back slapping, smiling kids and an absence of any sensible advice, discussion of drawbacks or limitations and no real evidence. I thought immediately of Jim Wightman’s Nanniebots covered here some years ago (and debunked, like I think this will be). Even if it works to a degree, the false sense of security it engenders in users and parents will render children less safe, by encouraging them to trust false negatives.

  32. SPP said,

    February 2, 2011 at 2:57 pm

    As several have pointed out, in many areas of computer science conferences are the primary route for disseminating research, and most are peer-reviewed – so in itself this should not be a cause for criticism.

    There are plenty of things to criticise, however, regarding transparency of scientific method. Perhaps we should give the benefit of the doubt here – maybe pressures from research councils to engage in “outreach” has brought things into the public domain prematurely.

    One thing that does cause concern, though, is their quoting a “success rate”. Binary classification (of which this example is) always trades off false-positive probability with the false-negative probability. Quoting a single “success rate” is pretty meaningless.

  33. Amorvincit said,

    February 2, 2011 at 3:06 pm

    My suspicion is that birth time is non-random, an assumption underlying even the reduced odds of 518,400 to one. Aside from “so many hours after a shift change/rounds at which someone is most likely to decide on induction”, Jolly (1972)* claims [in the abstract] “More human births occur between 1 and 6 a.m. than at other times of day. Some prosimian species give birth by day, many Old and New World monkeys by night, while apes apparently have random birth hour. Human birth hour is probably still subject to selective pressures.” Of course I can’t read the original paper unless I pay $26…but I think I trust Jolly (although is that 1-6am just in western hospitals?). This would put the births in question 2-4 hours outside the peak time but presumably the odds within the peak time would be significantly less than 518,400 but a distribution would be required to enable calculation of the odds.

    BTW – I am not sure that the statement “you’ll see this coincidence once every three years” is helpful without some rider like “on average”. Otherwise I think it is the equivalent of saying “a fair coin will come up heads every other flip”.

    *Jolly A (1972) Hour of Birth in Primates and Man Folia Primatol 18:108-121

  34. msjhaffey said,

    February 3, 2011 at 10:28 am

    As a matter of interest, Ben (and others), have you ever
    tried making a complaint to the Press Complaints Commission about
    newspaper reports like these? Their rel="nofollow">Code of Practice states in paragraph
    1(i) The Press must take care not to publish inaccurate, misleading
    or distorted information … which in theory at least appears
    pretty unambiguous.

  35. MattCh said,

    February 3, 2011 at 3:16 pm

    With regards to the pedo app – is this article actually accurate though???

    When I look at the press release it states that the App is developed by a company rather than Lancaster University. It might be a spinout, but that doesn’t necessary mean the University has any involvement in what it does.
    Likewise at the bottom it says that the work is funded by the NWDA. Digging around on the NWDA website I found this www.nwda.co.uk/news–events/press-releases/201001/child-protection-for-msn.aspx

    This implies it is an industrial development, so it is not suprising that specific details aren’t being released.

    With regards to computing publications, as others have pointed out, conferences are the normal venue for academic work. Likewise, results are likely to be the very last thing that are published. I’m not sure it would be possible to have an evaluation paper published with out also desribing how it worked (or pointing to somewhere else where this is described) – in this case, I can’t really see how it can be properly evaluated without knowing how it works – presumably it is the approach that needs to be evaluated???

  36. George999 said,

    February 4, 2011 at 8:08 am

    I think everyone, even Ben, is being a little unkind to Zoe
    Harcombe. Her ‘reasearch’ isn’t entirely original, because a huge
    amount of it is a rehash of that of Gary Taubes and Barry Groves.
    If we take Gary Taubes, he is a huge sceptic of ‘received wisdom’
    in science and his book Good Calorie, Bad Calorie challenges the
    appallingly bad science that forms the basis of government dietary
    dictats. His book is something that I’m sure Ben would approve of.
    Zoe is correct in challenging the promotion of ‘five a day’. There
    is no scientific background for it, it was merely adopted by the
    Food Standards Agency simply because the US Department Of
    Agriculture (sponsored by cereal growers no less!) adopted it. Why
    does Ben attack Zoe when she is merely stating that the Food
    Standards Agency hasn’t done any research itself? It is okay for
    the FSA to tell us we need to eat 5 a day without any scientific
    research, but its not okay for anyone to challenge it. To my mind
    THAT is bad science, and Ben has fallen into that trap. On the
    subject of fruit and veg, and five a day, fruit and veg are not one
    and the same. Green veg yes, absolutely fine. Starchy veg, take
    care if your insulin system can’t handle it. Fruit, be VERY wary
    because of its fructose content. Fruit and veg are nice, but
    they’re not essential. If you hold it dear to your heart that they
    are, then you probably need to question your own science. Why do
    you believe it? Probably because a bad scientist told you, or
    rather because a civil servant copied the Americans who never did
    any science on the subject. FWIW, Zoe promotes liver as the perfect
    food to obtain vitamins, and she is right. Liver has something like
    ten times the concentration of vitamins of an apple, and also
    enables the fat-soluable ones to be absorbed. While some vegetables
    may have vitamins and minerals, a lot of them can’t be absorbed
    (spinach is a case in point). Back to Zoe. There are two types of
    diet advisers, nutritionists and dietitians. Anyone can be a
    ‘nutritionist’, and that is what Zoe is. Dietitians are a regulated
    body, but unfortunatly they are only allowed to follow the party
    line on diet, and sadly the party line is what caused our obesity
    explosion. I am deeply suspicious of the party line as there is
    appalling science behind it (read Gary Taubes book), so in this
    case, I would prefer to listen to someone like Zoe who is allowed
    to think freely without being struck off from her
    association.

  37. Guy said,

    February 4, 2011 at 9:12 am

    Matt, that’s the whole point. It would be very easy to design a simple 9 kids with one adult test for it and see how often it picks out the adult. As outlined about, sensitivity and specificity could be published with any “black box” method. We have no need to know how it works but whether industrial app or not, we have a right to know whether it works. For that we just need to see the evidence.

    Otherwise it’s like the awful non functional bomb detectors our government has been encouraging the Iraqi police to buy, which then kill people because they don’t work!

  38. Vincent said,

    February 5, 2011 at 12:23 am

    “A child has been born … just like their two siblings” Bad grammar bothers me as much as bad science. “Its two siblings” would have been correct.

  39. thoralf said,

    February 6, 2011 at 11:43 am

    @Vincent:

    Relax, singular “they” is fine, even for children:

    A: “It’s one of your children on the phone for you. Not sure which one — maybe Mike, maybe Betsy.”
    B: “Tell them I’ll call them back in a minute, I’m in the middle of something right now.”

    No English speaker I know would say: “Tell it I’ll call it back later..” or, “Tell him or her that I’ll call him or her back later…”

  40. Robert Carnegie said,

    February 6, 2011 at 2:47 pm

    28: oh, “raw fruit” do you mean? I think the brand may be “Eat God’s Way” but it isn’t clear to me that that refers to one particular plan. I found “Hallelujah Diet”, and a bunch of “testimonies” at www.hacres.com/library/testimonies/heart-health/p5-15 You will also read of “The Maker’s Diet”. Otherwise the topic seems to be dominated by illiteracy – that is, bad spelling. www.hem-of-his-garment-bible-study.org/bible-foods.html wants to know “Have you ever heard of the Levetical dietary laws?” and www.eatgodsway.webs.com/ sets about “letting others know that there body is not their own, but it is a temple of God.” “There are recipes for vegans, vegetarians, and even those who eat normally.” “Online Recipies”, that is. (I thought you said it was no meat?) (Fish doesn’t count as meat, perhaps, although it does if you’re vegetarian.)

    As far as I remember from the story of Cain and Abel, God himself likes cooked meat – smoked, probably even more unhealthy – and doesn’t go veggie. The cartoonishly illustrated “Good News Bible” has a memorable sketch of Abel’s barbecue smoke rising nicely and that of Cain’s vegetable feast going sideways. Poultry is also acceptable but all meat must meet EU-type quality standards. As to who really gets to eat the cooked offerings, it’s the priests, apparently. Isn’t that interesting.

    Actually, less red meat and more fruit and veg., raw where appropriate, is a fairly good eating plan, as far as I recall.

    36: by the telling of it, Zoe Harcombe is the latest Gillian McKeith, all the way down to the “controversial” academic qualification. Indeed we should not be unkind, we should be excoriating. We should write to the newspaper to declare we are not going to use recycled toilet paper any more in case we accidentally wipe our bum with a Zoe Harcombe article. Well, perhaps not, that would be rather silly.

  41. Robert Carnegie said,

    February 6, 2011 at 2:58 pm

    And perhaps I really shouldn’t bang on about this, but there’s also “The Divine Diet” according to www.articlesbase.com/wellness-articles/the-divine-diet-for-healthy-living-2870770.html In this one, “Everything that lives and moves will be food for you. Just as I gave you the green plants, I now give you everything. (Genesis 9:3)” When you read the small print, you do not have to eat -everything- after all, which is lucky. “Author is a licensed realtor with a passion for natural remedies. Author received certificates in Eat God’s Way and Nutrition Consultant from Life Extension Center. Find our more at Miracle Juicing.” Interestingly, -that- link doesn’t mention the bible at all, but the MJ book itself is full of it. If you do have a bible already, then read that and save the money. :-)

  42. Vincent said,

    February 6, 2011 at 6:34 pm

    thoralf, thanks. I agree with “them” in your example. I know they “is fine” these days, but one can still be bothered by a usage in a given context. Or in other words, it’s not a question of rules.

    Apologies to Dr Goldacre for this irrelevant side-issue. I won’t do it again.

  43. MattCh said,

    February 7, 2011 at 4:20 pm

    @ Guy

    The problem is the evaluation you suggest wouldn’t have any credibility as it couldn’t be verified. They could test their app with 100 children and 100 adults, write a paper about it, and claim it got 70% right, or perhaps change their mind and claim 80% or 90%, etc. We would be none the wiser, and to be honest I don’t think such a paper would get published.

    The thing is this isn’t like doing a drug trial or something, where you can evaluate the end result on the user – in this case it requires evaluating the technology/algorithm itself and I don’t think you can credibly do this without actually knowing how it works. I can’t say I’ve ever seen a computer science paper where ‘how it works’ is not also described or referenced to – unless it is a well established system.

    So in some ways, the fact that the press release is quite vague on how well the app works (is it talking about accuracy, efficiency, etc compared to a server – who knows???), is probably better than providing figures that can’t probably be verified as details of ‘how it works’ can not be made publically available

  44. WilliamJay said,

    February 15, 2011 at 2:35 pm

    @28, jenniferbridge

    “Slightly random question – but sort of related to the diet above. I’ve got a a friend who is a former anorexic / bulimic and she is evagelising about the Allelujah diet (front cover “Eat God’s Way”)”

    I thought the book where god told people what to eat was called Leviticus, perhaps they’re re-branding?

    “My plea is “CAN ANYONE SEND ME RANDOMISED CONTROL TRIALS OF THE ALLELUJAH DIET?” Please?”

    I would have thought the answer to that was blindingly obvious.

  45. WilliamJay said,

    February 15, 2011 at 2:40 pm

    @36 George999

    “Zoe is correct in challenging the promotion of ‘five a day’. There is no scientific background for it, it was merely adopted by the Food Standards Agency simply because the US Department Of Agriculture (sponsored by cereal growers no less!) adopted it.”

    Am, thanks for the conspiracy theory but cereal growers don’t grow fruit, they grow grains. Strangely grains aren’t one of the five-a-day. Either the cereal growers are involved in an extraordianary circuitous conspiracy, by getting everyone to eat fruit that they don’t grow or there may be a flaw in your analysis.

  46. WilliamJay said,

    February 15, 2011 at 3:12 pm

    @ 43. MattCh

    “The thing is this isn’t like doing a drug trial or something, where you can evaluate the end result on the user – in this case it requires evaluating the technology/algorithm itself and I don’t think you can credibly do this without actually knowing how it works. I can’t say I’ve ever seen a computer science paper where ‘how it works’ is not also described or referenced to – unless it is a well established system.”

    I don’t see what the problem is. Infact there is no problem. Take a 100 kids and 10 adults, get them to hold write chat conversations where the adults pose a children. Feed the conversations into the testing device, it guesses if they’re children or adults. It either gets them right or not.

    “The thing is this isn’t like doing a drug trial or something, where you can evaluate the end result on the user … The problem is the evaluation you suggest wouldn’t have any credibility as it couldn’t be verified.”

    Am well, you take the guess the software made about wether person X was an adult or a child, then you look at person X and they will either be an adult human or they will be pre-adolescent human – that seems pretty easy to verify to me. Unless the testers include few dwarfs in as well, just to throw us a curve ball.

    “They could test their app with 100 children and 100 adults, write a paper about it, and claim it got 70% right, or perhaps change their mind and claim 80% or 90%, etc. We would be none the wiser, and to be honest I don’t think such a paper would get published.”

    In science we use a thing called a double blind trial.

    The idea is that the person doing the experiment doesn’t need to know the data they’re experimenting on – in this case the age of the authour of the text sample – infact they ought not to know the age of the authour of the text sample before they predict the age of the author; it wouldn’t be a double blind trial if they didn’t, nor would it be a very convincing prediction. Their predictions are then compared with the actual results, that some third party has being hiding from them or that hasn’t happend yet – that’s the ‘blind’ part.

    They’re supposed to make their predictions based on their software, then a neutral third party will reveal the actual ages and we can see how well their software’s predictions matched the actual ages.

    And at no point in this experiment do we need to know how the software works, as Guy says it can be treated as a black box.

  47. Kevin P said,

    February 27, 2011 at 6:45 am

    The paedophile identification app is a free download on the iPhone app store (search for ChildDefence, with British spelling and no space), so anyone who wants to do their own trial and report the results should be able to (unfortunately I don’t have a compatible phone so I can’t help). There’s also an email address on the spin-off company’s website to request more information, child [dash] defence [at] isis [dash] forensics [dot] com, which might be more helpful than the university’s press office.

  48. Les Reed said,

    March 11, 2011 at 4:12 pm

    Having checked some of the recent abstracts of papers on the role of fruit and vegetables in preventing cancer it seems that Zoe Harcombe is not completely wrong in her article. She is certainly extremely foolish over the PHD claim and is certainly over stating her argument but the studies I looked at online from the past year do appear to question the “5 a day” orthodoxy. While we may not trust Harcombe to provide shortcuts in analysis of primary research it seems we cannot trust mainstream nutritional advice either.

  49. edsite said,

    March 22, 2011 at 8:49 am

    Further to what Tim O’Leary said, a far more disturbing application of such a programme – providing it works – which again would require no knowledge of its inner workings, just knowledge of the police´s faith in its ability, is the possibility of using it to hone one´s ability to pose as a child.www.topedhardystore.com