Counting Cats in Zanzibar Rotating Header Image

Mercury rising in Uranus…

I haven’t done one of these for ages…

When I was a kid I had to do a an English comprehension thing on an article about astrology. I recall it well. Basically it was saying astrology was right because a recent British survey had found a significant statistical anomaly. There was a large peak of rrade union leaders who were Aries (an aggressive, confrontational sign) and a large number of nurses were Pisces (a caring, compassionate sign). These were both well above the expected statistical norm.

Explain what that means and if you think it meaningless why is it?

I knew why.

PS. As far as I recall no mention was made of the Sun sign of then current head of the RCN.

19 Comments

  1. Michael says:

    Astrology proves one thing beyond all doubt… There is one born every minute!

  2. JuliaM says:

    I wonder if the nurses in the case of the man dying of dehydration in a Tooting hospital were Pisces?

  3. RAB says:

    Could it be that most people like to do their shagging in the warmth of the summertime, so disproportionately more people are born, 9 months later, under the signs of Pisces and Aries?

  4. timdifano says:

    I’d say that this maybe a selective use of statistics, how many other job titles did they survey before finding the answer they wanted?

  5. bloke in spain says:

    Don’t knock astrology. Chica in the bar was claiming Pisces last night. Doing fishy swimming motions with her hand to get across the language barrier. Turns out, being Virgo’s not a disadvantage. Although the sign language was complex. There seemed to be some interest in testing the validity…….

  6. NickM says:

    RAB,
    I thought someone might suggest something like that. But I don’t think that is. I admit I don’t know seasonal variations in births but that was not what I was thinking of. There is though something implicit in your argument that is essential to this puzzle.

  7. Single Acts of Tyranny says:

    Don’t know about nurses (faulty data perhaps) but surely “trade union leaders” don’t make up a statistically significant sample. I admit many are probably chippy, but there may also be a selection bias, (i.e. pick those who confirm the theory, ignore the data that does not). What after all, is a trade union leader?

  8. RAB says:

    You enigmatic swine Moriarty!

    Lemmie see now… Shagging… Warmth…Summertime..

    Oh I give up, we Taurus types never get your puzzles, too grounded see. And if this were really an English Comprehension test I might do alright at it, I was very good at them, picking out logical fallacies and all. But you haven’t given us the full text have you, just the riddle? Swine again! :-)

  9. dcardno says:

    Or other confirmation bias, SAoT: pick several dozen professions and trades (each with potentially the same issues you point out wrt ‘trade union leader’), then examine whether any of them show anomolously high or low members of an orthagonal class (astrological sign, handedness, public school, hair colour, birth order, what have you) and publicise the shit out of it. About the other groups that don’t show an anomoly (or that counter your pre-conceived corelation) – stay mum.

  10. Tony says:

    Seems to me there are three issues here:

    1. Data mining: there are thousands of ways of groupings people. It is inevitable that statistically significant correlations will be found if someone goes looking for them. Also, 2 sigma (95%) confidence is ridiculously weak. 5 sigma (1 in a million) as used for the higs boson would be more impressive.

    2. People tend to read up on the supposed characteristic of their sun sign. So pisceans will believe they are more suited to caring professions such as nursing.

    3. Not so relevant here perhaps but the month in which you are born determines the age at which you start school. And this could influence your future career.

  11. Lynne says:

    Well I’m Pisces and I can tell you that the caring and compassionate bit is pure bollocks. Sod nursing, I wanna be a demolition expert.

  12. zack says:

    I think Tony is pretty close. My take is thus:

    Correlation does not mean causation. Most nurses are women, does this mean most women are nurses? No.

    Lets look at a more controversial subject. People who have less education/lower IQ’s tend to have lower incomes then people with more education/higher IQ’s. Blacks (in America) are on average poorer then whites – does it then follow that blacks are naturally dumber then whites? Not at all – Thomas Sowell has a great article debunking this fallacy ( http://www.jewishworldreview.com/cols/sowell100402.asp )

    The article was taking a (weak) statistical correlation and making unfounded assumptions based on these correlations.

  13. NickM says:

    SAoT and Tony got it. OK, this is my take…

    SAoT’s point is true especially when taken with the fact that nursing is probably one of the commonest careers in the country. Very diffiicult to compare two sets of such different sizes. It’s kinda like comparing the favourite take-away food of cat owners with python owners.

    As to date of birth and school-start times. Yes, there might be something in that. I had thought of that and I did wonder specifically if there might be something in that too. My thought on that was that nursing is structured career and nurses tend to start training straight after school and nursing schools have fixed intake times around the year. That might be relevant here.

    But the real point is the data mining one. Which is that you can splice and dice the population any which way until you get a conclusion you want. That is a very serious problem with how social sciences operate. Actually quite a few sciences can be operated this way. The question naturally arises with other countries and other caring/aggressive jobs. This would not be such a one off statistical oddity if the same results could be found if we looked at the Sun signs of say Bolivian wrestlers, French kindergarten staff, Australian fighter pilots and Danish priests.

    In any case the whole thing is proposed on the basis of a theoretical assumption which is very crude and based upon cultural assumptions that are perhaps not universal though astrology claims to be. It might just be myth, legend, TV, movies, books…

    In science the theory has to have some real theoretical validity before you go out looking to prove it. I’ll give you an example My brother did archaeology at university and on a dig he found a Twix wrapper in his trench once. Now either Roman Britain had plastic-wrapped chocolate snacks or a passing scrote had chucked it in and it had got bogged down in the mud with the rain the night before…

    I must write more on this sort of thing for it interests me greatly.

  14. NickM says:

    PS. I wrote that before zack commented.

  15. zack says:

    PS: last sentence should read “the article that Nick is talking about was taking a (weak) statistical correlation and making unfounded assumptions based on these correlations.

    PPS:. also, looking this up on wikipedia (to follow up on RAB’s thought, looking for holidays that may explain a jump in births), I found another problem – there are multiple Zodiac calenders, each one slightly different. The two major ones actually differ their signs by about a month – so ones Aries is the others Pisces. So the article nick is referencing, claiming that *this one* statistical correlation proves astrology, is equivilent to saying that ‘philosophy has been proven correct’; ok, which philosophy? Materialism? Christian Humanism? Confucianism? Platonism? Aristotelianism? Stoicism? What philosophy?

    Also, (to be more scientific) to make the claim that something is Proven, you need to show that it’s predictions/claims are correct in ALL particulars, not just one. The theory of luminiferous aether may have been compatible with the observations of the 1800′s, and provide an elegant explanation of some phenomenon, but it didn’t explain/fit with newer observations, and had to be discarded. The same is true with astrology.

  16. zack says:

    PPPS: damn it Nick, we keep missing each other! Well, I’ve said what I’ve ment to say. Looking forward to your feedback on my comments.

  17. NickM says:

    PPPPS,
    timdifano and dcardno also make fine points. Sorry folks you were held in the spam filter.

    Don’t worry zack! I didn’t know that BTW. A further point of course is that astrology involves much, much more than Sun signs. I have drawn a horoscope! I had a bit of spare time before my MSc in Astrophysics and I thought I’d look into it because I was somewhat interested in the history of the subject. I was trying to get a handle on Kepler et al.

  18. Philip Scott Thomas says:

    I was trying to get a handle on Kepler

    One of my favourite scientists. Not only was he interesting as a person, but I also cannot read his name without being reminded of that ancient joke about liking Kipling.

  19. Nick says:

    The weird thing about statistical significance is that its validity changes according to how many tests you run to try to show your proposition: http://xkcd.com/882/

Leave a Reply

%d bloggers like this: