Voting patterns in metro areas - ANC is well ahead

TNS Research Surveys, South Africa's leading marketing and social insights company, conducted a survey between 23 October and 9 November 2008 to examine what parties people say they would vote for in the next national elections. This survey was conducted amongst a sample of 2 000 SA adults from the seven major metropolitan areas of South Africa, interviewing them face-to-face in their homes, with a margin of error of under 2.5%.  This study reflects only metro areas but is fully representative of all people in metro areas.

In South African politics at the moment, it is important always to note the fieldwork dates and link these with events at the time.  A key conference was held in Sandton on 2 November - the "South African National Convention", out of which was eventually born the Congress of the People (COPE).  Accordingly, it is important to examine the results of this survey before 2 November and after 2 November separately, also ensuring that these two sub-samples are weighted to the same profiles.  These results are also compared with a similar study conducted in early September 2008.

ANC stays in control in metro areas but uncertainty rises, as well as those refusing to answer

People were asked to say for which party they would be most likely to vote in the next national election.

In the sub-sample polled prior to 2 November, the ANC was given by 41% (compared with 42% in September).  However, post 2 November, this dropped to just 30%.  The new party was given by 3% in both the latest sub-samples (note that, since the name of the party was uncertain at the time of fieldwork, it was listed as "The new party being formed by former defense minster Mosiuoa "Terror" Lekota and former Gauteng premier Mbhazima Shilowa").

So where does the ANC's 11/12 points go?

The principal effect is to increase uncertainty.  The proportion of people giving a "don't know" response rises from 16% to 19% (+3) (14% in September - so uncertainty has increased by five points since September, a just statistically significant rise).  In addition, the proportion of people who refused to divulge who they will vote for rose from 10% to 14% (+4) (September 12%) whilst those who say they will not vote changed very little (14% to 15%) (15% in September).  This suggests that there may have been a slightly greater sensitivity to divulging one's voting preference in the week after the SADC/COPE conference.  In addition, the DA moved from 11% to 13% in metro areas (12% in September), but this is not a statistically significant change.

As in September, no other party achieved more than 1% of the metro vote, making COPE, even in the first week in November, the third largest grouping.  Given that some of the additional 4% of people who refused to divulge their preference may well be COPE supporters, it is likely that the COPE figure of 3% in metro areas is a slight underestimate. 

How do different demographics fall?

Regretfully, the biggest differences in claimed voting patterns are still between race groups.

For blacks, the ANC moved from 62% in September to 60% in the sub-sample prior to 2 November and then to 46% post 2 November.  COPE achieved a 4% mention in both the latest subsamples, the ANC drop going collectively to the three categories of "I don't intend to vote", "don't know" and "I refuse to answer":

• "I don't intend to vote" rose from 11% to 16% (11% in September).
• "Don't know" rose from 12% to 15% (12% in September).
• "I refuse to answer" rose from 8% to 13% (11% in September).

For whites, the DA moved from 37% in September to 26% in the sub-sample prior to 2 November and then to 49% in the post 2 November sub-sample.  COPE received a 2% mention.  However, again, large movements in the three "opt-out" categories are evident:

• "I don't intend to vote" dropped from 21% prior to 2 November to 13% post 2 November (17% in September).
• "Don't know" dropped from 26% to 16% (18% in September).
• "I refuse to answer" remained at 18% (17% in September).

This suggests that indifference in the election rose somewhat for whites between September and November (52% in the opt-out categories in September, rising to 65% prior to 2 November) but that this dropped rapidly after the Convention on 2 November (47%), the DA being the beneficiary. 

For the other two race groups, the samples are too small to analyse pre and post 2 November with adequate precision.  However, looked at together, the COPE response rises from zero to 4%, the ANC shows no significant change (averaging 14%) and the DA loses some 12 points (from 26% to 14%) whilst the "don't know" response rises significantly from 22% to 32%.

Results by province

Region is another key differentiator (although, again, bear in mind that these results are only for metro areas).  The data below is for the full sample as the sub-samples pre and post 2 November are too small to analyse by area (September figures are in brackets).

Although the pre-post 2 November shifts cannot be shown it IS evident that the new party does have some legs in all areas except, perhaps, Durban.  The ANC showed some decline in Cape Town.

Service delivery expectations?

Support for the ANC Is high amongst those in squatter shacks (62%), matchbox houses (62%) and those who are unemployed (53%).  This suggests that these people feel that the ANC is still their best chance of a better life.  However, this puts considerable pressure on a new government on service delivery.

Other demographic differences of note in metro areas

Differences in potential voting patterns in metro areas show no notable difference by age - COPE had interest amongst all age groups at the time of the study.  In terms of black language, the new party has the lowest support amongst Zulus (2%), but this rises to 7% amongst all other black languages.

How do people feel Jacob Zuma will perform as a possible President of SA?

People were asked if they thought that Jacob Zuma would do a good job as President of South Africa.  Over the full sample, 36% said "yes", 40% said "no" and 25% gave a "don't know" response.  However, again, we need to analyse responses pre and post 2 November:

• Those answering "yes"
  • Pre 2 November - 37%, post 2 November - 26%
  • Blacks: pre 52%, post 37%
  • Whites: pre 5%, post 2%
  • Coloureds and Indians: pre 17%, post 10%
• Those answering "no"
  • Pre 2 November - 39%, post 2 November - 45%
  • Blacks: pre 25%, post 29%
  • Whites: pre 65%, post 76%
  • Coloureds and Indians: pre 62%, post 67%
• Those answering "don't know"
  • Pre 2 November - 24%, post 2 November - 30%
  • Blacks: pre 23%, post 34%
  • Whites: pre 31%, post 22%
  • Coloureds and Indians: pre 22%, post 23%

It is notable that the black "yes" figure declined, the response moving mostly into the "don't know" category".  The white results moved from the "don't know" category to the "no" category.

Our take-out

The advent of the breakaway grouping known currently as COPE and all the publicity surrounding this grouping, much of it generated by the ANC's own reaction to the party, has led to some decline in the certainty in many people's minds (especially blacks) as to who they might vote for.  It seems that there is a fairly substantial body of ANC supporters in metro areas who are now less certain that they will vote for the ANC.  Stated support for COPE is 3% but, given that refusal levels rose after 2 November, this may be a slight under-estimate.

Amongst whites, the advent of COPE seems to have stimulated greater interest in the election with the DA being the major beneficiary of this interest.  It may be that there is a feeling that the opposition could be strengthened by the new party.

That the perceived possible ability of Jacob Zuma to be an effective president also shows a decline is also noteworthy.  Again, blacks moved into the uncertain category whilst whites moved more into a negative arena.

It is clear that the breakaway party has created a greater interest in the election in some quarters but has crated greater uncertainty in the minds of some ANC supporters.

Technical note

The studies were conducted amongst a sample of 2 000 adults (1260 blacks, 385 whites, 240 coloureds and 115 Indians/Asians) in the seven major metropolitan areas: they have a margin of error of under 2.5% for the results found for the total sample.  The studies were conducted by TNS Research Surveys (Pty) Ltd as part of their ongoing research into current social and political issues and were funded by TNS Research Surveys.

Source: www.tnsresearchsurveys.co.za