The latest AFROBAROMETER poll results' credibility questioned

According to the latest results of Afrobarometer, if elections are to be held now, ZANU-PF presidential candidate would win by 44%, MDC-T Chamisa 28%, MDC Alliance 2% and others 1%. This should be ridiculous, MDC-T Chamisa party is on record declaring that it will field its candidate Nelson Chamisa under MDC Alliance. The inclusion of MDC-T Chamisa and MDC-Alliance could have been done deliberately to confuse the respondents. Why should respondents vote for MDC-T Chamisa party president 28% yet they know that Nelson Chamisa will be standing under MDC-Alliance. Why were they silent about MDC-T Khupe?

This piece seeks to demonstrate that the latest Afrobarometer results failed the test of credibility and conformability. The heart of survey polls is in the sampling techniques. The way Afrobarometer handled the sampling procedures forced me to declare their poll results null and void.

The Afrobarometer claimed that they conducted their survey with 2400 respondents (but did not confirm whether they used the voters roll) and the margin of error was +-2. They further claimed that they selected the 2400 respondents via random sampling (that is where the credibility and conformability of the survey can be questioned by the evidence below)

Afrobarometer seems not to know how randomization works. Under random sampling, each and every registered voter should have equal chance of selection. The selection is mostly done in two ways; first it can be the 'blindfold method' they could have taken all the names or IDs of registered voters (5,310,734) into a basket and picked randomly their 2400 samples or respondents. After picking them, they should then make a call to them and agree to meet for face to face interviews and I am confident that Afrobarometer did not do that. Evidence on the ground suggests that they could have used convenience sampling (interviewing any registered voters that they met in provinces)

The second method for random selection is via random tables that can be created and sorted by the computer. All the names of the 5,310,734 registered voters should be inserted in the random table and the 2400 respondents were supposed to be selected/picked randomly. After the selection of 2400 respondents, afrobarometer was supposed to call them for face to face interviews (there are registered voters who may be out of the country/some are dead people who are still in the voters roll) if selected by blind pick method or random tables, Afrobarometer was supposed to look for them and interview them . Afrobarometer did not give us any information about such cases.

According to Afrobarometer, of the 2400 respondents, 50% were males while 50% were females. This is a rare coincidence (it is very difficult for random sampling to pick equal number of male and female). The kind of coincidence may only happen if there is cheating during the selection. The danger of using randomization (without stratification) is that, all the samples may fall under one gender or in urban area compared to rural. I am convinced that Afrobarometer decided and agree to have 50% males and 50% females before the selection of respondents. Why then lying that it was via randomization?

Afrobarometer later agreed that they stratified the samples according to regions (provinces) that was an indication that the random sampling was not properly used by their team as per their claim under methodology section. I want to take this opportunity to request Zimbabweans to join me in condemning Afrobarometer, they violated the sampling procedures. This will make their findings to become null and void, their credibility has been tainted.

According to Afrobarometer, 62% of respondents were convinced that Zimbabwe is heading for a wrong direction and 63% blamed ZANU-PF government for the wrong direction. Surprisingly, that ZANU-PF government that is taking us to a wrong direction will win by 44% ahead of MDC-T Chamisa 28% followed by MDC-Alliance 2%. Something is not adding up. The voters blamed the government for bad governance but voted overwhelmingly for that same evil government.

However, Afrobarometer claimed that 26% of the voters may not be decided, that means there is still time for opposition to work hard and capture them to their side.

Conclusion
The results for Afrobarometer are questionable and should be declared null and void. If they are correct, chances are that they were doing their face to face interviews at ZANU-PF functions/rallies or they were paid to 'cook the results'. Therefore, I can now safely declare the recent Afrobarometer results ZANUPFBAROMETER

I want to urge police to arrest Afrobarometer team and send them back to grade seven, our constitution is against false claims and the cooking of data like what is happening at Herald.

----
Don Chigumba is a mixed methods research specialist can be found on twitter @Donchigumba