Please turn on javascript in your browser to play chess.
Science Forum

Science Forum

  1. Standard member Grampy Bobby
    Boston Lad
    05 Oct '13 19:49
    Statistics: A Poem (by Rachel Croxton)

    http://www.youtube.com/watch?v=95YYnLWJOdM

    Placed here rather than in the Culture Forum for a specific reason. I'd like your help with verification of this delightful poem's content accuracy for a friend earning her Masters Degree in Management with a demanding Course in Statistics. Thanks.
  2. 05 Oct '13 20:20 / 2 edits
    Grampy Bobby:

    I assume you simply did not think of this, an accident. I simply can't beleve that you not by chance but by will, intetionally, posted a youtube link. Huh? No purely accidental for sure.

    DO NOT BE LAZY AND just assume we want to spend our precious time with hearing (a bad youtube video, I DON'T ANYMORE watch them myfelf. BY PRINCIPLE. Watched too many bad ones).

    THIS IS WHAT I WOULD LIKE you to please please, do: Write it down! Then I (perhaps) am going to consider your request, your demand. I don't think anyone else is going to do this for you so do it yourself.

    Please?
  3. Standard member Grampy Bobby
    Boston Lad
    05 Oct '13 21:44 / 1 edit
    Originally posted by bikingviking
    [b]Grampy Bobby:

    I assume you simply did not think of this, an accident. I simply can't beleve that you not by chance but by will, intetionally, posted a youtube link. Huh? No purely accidental for sure.

    DO NOT BE LAZY AND just assume we want to spend our precious time with hearing (a bad youtube video, I DON'T ANYMORE watch them myfelf. BY PRI ...[text shortened]... and. I don't think anyone else is going to do this for you so do it yourself.

    Please?
    [/b]
    Statistics: A Poem

    This poem is for all you guys
    Who have some data that you need to analyze.
    Listen closely, cause my rhymin' will show
    All the IB statistics that you need to know

    First step: gather up those data observations
    Take the average, that's the mean, then standard deviation.
    The mean is a measure of location,
    The center of a population.

    Unless you want to be a comedian,
    Don't confuse it with the median.
    Of course, the median's not much of a riddle
    It's simply the number in the middle.

    So just stick to the mean and you won't get confused,
    It's the best measure of central tendency the statisticians use.
    You could compute the mean in your slumber.
    Just sum the scores and divide by the number.

    Normally, the measures near the mean are many.
    And in the tails, there's hardly any.
    Now, in your measurements the scatter
    Is something that really matters.

    Cause it tells how close to the mean you're likely to be,
    If a single measurement is all you can see.
    Or if it will take a lot of measurements to be able
    To get a mean value that's stable.

    To calculate a data set's variation,
    First find the mean and then each score's deviation.
    Each deviation then you square,
    And sum them up, all values share.

    Divide the sum by N minus one.
    Take the square root, and then you're done.
    The standard deviation, called SD,
    Is a good gauge of a trait's variability.

    So when showing your data in a graphical plot,
    The mean is usually drawn as a single black dot,
    And with the dots there usually are,
    Standard deviations shown as error bars.

    Forget this number and face some IB hate,
    Plus or minus one SD includes a percent of 68.
    So we've discussed two moments of a distribution,
    Calculated with each measurement's contribution.

    But there's another part of statistics that is totally the best,
    It helps you compare two sets of data — the t-test.
    The t-test tells whether a difference you see
    Reflects random scatter or reality.

    Compute the t statistic and you can tell at a glance
    Whether the result you found was likely due to chance.
    T-testing is tough, but I'll help you survive:
    First choose a confidence interval, usually point zero five.

    But before you'll be able
    To use the statistical table
    You'll need to also know
    The degrees of freedom. Oh No!

    To get degrees of freedom, here's what you do,
    It's the number of samples minus two.
    When the measurements are paired, it's even more fun
    It's simply the number of pairs minus one.

    You can even compute t by hand if you dare
    Just calculate the difference for each pair
    The mean over SD is where you begin,
    Then divide the result by the square root of N.

    Now compare to the t-table, it's no big deal
    To tell if your difference is by chance or real.
    Or if the value you place on your time is greater,
    Skip these instructions and use a calculator.

    Your lab partners will think you're incredibly intelligent
    If you're able to tell if your data is significant.
    Prove your result is significantly different from zero
    And you're sure to be a hero.

    But watch out for one thing, a danger of sorts,
    From a value that computers often report
    They often provide when a t-test is done
    The coefficient of correlation.

    This is great to know since it tells you whether
    Two things you measure tend to line up together.
    But there are terrible IB threats
    For any student who ever forgets

    That even strong correlation
    Can never imply causation.

    (Published on May 29, 2012. An outline of IB statistics including mean,
    standard deviation, t-test, and causation vs. correlation.)

    Note: Hi, bikingviking. I've got Viking Ancestral roots too. Thanks for your request. Clarifications: Laziness isn't now nor has ever been one of my shortcomings. A strong Swedish/New England Work Ethic has characterized my daily regimens since a long 12-month paper route on bicycle (snow shoes in winter) as a nine year old boy. Length of the poem was the reason for suggesting a few minutes listen to an excellent quality and delightful reading by the Poet, Rachel Croxton. Judgmental on your part to dismiss it as "a bad youtube video" on the basis of your previous bad experiences. Kind regards, Bob
  4. 05 Oct '13 22:18
    I did not assume it was bad or so. No, but was not in the mood for taking chances and listen to a Youtube video. I think you understand.

    Must have taken you some time to write, good there. Don't know if I can be of much help though (even though I got highest score in my old statistics course). To me everything sound accurate. But as said, I am unable to really tell.

    I remember something about that in game theory you could use the some ... have to quote
    "Divide the sum by N minus one.
    Take the square root, and then you're done"
    N minus one. Then there is another wasy to do this as well , perhaps N minus 2? Only remenber that there was two ways of doing it and that both was correct. Then something that is not in the poem is how to plan the statistical survey. You could i.e. use a paper with questions. Then send it to every female born in 1948, living in a certain area. Then you have to analyze the results and that's kind of tricky. Another thing you could do is measure a physical parameter of some kind. Perhaps several and see if they correlate. Sometimes you get a curve with a bump in the middle (in population surveys). That usually implies that you think you have in fact measured on two populations, not one. Then there is some "tricks" you can apply which is not strictly allowed but that can help you estimate things if you have too few observations. Of course these tricks can be wrong but mostly they are helpful. Then you can user some "grids" tequniques for analyzing, Think a cube. Would take some time to explain (the grids and cubes, you plot your results on the cube) and frankly I am unsure if I am able to do so at all. Then there is things which is not really suited for measure with statistics (no matter what statistics you use it will not tell you anything). This is also hard to explain and is most likely the result of an outside factor effecting the "thing you measure on". That means that you could use statistics on that and get results. Then the results is wrong because it is a speacial case. Then there is the fact that you can use other peoples numbers (if i.e. available publicly) and you can't be sure if those numbers are correct so you assume they are, but you do not know that. Then there is a problem that people are trying to manipulate numbers for various reasons, mostly because the answer, we don't know is not good enough.

    It was a nice poem by the way, it does not explain statistics but it was nice. Meaning I would not use it as "facts" but as "a quriosity", to break the ice. Getting people interested in statistics perhaps. Statistics is hard, poems are easy. Most of the time.