South Africa

Zuma’s apparent minimal collateral damage as new polls point to potential ANC landslide

By J Brooks Spector 24 March 2014

The most recent polling data seems to indicate a major, sweeping victory is in the making for the ANC in the nation’s next election. If true, it is a real turnaround for the party and could demonstrate its resilience in spite of being hammered by the Nkandla controversy. J. BROOKS SPECTOR takes a closer look.

Despite strenuous efforts by both the Democratic Alliance and the Economic Freedom Fighters to register sympathetic new voters and rev up support for their parties’ respective electoral chances, new polling data seems to be pointing to the African National Congress’s chances of achieving a virtual two-thirds vote from South Africa’s electorate.

These results from the Ipsos polling organisation, first released by the Sunday Times on Sunday, seemingly contrast starkly with earlier polls by the very same Ipsos (as well as some private polling by the DA) that had pointed to a flailing ANC and a rising DA share of the vote.

According to the newest data, and significantly at variance with widespread public speculation by numerous commentators and analysts, the ANC is now pinging at a 66.1% level of voter support, with the DA trailing at 22.9% and only some 3.7% going to the new Economic Freedom Fighters of Julius Malema. There had been optimistic predictions from some DA corners that the party was headed for as much as a 30% share of the vote. If this newest survey data is accurate – and if it holds up through the next month and a half into the election on 7 May – that could actually give the current majority party a marginally higher level of electoral support than in the country’s national election in 2009 when, once the ballots were recorded, the ANC weighed in at 65.9%.

There is, however, a serious caveat noted by the newspaper in reporting this data. The interviews with the people contacted for this survey had all taken place before the release of the final Nkandla report by the Public Protector over who had paid for the so-called fire pool, the chicken run and the automated cattle kraal, in addition to all those other personal lifestyle enhancers.

Nevertheless, Ipsos says its data seems to indicate that ANC-inclined voters tend to feel the drama over Nkandla is more an issue specific to President Jacob Zuma personally, rather than something that was his party’s responsibility. Those results should also be read with another survey from February, also by Ipsos, that around half of South Africans believe the president and his government are not carrying out their jobs effectively, based on face-to-face interviews with some 3,564 randomly selected adults. If nothing else, this jumble of data would seem to indicate a certain degree of confusion among voters as to whether it is the party’s standard bearer or the party that should be the key, determining variable for voters when they make their mark.

In response to these newest results, the DA’s national campaign manger, Jonathan Moakes commented, “Ipsos Markinor has a record of consistently understating DA support and overstating ANC support. In 2009, they predicted that the DA would get 13% and that the ANC would get 67%. The 2009 election result gave the DA 16.7% and the ANC 65.9%. In 2011, they predicted that the DA would get 19.1% and the ANC would get 58.9, and after doubtfuls and people who refused to answer were allocated in proportion to party support, had the DA on 22.6 and the ANC on 69.1. The 2011 election result gave the DA 24.1% and the ANC 62.9%. Our internal polling reflects that today’s poll is continuing this trend. We are confident that the DA will succeed in reducing the ANC’s share of the vote to well below a two-thirds majority.”

Nonetheless, News 24 reported, “Unless a dramatic shift in voter alliance takes place over the next 45 days, the Democratic Alliance’s dream of securing 30% at the polls is fading…. [T]here is a widening gulf between DA leader Helen Zille’s ambitions and the party’s actual support…. However, last month the 30% dream was brought to a halt by Zille when she tweeted: ‘Just to make it very clear: our target is NOT 30% of the vote.’ The Ipsos survey found that the DA will probably have to be satisfied with a 23% showing at this year’s elections.”

Khusela Sangoni, speaking for the ANC, said that they are confident of receiving a big, decisive majority, but that they are not paying very much attention to the ebb and flow of survey data like that most recent Ipsos poll. One key political party tactic, of course, is to understate expectations so that a victorious party can claim to be astonished, delighted and – above all – appreciative of an electorate’s wise decision-making in reposing its trust in them.

Concurrently, and perhaps paradoxically according to the survey data, the Democratic Alliance appears to be growing stronger in the Eastern Cape, where new survey data showed its current level of support to be 26% of the electorate – nearly three times its 9.9% showing in 2009. The survey concluded this much larger number might well be a function of the virtual collapse of the Congress of the People as an electoral force. By the time votes are finally cast, Ipsos has projected the DA would ultimately achieve at least 15% of the province’s votes.

Interestingly, the Ipsos survey it reported on in January had found that the ANC’s support had dropped some ten percentage points in comparison to an equivalent period prior to the 2009 election – i.e., in November 2008. According to that Ipsos’ “Pulse of the People survey”, a random sample of South Africans of voting age or higher were asked which party they would support – if the election were to take place the next day. Respondents then filled in their own choices on a “ballot paper” to help create a tangible sense of a secret vote. In this series of surveys, overall support of the ANC had dropped off 10%. Put another way, the party had lost 19% of its overall support – from a level of 63% in November 2008 to only 53% in November last year (without the undecideds included).

Moreover, most of this drop came from the interval, November 2012 to November 2013. Ipsos argued that political uncertainty, leadership issues, the aftermath of Marikana, Nkandla, service delivery protests, the creation of several new political parties, among other reasons, could all have contributed to this decline in support. But, if its most recent polling is to be believed, the ANC has more than regained earlier lost support.

And this resurgence of support has come, despite the intensifying of many of the same issues Ipsos argued had been the likely causal factors for the ANC’s decline in support. In its January survey, 7% of those surveyed indicated they would not vote at all, 6% more declined to answer – and 5% still did not know which party they would ultimately select for their ballots. A further caveat is that this earlier survey, released in January, had actually been done just prior to the death of ex-president Mandela – an event that could well have influenced responses. At the time it was released, Ipsos said the next survey would reflect that event, as well as others.

Naturally enough, such swings in support levels bring into focus questions about the reliability and accuracy of such polling data – and how the raw data is to be interpreted. In very general terms, the accuracy of any poll is affected by its “coefficient of reliability” and its “margin of error”. In other words, one can say something like: a poll has a 95% (or 90% or whatever) chance of being right, but with a margin of error of 3 (or 2 or 4) percentage points either way, dependent on how many people were asked to respond and the vagaries of the statistical methods used to create the smaller sample that is meant to stand in for the entire population – rather than ask everyone in the country. In both cases, these figures are also affected by how questions are phrased, whether the sample surveyed is an accurate reflection of the larger population, and how “don’t know”, “not sure” and “undecided” are adjudicated, distributed between choices, or simply listed as uncertain as some way, among other issues.

To some degree, polling still remains something of an art, besides being a straightforward statistical process. The more success a polling organisation has over the years and the more often it does these surveys on the same or similar questions (as measured by the eventual voting), the more it can fine-tune its sampling methodologies, and the manner in which is asks its questions.

In standard polling practice abroad, it has been common to conduct telephone polls, drawing the numbers randomly from telephone listings (on the assumption that nearly everyone has a telephone line), and then adjusting and tweaking the numbers (and thus the people) included in the survey because of people who do not answer their phones, or who refuse to be interviewed by strangers about a fairly intimate set of ideas. In recent years, moreover, the growing dominance of cell phones, especially among younger people, and the consequent decline of landlines has made carrying out a good survey more complex – and a process that requires still more tweaking of its methodology and modelling.

In a country like South Africa, it is even less likely that landlines are universal and there is a strong reluctance by many to reveal their personal views to a total stranger by phone. As a result, good surveys require face-to-face interviews (presumably in the interviewee’s home language). Such methodology is something that significantly increases the cost and magnifies the difficulty of carrying out such surveys – and thereby makes it less likely they will be done frequently.

Ipsos has described its sample for this latest survey as being demographically representative as well as representative of both urban and rural voters. To achieve their result, they carried out 2,222 face-to-face interviews, and interviewees were disqualified if they were not registered voters. Ipsos actually structured its (somewhat varying) results based on three scenarios – heavy, moderate and light voter turnout and the three models distributed undecided, don’t know, not sure respondents over the various parties to recognize the fact that ultimately a voter makes a choice. An important footnote to the Ipsos polling was that if the final data discarded the 7.4% who declined to respond with their choices, under the moderate voting turnout scenario, the national level of support for the ANC came in at 61.5%, for the DA at 20.5% and for the EFF at 3.4%. This is significantly lower than the near two-thirds majority, but still substantially higher than those January figures.

In that earlier poll, the ANC had been accorded a 65% with a low voter turnout model, a 64% level in the moderate voter turnout level and a 56% level at the high voter turnout scenario (with undecided distributed). Data like this helps inform the implicit strategies for the ANC and DA. The ANC would seem to be trying to encourage highly motivated members and guaranteed supporters to show up on voting day, while the DA’s modus operandi would appear to be to work for as many new – and presumably disaffected of the current political order – voters to show up at the polling stations, come 7 May.

Meanwhile, this same weekend, the Afrikaans Sunday paper, Rapport, reported on the opinions of various political scientists trying to figure out what will ultimately play out in South Africa’s national election. In their article, University of Johannesburg political science lecturer, Piet Croucamp, argued that various factors will keep the national ANC total to under 59%, while Unisa Institute for African Renaissance Studies scholar, Shadrack Gutto, said Gauteng province was most likely out of reach for the ANC, given recent the net impact of the Nkandla report. Gutto was of the opinion the ANC could try to control the damage by withdrawing Zuma as presidential candidate, even at this late stage. That would serve the party better in appointing a popular, “squeaky-clean” candidate to lead them into the elections.

Croucamp added that many Numsa members, now that their union was effectively estranged from Cosatu, may well choose to vote for the Workers and Socialist Party, or simply stay home, rather than vote, as before, reliably for the ANC. Moreover, many Amcu members, particularly those who hail from the Eastern Cape but work on the mines up north may well vote for Bantu Holomisa’s United Democratic Movement, giving that party a significant boost over its 2009 total.

However, much speculation and would-be analysis by commentators still tends to overlook the impact of traditional voting patterns in the large rural areas of South Africa’s hinterland. In a system that has real echoes with machine politics as practiced for decades in big American cities, local traditional leaders let it be known who to vote for, who butters the local bread. And voters tend to oblige in large part to be remembered when the largesse is distributed.

But the longtime observer of local life and political behaviour, Jonny Steinberg, has reminded that for many in these communities, the things that really resonate with people are the indisputable facts that things have gotten better for them. There are more schools, housing is better, health care has improved and those social grants have made a huge difference in the lives of ordinary people. The concerns of city folk, like government maladministration over tenders and transparency, can be far away from daily life – and perhaps even the mess that is Nkandla. Add to that, yet another observation by a friend who pointed me into considering that all these attacks on Jacob Zuma may even be shifting sympathies his way – as in it being a case of those smart city folks picking on one of us – as measured by the positive shift in support for his party, as measured by the differences between the January and this most recent Ipsos poll.

Regardless of these shifts in support, however, over a month and a half left still remains in the campaign, until the actual voting on 7 May. Much may well ride on how the ANC responds to the revelations in the Nkandla report by the public prosecutor, or how adroitly the opposition parties rally behind a coherent and cogent message of change and reform – and, most importantly of all, of course, how the electorate ultimately responds to such calls. DM

Photo: South Africa’s President and leader of the ruling ANC party Jacob Zuma (C) greets his supporters as he arrives for the launch of his party’s election manifesto at Mbombela stadium in Nelspruit January 11, 2014. REUTERS/Ihsaan Haffejee.

Read more:

  • ANC set to secure two thirds: Poll at News 24;

  • Significant DA growth forecast in EC at the News 24;

  • DA’s dream of 30% wanes at the News 24;

  • ANC gaan by stembus boet vir Nkandla at the Rapport;

  • ANC’s support down to 53% among eligible voters – Ipsos;

  • South African general election, 2014 at Wikipedia;

  • ANC’s 45% poll panic in City Press.


While we have your attention...

An increasingly rare commodity, quality independent journalism costs money - though not nearly as much as its absence.

Every article, every day, is our contribution to Defending Truth in South Africa. If you would like to join us on this mission, you could do much worse than support Daily Maverick's quest by becoming a Maverick Insider.

Click here to become a Maverick Insider and get a closer look at the Truth.

Stimulus Package

Ramaphosa steps up to economic realities with R50bn package

By Greg Nicolson

"Look for lessons about haunting when there are thousands of ghosts; when entire societies become haunted by terrible deeds that are systematically occurring and are simultaneously denied by every public organ of governance and communication." ~ Avery Gordon