September 19th, 2009 by Ben Goldacre in bad science | 12 Comments »

I’m very happy to post a quick correction.

In this column:

1. I gave the newspaper names incorrectly. It was the Sunday Times, not the Times, and the Sunday Express, not the Express.

2. When The Guardian edited my column, they removed the response from the Sunday Times. However this response was included on the version, where I always post my unedited copy.

The Guardian correction is here:

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! ++++++++++++++++++++++++++++++++++++++++++

12 Responses

  1. Ben Goldacre said,

    September 19, 2009 at 5:07 pm

    Oh, I should expand on the Sunday Times’ original response, which I reported unchallenged in the original piece in order to save word length. Reporter Jonathan Leake told me that Prof Ryskin had approved his copy. In fact this was untrue: I have the emails between Leake and Ryskin. Ryskin was sent a version of the article which said that changes in the earth’s magnetic field were related to the movement of the oceans – which Ryskin did not object to – but of course the paper printed the incorrect claim, that the earth’s magnetic field was caused by the movement of the oceans. So Ryskin did not approve the Sunday Times’ error.

    When Professor Ryskin sent me the draft he was sent, he said:

    “…notice that the first sentence of the draft had “linked to” where the final article has “produced by”. While I kept insisting that the origin of the field should not be speculated upon (see below), I was willing to acquiesce to the words “linked to”, which may be viewed, arguably, as reflecting my published paper. The words “produced by” is a totally different story; this was a big surprise to me when I read The Times. Science requires absolute precision of wording; this may not be obvious to everybody.”

    I note with some pleasure that while the Sunday Times quietly changed the online version of their article about Ryskin’s research after I contacted them to say I was going to write on it, they have never published a correction for their misrepresentation of his work.

  2. dutch said,

    September 19, 2009 at 7:03 pm

    Isn’t the crux of the complaint by Leake that you managed to associate him, and his story, with the somewhat scummy Sunday Express story on the Dunblane kids? No apology for that?

  3. jdc said,

    September 20, 2009 at 12:04 am

    One of the (many) reasons I still buy The Guardian is that they have the integrity to promptly and prominently correct errors – it’s a shame all newspapers don’t do this*. One of the (many) reasons I read blogs like Improbable Science and Bad Science is that the authors are happy to publish corrections and clarifications – it is a shame all writers don’t do this for their blogs / books / websites.

    *Some newspapers will change inaccurate articles or headlines without noting that they have done so, some are reluctant to make any corrections. When I have complained to editors about inaccurate headlines / articles I have been ignored. When I have followed up with a PCC complaint, changes have (sometimes) been made but not acknowledged. I think acknowledging corrections and clarifications is almost as important (or perhaps as important) as making corrections.

  4. Methuselah said,

    September 20, 2009 at 2:07 am

    The Gruniad don’t always fess up their mistakes, as their latest attempts to assign the E. coli outbreak story to someone who knows the difference between a virus and a bacteria demonstrates. I gave up on following the goings ons but they didn’t respond to an email correcting their initial confusion:

    Apologies for the self-spruik.

  5. wertheim said,

    September 20, 2009 at 5:32 pm

    Hello Dr Goldacre,
    I don’t know how to contact you other than through this site. Just want you to know that I read your book Bad Science with pleasure (although the media situation in the Netherlands is bad too, it seems to be somewhat less extreme as it appears to be in the UK). But I thought there might be an error in the book on page 214. I like your suggestion of looking at the symmetry of a distribution as described here. It should be indeed symmetrical. However, the distribution in the graph you mention on this page appears to me to not symmetrical but very much skewed to the left, because the horizontal axis is logarithmic. Am I right?
    Another comment I would like to make refers to your discussions about placebo effects. I fully agree with your arguments here, but I wonder why you do not mention the experimental design known as “colored placebo design” in cases where you do research on healthy subjects. It allows for the precise measurement and distinction of a placebo effect, and a real treatment effect, without the ethical problems involved with the usual double blind designs. I used it to investigate effects of amphetamine as compared to that of barbiturates on reaction times. I’ll use this as an example: First you split the to be investigated sample population in two groups. Group A receives amphetamine, barbiturate and a placebo on three separate occasions. You tell them what they get and on which of the three occasions. So you are perfectly honest and inform them about the effects to be expected. Also that the placebo pill might not have an effect. Group B gets the exact same treatments, i.e. you tell them also the expected effects of the three drugs and when they get them. However group B in reality only gets placebo pills. So you lie to this group but that’s no ethical problem if they are healthy subjects, because they do not get the drugs. When you now subtract the effects observed in group A and group B from each other, the difference corresponds to the real drug effect, because the expectations are the same in both groups. And the findings of group B give you the precise placebo effects. Moreover, when you compare the placebo trials of group A with group B you should find no difference. If you do, it means your groups were not properly matched. So this is a control of your matching procedure. It seems to me that, when you do drug testing with healthy subjects this design is better than a normal double blind trial. If I would have used the traditional design, i.e. one treatment with the amphetamine, one with the barbiturate and one with the placebo, in a double blind design, you may get into a methodological problem. In my case subject who are randomly assigned to the treatment conditions will on their first visit probably find the reaction time tasks dull and thus believe the had a placebo or barbiturate. When they come for the second time that is likely to happen again. So when they come for their last visit their expectations will soar high that this time they might be given the amphetamine treatment. One third of them does get it indeed. So the results of the amphetamine condition will be strongly biased in one direction and those of the barbiturate condition will be biased in the opposite direction. That would confuse the interpretatio of the data. The colored placebo design does not suffer from this.
    Alex H. Wertheim

  6. billpg said,

    September 20, 2009 at 11:42 pm

    “I gave the newspaper names incorrectly. It was the Sunday Times, not the Times, and the Sunday Express, not the Express.”

    So you got the name of the papers right, but failed to mention that it was in the editions published on a Sunday, and *that* warrants a correction?

    If you want to quote me, remember that I’m the *Sunday* billpg, and referring to me as just billpg is simply wrong.

  7. Hobnob said,

    September 21, 2009 at 8:35 am

    Actually, it’s a genuine error: although the Sunday Times and The Times have the same owner, they’re separate papers with different staff and editorial. The relationship is the same with eg. The Guardian and The Observer, which often run simultaneous promotions but are not just different editions of the paper. ‘Sister publications’ is, I believe, the term.
    My guess is that there may be some connection with the sunday trading laws?

  8. Ernesto Collianus said,

    September 21, 2009 at 3:27 pm

    Anyone care to comment on Milk Thistle?

  9. Ernesto Collianus said,

    September 21, 2009 at 3:27 pm

    or is it Thistle Milk?

  10. billpg said,

    September 21, 2009 at 8:45 pm


    I recall a cafe around where I grew up, the owner ran it during the day and in the evening they rented it out to someone else to run. The evening management had separate business arrangements. Both traded under the same name.

    Say a journalist had mentioned the cafe, but didn’t clarify if they were talking about by-day or by-night. Is that correction-worthy?

    (I’m not saying the correction is bad thing, just seems unnecessary.)

  11. Daibhid C said,

    September 21, 2009 at 10:50 pm

    If the journalist was criticising the cafe, and the daytime-manager was the one responsible, I think the nighttime-manager might have reason to feel aggrieved, yes.

  12. csrster said,

    September 22, 2009 at 7:48 am

    And actually it was the Scottish Sunday Express, according to the Guardian.