One of our early forays into the world of statistics is the concept of probability and chance. In Maths lessons they are demonstrated bytossing a coin or throwing a dice. And we know intuitively that there is an equal probability of it being heads or tails, or any one of the dice faces. We learn this by observation long before we even know we are learning.
We learn about normal distributions in Infant school when we learn that some of the children are relatively very tall and some are very small but that most are pretty much the same height , in between. We learn that design is rarely random - that certain first names are more common than others. Throughout my life the Catherines and Helens and Stephens and Davids have outnumbered the Geraldines.
Later as we study statistics in more depth, we use random number tables, and we learn about different ways of sampling. As an auditor I have selected judgemental samples (that one looks interesting or important, that one is big), random sampling...random number tables are useful in this. In Monetary Unit Sampling every pound has an equal chance of being selected, thus biasing the sample to the larger items.
Except that in 2006 everything is computerised. My life is filled with computerised randomness. I mainly have my mp3 player on random. I have a screensaver that sends photos from my hard drive randomly across my screen, so one moment it could be a nepling looking cute, the next a hillside in Scotland, followed by flowers from the garden at Gert Cottage. The main page of this blog will kick you up a random photo taken from a folder on the server called 'random'.
But it troubles me. I do not believe by observation that the selection is purely random. On my mp3 player I sense that certain albums get featured over a short period. This week it seems to be Der Rosenkavalier and The Best of Nanci Griffith. On 4th May en route to the Crossover Classical Brits all but two tracks were from Plácido Domingo or Rolando Villazón, both of whom I was looking forward to seeing that evening. Of the remaining two, one was from Gerald Finley whom I had seen twice the previous weekend, and the other was Ideology by Billy Bragg - and it was election day! Now, I realise my mp3 player could not have known these facts.
But in any case it was not a random selection. A long time ago I used to play occasionally the Lottery and chose my numbers by using the random number generator on my calculator. (It's the sort of thing auditors do...). And yet they never appeared to be random. I know that the lottery is a perfect illustration of Permutations, with each permutation standing an equal chance. Each number standing an equal chance. But my calculator always seemed to bunch them. So I would end up with three single digited numbers and three in the Forties. That could not be random, it didn't look random. At least they never kicked me up: 1,2,3,4,5,6 because everybody knows that that sequence will never win the Lottery. It's like Double Six is special on the throw of the dice.
My photos are not random - try it and see, refresh the front page, see how quickly the same picture recurs, although the folder contains 127 images. If I had patience, I would make notes of my observations and see how they compare to what we would predict using pure probaility theory.
And then I would know whether I am blinded by my identification of random digits as things (photographs, music tracks) or whether computers select random numbers by first breaking the population into sub-populations and then concnentrates on just one segment. If I could prove the latter, it could have major implications for audit and other disciplines that use computerised sampling techniques.
It's depressed me now.
