The Council for Scientific and Industrial Research (CSIR) has enjoyed considerable success in its operation to predict the outcomes of the recent local government elections in the crucial period (for the media and political analysts being interviewed) between the closing of the polls and the announcement of the final results. “We were accurate to within 1% of the final result in almost every municipality,” highlights CSIR planning head Dr Zaid Kimmie.

“We are trying to bridge the information gap in the early hours after voting has concluded in an election, when you only have early results coming out, which will not reflect the final result. We’re trying to predict that final result,” he elucidates. “We also want to show how useful mathematics and statistics are.”

The CSIR developed special algorithms for this purpose. These were originally created by a small team led by now-retired CSIR nuclear physicist Jan Greben. These have since been incrementally improved. The algorithms are run on a standard desktop computer, but a laptop could be used instead. Indeed, the CSIR election prediction team jokes that they could be run on a smartphone, if they developed an app for it. “It’s not computationally complex,” explains Kimmie.

The algorithms do not make use of any polling data. Indeed, in South Africa, exit polls are illegal, except those conducted by the Human Sciences Research Council on behalf of the Independent Electoral Commission – and these polls do not ask how people have voted. The CSIR process simply extrapolates from districts that have reported their results to districts that have not reported their results. However, the practice is more complicated that the principle.

“We use some quite fancy maths,” he points out. “We split the country into little groups of voting districts and we extrapolate within these groups.” To give a simplified example: districts that, say, have previously voted for the Democratic Alliance (DA) would be grouped together and the outcomes for those which reported their results early would be extrapolated to those that had not yet reported. In reality, the districts and clusters are defined in a more sophisticated way than this, including by use of socioeconomic factors.

The CSIR had only one significant failure in its predictions – for Tshwane (the metropolitan authority that includes Pretoria). “We were still within 1% of the actual result, but we got the order of the parties wrong!” The predictions team forecast that the African National Congress (ANC – the ruling party in the metropolitan area) would get the most votes. In fact, it was the DA that came first.

“We overestimated the ANC’s performance by about 1.5%, or 15 000 votes (out of a metropolitan electorate of about one-million),” he notes. “The DA won Tshwane in the end by about 11 000 votes.”

The one problem with the prediction system concerns the performance of the small parties. “The 1% error is generally insignificant for the big parties, but significant when it comes to the smaller parties.” The CSIR team defines big parties as parties that get above 7% of the national vote (namely the ANC, the DA and the Economic Freedom Fighters), as well as the Inkatha Freedom Party, which is also very significant in the province of KwaZulu-Natal. But there are numerous small parties that get only 1%, 2%, 3% of the vote nationally, provincially and often locally. “So this makes it difficult to predict the performance of the smaller parties,” he points out.

“This is intended as a serious aid to the media,” observes Kimmie. “We try to give good information early on to help the analysts. And it’s nice to be first with the results!”