BEWARE OF STRAW TELEVISION POLLS.

A number of television anchors from reputable media houses used “news polls” to consistently predict that the opponents of the proposed constitution would win the August plebiscite. The leadership of the No Team buoyed up by predictions of their success by the hitherto straw vote of the TV polls, rubbished as doctored the findings conducted by professional pollsters that indicated otherwise. Yet on the referendum day, the actual referendum results confounded both TV pollsters and the leadership of the naysayers. Proponents of the proposed constitution won with an unprecedented landslide margin.

This development has prompted many people to question whether it is proper for any Tom, Dick and Harry to purport to predict the outcome of an election based on a few text messages send by viewers either in support or against the proposed constitution.

Pundits fear that given the average or low literacy levels of many people, they are likely to believe such findings as gospel truth. The losers are likely to interpret an unfavorable outcome as arising out of an election malpractice, hence the risk of polarization. Just to put the scenario into proper perspective, a friend of mine bluntly told me that polls (TV) had forecasted a very big win for the naysayers and that any other outcome would be unacceptable. As I look back at the just concluded referendum, I shudder at what would have transpired had the referendum outcome been closely contested.

Unbeknown to many people, polling is a painstaking exercise that takes days or even months. This means that it can only be carried out by professionally trained pollsters. But even for trained pollsters, it still takes years of hard practice for them to release credible poll findings. It involves sampling, sampling units that must be representative, administration of a carefully thought out questionnaire, and interpretation of the data generated among many other professional requirements. It is not a matter of just receiving text messages and phone calls from viewers and hastily drawing untenable conclusions.

In any case, less than ten per cent of the Kenyan households have access to phones. Fewer even have television sets; hence participants in such a TV poll cannot be a true representative of the electorate. Furthermore, such findings suffer from credibility since they lack external validity and can thus not be generalized beyond the limited scope of the TV poll itself.

TOME FRANCIS,

BUMULA.

http://twitter.com/tomefrancis