Opinion polls don’t always get it right when predicting results
IF the chattering classes and opinion polls are to be believed, the ANC is facing a difficult election battle tomorrow. Significant challenges include the headache of factionalism, reputational fallout from ongoing allegations of state capture and a legally delinquent SABC.
Ipsos-eNCA polls suggest these issues will cost the ANC dearly. It will not only lose its majority in three cities, but will be beaten by the DA by a sizeable margin – 5% in Johannesburg, 17% in Tshwane and 14% in Nelson Mandela Bay.
The weekly opinion poll results are obtained through cellphone sample surveys of 1 500 people selected from 2 500 possible participants.
Representing less than 0.04% of the more than four million voters registered in the three metros in 2014, this sample is very small.
While opinion polling is common across the world and intriguing, it is also fallible. Opinion polls failed to predict the outcome of the recent Brexit referendum, for example.
Ipsos recognises a potential sample error of up to 6% and its latest weekly poll shows 14% to 16% of respondents are undecided – implying that a swing of 15% or more towards any party could substantially alter the election’s results.
Previous elections are a more reliable indicator of future outcome because they are a reflection of all voters, not just a sample.
However, historical figures fail to reflect recent and potentially rapid changes in voter sentiment.
Notwithstanding the problems facing the ANC, it would be shortsighted to overlook the fact that the party received just less than two-thirds of votes in the 2014 provincial elections and sentiment would have to shift cataclysmically for this support to erode across the country.
Provincial elections are the most useful comparison to local elections because they require voters to register in a specific area. But they are not directly comparable, given differing turnout figures.
The risk for the ANC, therefore, is where it holds power on the margins.
The eNCA Election Challenge – to predict the outcome of the “Tale of Three Cities” – is apt, given that the ANC’s support in the 2014 provincial poll dipped less than 50% in Tshwane and Nelson Mandela Bay, and was 52.3% in Johannesburg. The ANC received 55% of votes in Ekurhuleni in 2014.
In Gauteng, support fell less than 50% in Midvaal and was 55% or lower in Randfontein and Mogale City.
Randfontein is to be merged with Westonaria into Rand West City in August.
These figures are particularly worrying for the ANC when contemplating the downward trajectory of its support in all of Gauteng’s municipalities between 2011 and 2014.
Service delivery protests in Gauteng declined last year, after a high tally in 2014 – second only to a 2010 peak. This is an encouraging outcome for premier David Makhura and his Ntirhisano war room strategy to deal with hot spot areas.
Like the rest of South Africa, however, there has been an uptick in service delivery protests in Gauteng this year, although it remains unclear whether unhappiness on the streets will translate into an erosion of support for the ANC at the ballot box.
Tallying up service delivery protests with protests against councillor candidates and demarcation issues shows that – aside from Ekurhuleni, where there is a relatively lower number of protests – the outcry is similar in the contested metros.
It is also not clear from protest data whether there is a relationship between the ANC’s political vulnerability in key metros and protests over the last 18 months.
Protests are notoriously anomalous and many communities resort to them because they feel marginalised from established political systems.
Censure of the ANC may not translate into votes for any opposition party, but in apathy – which will strengthen the proportional representation votes of opposition parties, especially in marginally held municipalities.
If there is widespread apathy among ANC voters tomorrow, then Johannesburg and Tshwane are most at risk of being won by an opposition coalition, and possibly Mogale City and Ekurhuleni. Historically, especially in local elections, opposition parties such as the DA are better at mobilising their constituencies to vote. In affluent communities, it is typically easier for voters to reach voting stations.
At a ward level, a lower turnout will not necessarily translate into losses for ANC ward councillors – but there will be a proportional representation boost for opposition parties, especially in the larger municipalities.
The exception is wards dominated by informal settlements, where disgruntled citizens may turn to independent candidates or the EFF to express unhappiness about their living conditions, more pronounced by the harsh winter conditions.
It is also possible the cold weather will deter older voters from voting, making it even more important to harness the youth vote.
But again, opposition parties will benefit only in the allocation of proportional representation seats, not at ward level, especially the EFF.
The National Freedom Party’s (NFP) disqualification following its failure to register for the elections will have profound repercussions for the outcome in KwaZulu-Natal.
The NFP won 11% of the KwaZulu-Natal vote in the 2011 elections. In 18 municipalities it garnered about 10% support or lower, and in 10 municipalities it won 20% of support and more.
It split the IFP vote, leaving 18 of 50 councils in KwaZulu-Natal “hung” in 2011, leading to a locking out of the IFP through a series of ANC-NFP coalitions. This will make the NFP’s absence from this year’s poll as dramatic as its arrival on the ballot paper in 2011.
The NFP has stated categorically that it will support political parties other than the IFP – including the ANC – but it cannot be assumed that its supporters will vote for the ANC.
Many NFP voters might not vote or will support independent candidates. With the Independent Electoral Commission reporting a 12% increase in independent councillors in KwaZulu-Natal (a significant number are former ANC members, following a recent provincial leadership battle), the NFP’s no-show is a significant risk to the ANC’s fortunes.
The ANC in KwaZulu-Natal gained significantly from the IFP vote being split, but this will not be a factor this year in the 18 coalition councils where no party had a majority in 2011.
With the EFF’s entrance onto the local government stage, and a difficult political and meteorological election season for the ANC, this may prove to be one of democratic South Africa’s most interesting polls.