The Star (Nairobi)

16 February 2013

Kenya: Tyranny of Numbers Versus Opinion Polls

Researchers Tom Wolf and Caesar Handa dismissed the "tyranny of numbers" analysis, put out by veteran political commentator, Mutahi Ngunyi, on Citizen TV, saying it was not only unscientific but was also unreflective of historic voting patterns.

Handa of Strategic PR said in voting, ethnicity is not the only variable. Other variables like level of education, gender and age can change voting dynamics. He says this is why pollsters publish the demographics they work with.

Ngunyi's "Tyranny of Numbers" did not have any reasonable demographics according to Handa. Besides, Handa says there are other ethnic blocs apart from the ones "Ngunyi lumped together" which will participate in the polls. He says in a democracy the principle of universal suffrage, one man one vote is key.

"It did not have any demographics. It was based on one assumption that Kenya is made up of certain ethnic blocs which vote in certain ways, that assumption is wrong."

According to him, only a scientific poll can accurately predict results of an election that has not yet been held. He says by and large, scientific polls are credible to the extent in which they remain objective and in which they adhere to minimum research standards.

Any body set about to do serious research who gets the correct sample size, distributes it appropriately, frames questions properly and does proper data analysis will get results "close as possible to what voters will actually vote."

Asked about lack of credibility arising from pollsters use of census results as opposed to voters register demographics, Handa concedes that ideally, the frame ought to be registered voters. He however says in the absence of updated voters register, the next available frame would be the population census because it shows the proportion of Kenyans who are 18 and above.

"And its a close, good indication of voter figures. That is why IEBC uses the population census to estimate the number of people who should be registered in particular regions. We may not have the updated voter demographics but we know we have 14 million voters and we know where they are," he said.

He said although opinion polls have chances of biases here and there, this is catered for in the margin or error calculation. The margin of error is calculated on the basis of possibilities of questions not being framed properly, missing data and the way questions are put among others.

Asked to give "tyranny of numbers" a margin of error, Handa said: "There is nothing like margin of error in that thing. That thing was not scientific. Even historically speaking I think it was wrong. He has not even come out to defend it."

Wolf who is traditionally associated with Ipsos Synovate group is opposed to "tyranny of numbers" right from its framing. He says "tyranny" suggests somebody is being oppressed and yet in a democracy numbers, any numbers or even a number matter.

For him the correct phrases to use is "safety of numbers" or "importance of numbers."

"Yes it is true certain ethnic groups are larger than others and therefore candidates from those groups may have some advantage than others. Fortunately, there is no single ethnic group in Kenya that approaches a quarter of the population," he says.

On this basis, Kenya would never suffer the tragedy of Rwanda- polarisation of two large groups. The "safety of the Kenyan numbers" is that no ethnic group on its own can ever aspire through the vote to achieve dominance.

Wolf also challenges Ngunyi's assumption that certain communities would vote for their presidential candidates man-to-man and woman-to-woman.

"We know from our survey data that there is no ethnic group in this country which is 100 per cent united behind their candidates. Within every ethnic group there are free thinkers, deviants and independent minds. It's false and misleading to make such a blanket assumption," he says.

He says the best example is UDF presidential candidate Musalia Mudavadi. Three opinion polls have in the recent past showed that he is not the leading presidential candidate in Western and in his Luhya tribe.

And so for him, scientific polls are the only credible means of crunching numbers. He admits that polls have statistical margin of error. He said there are also other issues such as excluding non English and non-Swahili speakers from the polls with the risk of creating a bias.

There are also other issues like respondents lying. The main weakness for voter intention surveys is voter turn out, he says. Although its very rare that voter turn out is rarely likely to go below 65 per cent and beyond 75 per cent, poll results reflect an extraordinarily higher results than this.

According to Wolf, the past record of pollsters in the country however bails pollsters out of these credibility issues.

"Going back to 1997 election, the only company that was polling, Strategic PR, got it basically right that Moi would be re-elected but with a smaller margin than 1992. They were outside margin of error but not by much," he says.

He says ever since and starting in 2002 elections when many pollsters joined the business of polling, all the pollsters always got it right but sometimes with minimal deviations from the margins of error.

Wolf is also ready to concede on one thing; that pollsters are challenged by closely called polls. No pollster can predict the results in a very closely called election because candidates can be defeated by 1 vote or even 100 votes.

"The closer the election, the more likely it is that pollsters cannot produce any prediction. And they should say so. I was widely quoted in the last election saying it was too close to call. And up to now, no one know who won that election."

He say the real "safety of numbers" is in many pollsters doing the same thing so that comparisons can be made across them. He says this should and include the possibility of doing "a poll of polls."

Like Handa, Wolf cannot give any margin of error to Ngunyi's figures because there was no sample used. He says Ngunyi did not even check past voting trends in each tribal bloc.

"So yes, lets look at the importance of numbers but lets not worship them and give them an importance that goes beyond their meaning at least as shown in an empirical research," he says, signing off.

Ads by Google

Copyright © 2013 The Star. All rights reserved. Distributed by AllAfrica Global Media (allAfrica.com). To contact the copyright holder directly for corrections — or for permission to republish or make other authorized use of this material, click here.

AllAfrica publishes around 2,000 reports a day from more than 130 news organizations and over 200 other institutions and individuals, representing a diversity of positions on every topic. We publish news and views ranging from vigorous opponents of governments to government publications and spokespersons. Publishers named above each report are responsible for their own content, which AllAfrica does not have the legal right to edit or correct.

Articles and commentaries that identify allAfrica.com as the publisher are produced or commissioned by AllAfrica. To address comments or complaints, please Contact us.