Holes in 2010 Afrobarometer election poll

A recent poll putting President Museveni far ahead of Dr. Kizza Besigye has generated controversy everywhere. In today's Daily Monitor, Timothy Kalyegira exposes the gaping holes in the said poll.

In 2006 a few weeks before that year’s general election, the Daily Monitor commissioned the public opinion survey firm AfroBarometer and their Kampala partner Wilsken Agencies, to conduct a poll on the popularity of the presidential candidates.

The results showed NRM candidate Yoweri Museveni in the lead. I sent an sms to the then Daily Monitor Executive Editor Peter Mwesigye warning that he should have taken better care in selecting which company to conduct a survey on behalf of the Daily Monitor.

I worked at Wilsken Agencies for a few months in late 1999 and over something as vital as a tense general election, I could not vouch for the integrity of the findings. The latest AfroBarometer opinion poll of the standings of the 2011 presidential candidates -- published across the front pages of Uganda’s two leading daily newspapers Daily Monitor and New Vision on Friday, December 17, 2010 to much public anger -- has not only vindicated the reservations I expressed to Peter Mwesige in 2006 but, I think, damaged the reputation of AfroBarometer beyond repair.

Early glitch
The poll was conducted between November 18 and December 6, 2010 in 71 of Uganda’s 112 districts and sampled 2,000 Ugandan adults. The result were that, as the New Vision put it in its lead story, “If Ugandans were to vote today” NRM candidate Yoweri Museveni would score 66 percent of the vote, FDC/IPC candidate Kiiza Besigye would get 12 percent, DP-1 candidate Norbert Mao 3 percent, UPC candidate Olara Otunnu 3 percent, Uganda Federal Alliance candidate Beti Kamya 1 percent and DP-2/independent candidate Samuel Lubega, PPP candidate Jaberi Bidandi-Ssali, and PDP candidate Abed Bwanika would all get 0 percent.

Let us start off with the most obvious and suspicious facts of this AfroBarometer poll. The first is that the numbers are all complete numbers --- 66, 12, 3, 3, 1, 0, 0, and 0. What sort of opinion poll would that be that would return perfect numbers without mathematical fractions? Notice that in Friday’s same edition, the New Vision stated that in its survey of June to July 2010, Museveni stood at 52.72 percent, Besigye at 16.06 percent, Olara Otunnu at 3.14 percent.

The truthfulness and impartiality of a survey by a newspaper such as the New Vision that has, since nomination day, insisted on publishing the campaign photos of only NRM candidate Yoweri Museveni, can be questioned. But it at least looks like a proper survey, with regular and fractional numbers.

The second and much more glaring flaw about the 2010 AfroBarometer opinion poll --- one that I am surprised has not been pointed out amid the public outcry on Friday and all weekend against the results -- was the total percentage number.

If the AfroBarometer poll is credible, how come it does not add up to 100 percent (if the Latin expression per cent means “out of a hundred”) in some cases? NRM is, for instance, said to command 72 per cent of the people’s love, FDC 28 per cent and UPC 15 per cent! That is more than 100 per cent. I suspect a deeper look at the finer details will reveal more of such inconsistencies.

This Afrobarometer poll reminded me of the heated 2002 phone exchange on Monitor FM (predecessor to KFM) between the then exiled FDC president Col. Kizza Besigye in South Africa and the then Chief of Military Intelligence, the late Brig. Noble Mayombo. Mayombo narrated to the listening audience specific days and times at which Besigye had allegedly made phone calls to the rebels of the LRA, ADF, and a new, mysterious rebel group called the PRA. Somehow, Mayombo’s intelligence report showed that all Besigye’s phone calls always fell at the top of the hour. It was “On July 12, at 3:00p.m., you called Vincent Otti. On October 7, at 11:00a.m., you called James Opoka.” The calls were never as irregular as 12:23p.m, 6:38a.m, 9:19p.m. They were all perfectly rounded numbers.

The immediate and obvious conclusion: the AfroBarometer opinion poll was in professional social sciences and even primary school mathematical terms, wrong.

This leads to the next question: what was the intention behind this poll that is calculated in percentages but is not quite 100 percent? Once again it is obvious: while the western donors who commissioned it had intended it to be an opinion poll in the professional, informative sense of the word, somebody might have got wind of it and possibly pressured or threatened AfroBarometer to falsify the results, which we now see do not even add up to 100 per cent.

Destroyed credibility
What this suspected fraud is going to do, apart from forever destroying AfroBarometer’ credibility in Uganda and future business prospects, it will if anything harden public anger and determination against any rigging during the February 18, 2011 general election.

It will make the donors and news media not only less gullible about whom they commission to conduct opinion polls, but also bring greater alertness to the reality of rigging and manipulation of final results at the Election Commission computer and tallying rooms in Kampala.

It will be a most interesting irony that what started out as a poll intended to create the impression of an inevitable Museveni victory and have a demoralising effect on the millions of opposition voters will turn out to be the very catalyst not only to greater watchfulness during the voting and counting. It could be one of many sparks for violence that might erupt should such obvious efforts at fraud happen on and after February 18.