DOCUMENTS

ANC likely to hold its ground in election - Ipsos

Ruling party polls at 61% of vote on medium turnout scenario, also likely to hold Gauteng

Final Election Outlook 7 May 2019

With service delivery in the spotlight over the last few months, the most important post-1994 election in South Africa’s history will take place on Wednesday 8 May 2019 and, in spite of the electorate’s unhappiness about service delivery and other issues, it is unlikely that the low turnout figure and the phenomenon of ANC “stay-away voters” of the local government elections of August 2016 will be repeated.

7 May 2019

The past weekend saw the three biggest parties on their final rallies in Gauteng. Leaders made their final appeals to voters to come out and cast their votes in their favour.

Against this background it is quite sobering to remember that only a third (33%) of registered South African voters believe that the country is currently going in the right direction. More than half (52%) are saying the country is going in the wrong direction.

The picture is even more interesting if we look at the opinions on this question held by supporters of the three biggest political parties in the country.  ANC supporters are split in their opinions, while six in every ten supporters of both the DA and the EFF feel that the country is going in the wrong direction.

Party

The country is going in the right direction

%

The country is going in the wrong direction

%

ANC

40

42

DA

26

61

EFF

25

61

(Others said that they did not know or were uncertain, that is the reason why figures do not add to 100%)

These are some of the findings of the latest Ipsos Pulse of the People™ study conducted in March and April 2019. (See “Technical Detail” at the end of this press release.)

Modelling of voter turnout scenarios and possible party support on 8 May

The modelling of possible voter turnout and possible votes is a complicated analytical process.  Not only are there different turnout scenarios on a national level, but voters in different provinces have different turnout behaviours; different population groups and different age groups also behave differently – and moreover, voters are usually more convinced that they will vote before an election than on election day. There are also split votes – where some voters vote for a different party on the national and on the provincial ballot.

If previous elections are anything to go by, it is unrealistic to expect that ALL registered voters will turn out to vote on election day.  This is a major factor as the actual voter turnout can have a profound influence on the number of votes for each political party.  To try and get a workable answer to the question of possible turnout, we use the results of two survey questions, focusing on the desire to vote and on the likelihood to vote. 

How much do you want to vote in the next national and provincial elections this year?

 Desire to vote…

All registered voters %

ANC supporters

%

DA supporters

%

EFF supporters

%

I definitely want to vote

35

39

37

32

I want to vote

50

53

52

53

I don’t want to vote

13

7

8

14

Don’t know

2

1

3

1

How likely are you to vote in the 2019 National and Provincial elections?

 Likelihood to vote…

All registered voters %

ANC supporters

%

DA supporters

%

EFF supporters

%

Very likely

42

48

45

39

Likely

45

46

45

48

Not likely

11

5

9

11

Don’t know

2

1

1

2

More than 80% of voters were of the opinion that they would turn out to vote on election day. However, this is probably not realistic, given past turnout patterns.  Thus, based on the answers to these two questions and other questions in the study, an algorithm is developed to produce low, medium and high voter turnout scenarios.

From these numbers, a low voter turnout scenario will result if about 60% of registered voters turn out to vote; medium voter turnout scenarios will result if about 71% or about 75% of registered voters turning out to vote and a high voter turnout scenario will result in turnout of about 80% of registered voters.

Likewise, a view of what the possible outcome of the elections could be, is not as simple as asking respondents which party they are going to vote for.  As can be seen in the table overleaf, in Ipsos’ last pre-election survey, 5.57% of registered voters refused to fill out their “ballot papers” that we use in our interviewing process, while 4.19% indicated that they did not know which party they would vote for.

Obviously, we needed to get a clearer understanding of the views on political parties of these groups of (possible) voters.  Thus, their answers to a variety of other survey questions were used to determine their individual party choices:

The political choices determined by a simulated “secret ballot”

Direct questions about party choice

The party the respondent will never vote for

Views on political parties, ranging from “I reject this organisation completely and on principle” to “I will definitely vote for this organisation if there were an election tomorrow”

Assessment of political leaders

Other opinions expressed on political parties during the survey.

The figures regarding possible party choice (below) result from this rigorous process. 

With the choice of 48 parties on the national ballot paper, it is not possible to give accurate possible support figures for each of the smaller parties. 

In a low voter turnout scenario, the ANC voters seem to be more committed to vote, and therefore the party support increases notably.  In a high voter turnout scenario, the ANC succeeds in convincing voters who would otherwise not have voted to come out and participate.  Thus, both a low turnout and high turnout will be to the singular advantage of the ANC.  The support for other parties is far more consistent across the four scenarios.

Summary of party support on different turnout scenarios:

Chosen Party

All registered voters

(Ballot Paper)

%

Low turnout

Scenario

(60% turnout)

%

Medium turnout scenario 1 (71% turnout)

%

Medium turnout scenario 2 (75% turnout)

%

High turnout scenario

(80% turnout)

%

ANC

56.92

64

61

62

65

DA

15.23

17

19

19

18

EFF

9.45

10

11

11

11

IFP

2.16

3

3

3

3

FF+

0.76

1

1

1

1

ALL OTHERS

3.02

5

5

4

2

Will not vote

2.7

 

 

 

 

Refuse to answer

5.57

 

 

 

 

Do not know

4.19

 

 

 

 

A few national headlines

The ANC will still be the ruling party after 8 May and the DA will be the official opposition in the country.

The ANC draws support from all over the country, but especially from Gauteng and KwaZulu-Natal.

DA support is concentrated in the Western Cape and Gauteng.

The EFF support is strongest in Gauteng, the Free State and Limpopo.

And the provincial votes?

It is unlikely that the DA will win the 3 provinces they set out to do (namely the Western Cape, Northern Cape and Gauteng), but the Western Cape is within reach (although they might need a coalition partner to form a government).

The EFF should be the official opposition in Limpopo and North West, and possibly in the Free State and/or Mpumalanga.

The ANC is likely to achieve comfortable majorities (of 66%+ of the vote) in Mpumalanga, North West and Limpopo.

The ruling party should also achieve a majority of the votes in KwaZulu-Natal, the Eastern Cape and the heavily contested Gauteng.

The possible outcomes of the provincial vote in the Free State and the Northern Cape is not as clear.

Some of the smaller political parties have a definite support base in particular provinces –

like the IFP who is likely to win at least one in every ten votes in KZN;

the FF+ gets support from Mpumalanga, Gauteng and the Western Cape;

the strongest support for the UDM comes out of the Eastern Cape, where PAC support also originates from;

support for the ACDP comes mainly from the Western Cape;

AZAPO gets most of its support from KZN and

the GOOD party makes an impression in the Western Cape.

It is very important to keep in mind that survey research is not an exact science and that all results have to be evaluated within the margin of error, determined by sample size, response rate and sampling methodology used.  In addition, survey research is only considered a true reflection of the situation on the ground at the time of the survey.  This also must be kept in mind when evaluating results (for more information, see the Technical Detail paragraph.) 

TECHNICAL DETAIL

Fieldwork: 22 March – 17 April 2019

3600 in-home F2F interviews. Conducted in home languages of randomly selected respondents.

Countrywide representation

Filtered results by 18 and older AND registered to vote.

Results weighted and projected to official published IEC registration figures

Margin of error for this subsample (0.9% - 2%) – based on sample size, response rate and sampling methodology.

Issued by Ipsos, 7 May 2019