Who trusts the opinion poll?

Opinion polls are gaining currency in the EAC region and seem to have taken root in Kenya, markedly around election time, where they continue to fascinate and divide opinion about their reliability.

Thursday, July 05, 2012
Gitura Mwaura

Opinion polls are gaining currency in the EAC region and seem to have taken root in Kenya, markedly around election time, where they continue to fascinate and divide opinion about their reliability. A section of the Kenyan politicians, often loud and irreverent in their opinions, have been derisive about the polls. Some, predictably those trailing in the polls, have often questioned their veracity, especially as they relate to the popularity of a political party, or predicting the most likely presidential winner.  It becomes an amusing back and forth as the poll leaders rebut with effusive politicking about why they are ahead in the polls as the election campaigns gather pace. But with efforts to legislate the monitoring of electoral opinion polls in the country, despite the skeptics, it is with the firm understanding that they are absolutely necessary and cannot be dispensed with.In the days when public opinion didn’t matter to rulers through time, very few leaders had any idea that a revolution was brewing until it exploded with the mob "storming the Bastille.” This is what was replicated during the Arab Spring.  The story has been narrated of how the opinion polls came to gain their respect, pitting George Gallup, the father of the scientific poll, against the famous magazine, The Literary Digest.During the election of Franklin D. Roosevelt in 1936, Gallup’s informed opinion was that Roosevelt would "bury” his rival Alfred Landon. He, however, went on to famously predict what The Literary Digest would predict as to who would emerge the winning candidate. The magazine’s prediction, as foretold by Gallup, was Landon 56 per cent and Roosevelt’s 44 per cent.On his part, Gallup’s prediction was 63 per cent for Roosevelt against Landon’s 37 per cent. Roosevelt took the presidency by these very margins.This is how The Literary Digest had come up with their prediction. They picked names from records of car registrations and telephone directories and sent out millions of postcards to the people whose names were picked.Note that this was during the Great Depression at a time when the vast majority did not own a telephone or car in the US. The absurdity of the magazine’s method was, therefore, that the outcome of the opinion poll would only be of the people who owned telephones and cars.  Gallup’s method took a small sample that covered in due proportion the diversity of the population, including white, black, men, women, old, young, married, single, ethnic minorities, rich, poor, etc. It turned out to be astonishingly accurate, paving the way for the current, more updated methods of taking opinion polls.Today, the typical sample size is 1,000 to 2,500 national adults in all their diversity and regional representation, adjusted according to an anticipated margin of error. Though more people can be polled, it has been shown that the accuracy of the estimates derived only marginally improves with the larger sample sizes. Therefore, polling a million people or one thousand may yield the same result.The accuracy of public opinion polls may not always be exact, but they offer a fair picture of a population’s perception on any given issue if properly conducted.