PERSAUD Wilberne Persaud, Financial Gleaner Columnist
READING THE Gleaner lately unearths a fascination with knowing the future.
In this case, elections. Important as they are, I don't plan to say much except that pollsters have for years taken Jamaicans for a fantasy roller coaster ride - providing views on topical issues with a veneer of truth and scientific rigour that rarely existed.
Today, clairvoyants and others emerge. Look at that car, when I rub my nose, it will move.
I want to recommend two authors with hugely perceptive views on probability and prediction relating it to business and finance, but I first share part of a column from September 1997 discussing a plus or minus three per cent sampling error.
Ten years on, the comment is still valid.
"It is being claimed that the poll has a margin of error of plus or minus three per cent. This is fantasy. The polls are done on the basis of what is called a quota sample. I have heard mention of 47 and 40 areas being chosen for interviewing 1,200 people.
If we wish to speak of a margin of error and do so correctly, there must be a random sample.
The errors created in using statistics generated by sampling techniques are of two kinds, sampling and non-sampling errors."
To quote from Pfaffenberger & Patterson Statistical Methods (p.311): "It is important to understand that sampling error is a controllable, but unavoidable part of sampling. The amount of sampling error can be diminished by increasing the sample size.
To choose or select a random sample one would first need a listing of all the eligible voters in Jamaica. This would be in reality, the voters' list of perhaps 1.5 million people over the age of 18 years old. One could also use a stratified random sample.
One method of picking the sample would entail giving each name a unique number from 1 to 1.5 million.
By use of our computers today, we would then be able to pick 1200 names, each of which would have an equal chance of being represented in the sample. This would also require what is called 'sampling with replacement'.
For anyone in Jamaican political polling to suggest that in his sampling methodology, each potential voter has the same likelihood of selection is simply false.
Simply false
This method would be rather expensive. Also, it would not work for Jamaica. You see, the interviewer would have to go into a district and ask for a person by name. In Jamaica's charged political atmosphere, I need tell no one in what danger such an interviewer or enumerator would find himself.
These truths still hold whether the polls claiming this error margin are done by Anderson, Johnson, Stoneson, Wignall, 'Seeall' or 'Knowall'.
This claim outside of the methodology described, is simply false. The 'predictive accuracy' our pollsters either claim or in the end demonstrate is an art, assisted by statistical technique, enhanced by wishful thinking of both pollsters and their audience.
But to the more fascinating work of bestselling author Nassim Nicholas Taleb. In his new book The Black Swan: The Impact of the Highly Improbable, Taleb carries forward his exploration of randomness put forward first in his previous bestseller Fooled by Randomness: The Hidden Role of Chance in Life and in the Markets.
What he indicates is that highly improbable and unpredictable events have massive impact. We human beings look for predictable events and behaviours. We like to think thatif we emulate the millionaires in town and the successful role models we might know it is possible for us to do well too.
Rare and unpredictable
But the big important events are rare and unpredictable. So when we read a prospectus and come across the disclaimer: "Past performance is no indication or guarantee of the future performance of this fund" we read it but the meaning rarely registers.
Michael Edesess is concerned that we throw away money by paying investment advisers huge fees. Edesess book, more narrowly focused, looks at this phenomenon purely from the investment perspective: The Big Investment Lie: What Your Financial Adviser Doesn't Want You To Know ends by telling us that the best strategy is simply to "invest only in true low-cost index funds."
I recommend these books highly to our investing public, pension fund managers and the like. Of course, none of this can ever beat insider information - the stacked deck - which, though often used in investing arenas is nevertheless criminal behaviour in all jurisdictions.
wilbe65@yahoo.com
