From: Francis Tome

Dear Sir/Madam,

If credibility is defined as believability, then at the moment, credibility is the most of the worries that the pollsters must content with. Pollsters have attracted the odium usually reserved for prophets. Each faction contesting the up coming constitutional referendum believes it has what it takes to effect the biggest surprise of the election that will confound pollsters and pundits alike by winning the referendum by margins far much bigger than the pollsters` current estimates.

Much as I wish to dismiss such sentiments, I think that it is imprudent that we do not succumb to the dark impulses that lurk in the works of the pollsters lest we fall in the worst possible faux pas. Experience has over time shown that polls unwittingly or otherwise overestimate or underestimate the strength of the parties contesting in an election. Take the 1948 and 1992 elections of the United States of America as examples. Pollsters either greatly underestimated or overestimated the candidates` margins of victory. It is such inaccuracies in estimation that made people begin to cast shadows on the credibility of opinion polls.

To begin with the sampling procedures as well as inconsistent sample sizes have led to variations in results. Secondly, even if the basic poll data are valid, analyses of the data is at times too casual and quite superficial. In some instances, the raw data are simply presented as the public’s “opinion” without deep and careful analysis to probe nuance and possible bias.

Bias may be manifested in the manner in which the pollster constructs questions to measure the respondent`s responses. It will for instance be unsurprising if the percentage of the public in favor of “forbidding” hate speech were found to be lower than the percentage in favor of “not allowing” hate speech. From the foregoing, a pollster`s question can significantly alter an individual`s measured response.

In addition, results may be correct at the time they are collected, but as we all know, a day in politics is such a long time. Estimates are unlikely to be inconclusive since the vote can shift at the last moment.

The most stinging criticism is however leveled at the polls` appropriateness level. Such polls have been known to influence voters to favor a faction that seems to be enjoying a notable popularity at the time. The information that a certain faction is far ahead in the polls may discourage voters from voting at all or encourage them to vote for that particular faction that is enjoying a roller-coaster courtesy of the pollster thus affecting the results of the election. It is for this reason that I agree with those who are of the opinion that legislation be put in place to bar opinion polls from being conducted two months to the election date. This will ensure a free and fair election.

TOME FRANCIS,

BUMULA.

http://twitter.com/tomefrancis