Defend Truth


Final Forecast – Counter-factual findings: ANC 53%, DA 24%, EFF 14%

African National Congress (ANC) supporters wait for their leader during the final ANC rally prior to the elections, in Johannesburg, South Africa, 05 May 2019. The ruling party held its final rally ahead of the 08 May general election. EPA-EFE/YESHIEL PANCHIA

After its final poll on Saturday, the Institute for Race Relations has the ANC squeezing beyond the 50% mark, the DA holding the Western Cape and the ANC losing Gauteng.

The Institute for Race Relations (IRR), whose poll is run by Victory Research which is owned by former DA dynamo Ryan Coetzee and which polls globally, predicts the EFF will put 8 percentage points on its 2014 result, ending at 14% in the national election.

The think-tank says the ANC will win 53% and the DA 24% nationally which will mean that the governing party will reduce support and the DA will grow very slightly from its 2014 outcome of 22,23%.  The IRR poll is a counter-factual one in that it turns up lower results for the ANC and higher outcomes for the EFF than three other polls do. Its results are based on a respondent pool of about 2,200 people, all registered voters.

The IRR poll suggests the ANC will get only 45% of votes in Wednesday’s election in Gauteng – this is against the ANC’s own poll of 3,200 respondents which has found that it will win a majority of between 53% and 56% in the country’s economic heartland.  The DA gets 26% in Gauteng in the poll released on Monday, suggesting that the official opposition party is unlikely to meet its key goal of the campaign – to be the biggest party in the province and to be able to form a government. 

This is the most fluid national election to date, in particular with regards to the ANC and EFF, who have traded support one way and the other among around five to eight percent of black voters since September,” said the IRR’s head of politics and governance, Gareth van Onselen.  

From Saturday April 28 to Saturday May 5, pollsters undertook research to arrive at a “five-day moving average” outcome to increase the randomness of the sample (an important measure) and arrive at a 95% confidence level.  

Here’s what the pollsters saw: “For the first half of the week, the ANC managed effectively to squeeze the EFF.  From Saturday to Wednesday, the ANC had consolidated its support up to 68.9% turnout.  In turn, the EFF had dropped from 14% to 10% on lower turnout over that period. From Wednesday (May 1) to the final Saturday (May 4), however, the EFF fought back, systematically increasing its support from 10% to 14%.  The DA remained fairly stable throughout the week, averaging 24% on lower turnout and occasionally hitting 25%.”

In the Western Cape, other polls suggest the DA is fighting to achieve a majority in the provincial election, but the IRR says it will romp to a comfortable victory.  

While the party started at a 51% outcome in the IRR’s April poll in the only province it governs, by the time the researchers put down the phone on their final respondent, the party was polling for a 54% win.  

This is what an effective squeeze looks like,” said Van Onselen who said the DA’s tactics to push up support amongst the coloured majority in the province was working.  

The 2019 opinion and voting polls are beginning to reveal three things: the ANC will win the election; the EFF will make the biggest gains and the DA will not achieve its goal of being the largest party in three provinces.  

The DA had planned to use Election 2019 to prove its mettle at governing. The party moved its headquarters to Gauteng from Cape Town to put its energy into securing a victory in the powerhouse province.  The party won two of the major cities in Gauteng in 2016 which made it target a provincial win.  Gauteng is responsible for over 40% of GDP in South Africa.   

The IRR has stressed its survey is not a prediction and that its outcome is not absolutely definitive; it has assumed a voter turnout of 70%.   The difference between the IRR poll and the Ipsos polls are the following: the IRR respondent pool is representative but smaller; the IRR poll polls people by mobile phone numbers while Ipsos polls in face-to-face interviews using mock ballots. DM