Stephen Vasciannie

There has been no shortage of discussions about political polls, and about the role of the media in partisan politics. Some people believe these issues are apt to produce yawns, but others disagree. The latter are correct - the interplay between polls and media influence in Jamaica throws up a number of important and interesting issues.

In the first place, we are now officially sceptical about all polls. This scepticism takes different forms, and reflects different degrees of analysis. Some commentators on the radio take their cue from Motty Perkins, but - in fairness - do not develop their arguments in the way Motty does. So, when the poll debate comes up, we can expect at least two talk-show hosts immediately to say "the polls are only a snapshot". They say very little or nothing beyond this, so we cannot know what significance they draw from the snapshot point.

SNAPSHOT, SO WHAT?

We can do better than that. The snapshot point is true, but the value of the snapshot will vary according to particular circumstances. So, for example, a snapshot taken just before a general election may provide a clear indication of likely results and thus serve as a guide to political action. But, of course, this depends on the quality of the snapshot. I believe that the media should encourage some of its analysts to develop the means of assessing the statistical quality of the polls that are put before us.

A political election poll is a sample, built on the assumption that the sample - properly drawn - can reflect the views of the population. In statistics, if the sample is drawn on a random basis, then the evaluator can assess the margin of error for the sample. This is done by our pollsters, and on this basis they tend to tell us that their polls have a margin of error of plus or minus three per cent. Is this really true? And how can we test it? We can test it by looking up the statistical tables: If you survey about 1,000 people as representative of a population of, say, 750,000 voters, what is the margin of error?

The pollster may say, however, that this is insufficient because (s)he is not really undertaking a purely random sample. As economist Wilberne Persaud reminded us in last week's Financial Gleaner, Carl Stone could probably have been vulnerable on this point. But Stone's response was that his sample was drawn on the basis of standard statistical techniques as well as on the basis of his extensive knowledge of the Jamaican political environment. Stone knew the statistical sweet spots - so he could say, for instance, that certain parts of certain constituencies served as weather vane, giving a good idea of which way the constituency and the country was swinging.

The challenge, therefore, is for Stone's heirs and successors to indicate how they determine the actual persons to be interviewed as representative of the population. Persaud calls on the pollsters to do this, and I support him.

ELECTION STRATEGY?

But the debate on polls has not really moved in this direction. On the contrary, in at least one case, questions have arisen about whether the pollsters are part of the election strategy of one side in the electoral race. And, in another case, two editors have publicly argued with each other about the reasons for non-publication of certain poll results. I am not in a position to draw firm conclusions on these issues, but one thing is clear: poll-taking, either as a political art or science, has seriously been weakened in the build-up to the next election.

Finally, the media houses should sharpen their sensitivity to the possibility of bias. Of course, it is their prerogative to be biased if they want to be, but sometimes they seem to go overboard without realising it. If a certain politician coughs, it is in the news as important, positive news, while, on the other hand, we hear precious little about the campaign activities of many others.