Sunday, January 28, 2007

The Knock Out Mouse 2: The Poached Eggs Solution

In a comment on my last post, the indefatigble Wisconsin librarian Dave Lull draws my attention to The Journal of Spurious Correlations: Qualitative and Quantitative Results in the Social Sciences. This seems to answer the same need expressed in the Journal of Negative Results in Biomedicine. It's all too easy to publish positive scientific results, almost impossible to publish negative ones. But negative results are at last as important as positive ones. If they are not published, then science will slow to a crawl as thousands of scientists replicate experiments that they have not been told will fail. This is a very profound issue. It demonstrates how human failings can so easily compromise the super-human aspirations of science. Or, to put it another way, it is an example of how these super-human aspirations are, in fact, illusions. Whatever we do, it will always be infected by what Christians so resonantly call 'original sin'. That said, I have always been inordinately fond of spurious correlations. Indeed, I am not entirely sure they are all spurious. Take, for example, the poached egg solution to airline safety. On any flight, you should always carry a poached egg in each pocket of your jacket. This is because no air crash victim has ever been found with a poached egg in each pocket. Call me a sentimental old fool, but I find this very suggestive.

9 comments:

  1. Reminds me of 'failure analysis' - key strategy in streamlining business practice.

    ReplyDelete
  2. I have this, some would say spurious, argument in favour of fate or destiny. If fate does not exist, then everything that has happened was not bound or destined to happen. However this everything has indeed happened. Which of course proves that fate or destiny does exist.

    ReplyDelete
  3. But, Andrew, only if fate/destiny are identical with what happens, which is what needs to be demonstrated, otherwise you are commiting a petitio.

    ReplyDelete
  4. There is something a bit post hoc about that argument, Andrew. Is it even that one '...ergo prompter hoc'? I don't know, logic was never one of my strong subjects. Anyway, as one of your old favourites might say: amor fati. It's not a bad attitude, regardless of whether there is such a thing as fate. Let's pretend. Eh?

    ReplyDelete
  5. I'm a professional science editor and I guess you know too, Bryan, that the system of publication works via positive results not negative ones. It is a very big problem, the issue of "reinventing the wheel" because some guys didn't know that another lab had got a negative result but coudln't/didn't publish it. So all power to these negative results journals - there are quite a few of them now, and there has been much comment in the scientific journalism sphere about them. Dave Lull's example is a bit different as it concerns social science, which is a far more subjective discipline. (ie they argue all day about what a negative result means, or ask whether it is negative or postitive, etc. Those social scientists must be paid to argue they love it so much).
    What is vital, and what you highlighted in your earlier post, is the big biology projects of postgenomics. These projects take up so much in the way of resources, effort, person power, computer power, etc, that failing to capture negative or other "non useful" data is almost criminal.

    As well as journals to record negative results (which, sadly, will never get high impact factors so won't stimulate any but the most altruistic to publish there -- and big pharma R&D is not altruistic), there are internet-based initiatives.
    Alliance for Cell Signaling, published by my employer Nature Publishing Group, is a case in point. That is a collaboration between most scientists working on cell signalling. everyone pools their resources to characterise all the molecules in a signaling pathway. It is a great way to capture information that can be built on in the future when more sense can be made of it, and to enable comparisons between datasets, etc.
    (I forget the exact url of alliance for cell signaling, but you can get to it via www.nature.com)

    ReplyDelete
  6. What I mean by fate here is a kind of Absolute Fate or Total Pre-Determination. It is clear the argument is false but it seems strangely tricky to pin-point why.

    ReplyDelete
  7. I failed to invent a time-machine yesterday.

    ReplyDelete
  8. [...]I ship worldwide. 10 Year seller with 100% Feedback. Great Valentine’s Day Gift![...]

    ReplyDelete