Site menu:

September 2019
« Apr    

Recent Posts

Recent Comments

Site search



Of Opinion Polls & Politicians – Stop Shooting the Messenger!

By Matunda Nyanchama
Toronto, Canada
January 5, 2007

It has become a predictable trend in Kenya whenever an opinion poll is released. Those favoured by the poll will harp about it as if they have won already, especially where it pertains to political support.

On the other hand, those not favoured by the poll seem to take a hard line with attacks targeted at the polling agency, its conduct and the motive behind the polling.

The situation is so predictable that even those who have celebrated in the outcome of one poll at a point in time come out with guns blazing when a new poll result does not favour them.

Attacking polling agencies, in my view, is “killing the messenger” and failing to hear the message. Our political leaders do a disservice to themselves and the country when they attack polling agencies.

Opinion polls should be seen for what they are: a snapshot of public perceptions at a point in time. Indeed, the common wording is “if elections were held today …” which suggests that things could change or may have changed where poll numbers differ from previous ones. And poll numbers change precisely because of added information that may change perceptions of those polled.

These perceptions convey (usually) useful information for those to who the polls pertain.

Adding new information can change perceptions, an important issue that political leaders seem to overlook. Political leaders harping about the same issues, the same facts and the same song may add little to change perceptions. It is not new information they are presenting and hence carries little value in shaping opinion. For example, if we already know that X is a thief and we have known this all along, continuing telling us that X is a thief adds little else to our information bank. However, if one came back with new information (say) “in addition to X being a thief he also a murderer” then one has a chance to change perceptions.

There is a Kiswahili saying that “(mavi) ya kale hayanuki”, which literary translates to (excuse the language) “old shit does not stink”. If it has created a stench or sweet aroma once, it is not likely to cause more stench/sweet aroma later unless augmented with something else!

Political contest should be focused on creating an informed public on the values that the political leaders hold and their plans to solve the problems facing the electorate.

In opinion polls, respondents’ reaction depends largely on what they deem significant; issues they have come to terms with often will have little weight in these reactions.

Question: is constitutional reform the most urgent matter for Kenyans as the politicians are telling us? It is possible that Kenyans remember the fiasco that constitutional reforms process took, which ended up with the defeat of government in the referendum; and perhaps realize that the current political crop has no capacity to resolve the constitutional reform stalemate. And if this were true, their reaction would be to downgrade the importance of this issue in relation to other issues like education, health care, transportation, security, agriculture, etc. in opinion polls – shelve it away until such a time that there is leadership capable of attaining results.

This is not to say that reforms in the constitution are not important. Our constitution needs change for the benefit of Wanjiku. I doubt Kenyans want (in the immediate future) to go through another round of what they have been through, with the associated expense and political heat. This is especially so considering the polarization of current political leadership. It may well be that the present crop of politicians will have to exit the stage for real constitutional reforms to happen!

Opinion polls are useful if they convey information. Based on Shannon’s information theory, if an event is likely to happen and it happens then the information content of the event is low. However, when an event is least expected but happens all the same, the information value of the event is high. It is the reason; perhaps, people react negatively when not favoured by the poll: they expect to do well but the poll results suggest otherwise.

And this is where our political leaders should pay heed: when they are surprised by the results of an opinion poll, they should sit back, ask questions about the polling indications and address the issues raised by the results. Where the poll favours them, they should reinforce what good they are doing; where they appear to be failing, they should put in place mechanisms for improvement.

This is important, in part, because political leaders and their supporters have their world view of things: a world view that may be completely at variance with Wanjiku’s.

That said, opinion polls are not error free. Errors can result from the sample chosen; the methodology used for polling, the way the questions are worded; non-response and tailored responses of those polled.

The sample size and how well it represents the population usually yields an error expressed in percentage terms. A “dead heat” results in cases where poll outcomes are within the error.

Coverage error and bias result from the methodology used. For example, if one carried the poll through phone calls and yet there may be a significant number of people without telephones, then there will be an associated bias and error.

Polls results can be shaped by the nature of questions asked for questions can be tailored to result in specific outcomes. In some cases those polled fail to respond while others deliberately respond with answers that they know are not true.

All these are pitfalls of polling that are widely acknowledged. It is why one opinion poll, conducted by one firm should not be considered conclusive. However, if several polling agencies/companies obtain the similar results over and over again, then the poll results become substantially reliable. Those who complain about poll outcomes should commission their own with a different agency to verify a given poll outcome.

Opinion poll numbers can change with time based on changes in the population’s perceptions of the leaders or issues at hand. Information garnered from the subjects of polling can have either a positive or negative effect on the poll outcomes. It is important then that political leaders understand the messages they are putting out and whether or not these message help their causes or not.

Is this to say that politicians should be guided by poll numbers?

Absolutely not!

Leadership is about vision and means for attaining that vision. Leaders are supposed to visualize some end (the “mountain top”, to borrow Martin Luther King Jr’s words) where they are taking their flock. And at some point in time, the leader’s view of the mountain top may not be in sync with that of their flock. Such leaders will be perceived to be ahead of their time. For example, Galileo’s postulation that the earth revolved around the sun was regarded as blasphemous at the time. Yet today we take it for granted.

Polls are good indicators of what the population thinks; they are subject to error and manipulation. However, they convey important information that leaders should listen to, consider and understand how to address the issues raised by the polls. Attacking polling agencies is equivalent to “killing the messenger” and avoiding the message.

Through polls wananchi are sending key messages. Politicians should listen.

© January 5, 2007 Matunda Nyanchama


Matunda Nyanchama, a Canadian-based information security professional, is a past President of the Kenyan Community Abroad (KCA). He can be reached at

Write a comment