Patterns of voter participation in the 2019 elections

One of the key demands of the anti-apartheid struggle was that all South Africans should be able to choose their government. As we approach the 30th anniversary of the achievement of that goal, there has been much reflection on the fact that a large proportion of South Africans do not exercise their right to vote. According to the Gauteng City-Region Observatory’s 6th Quality of Life survey conducted in 2020/21, 28% of potential voters in Gauteng did not vote in the 2019 national and provincial government elections, including both those who were not registered (17%) and those who were registered but did not vote in the election (11% of potential voters). These numbers are based on recollections by respondents at least a year and a half after the election, and understate the extent to which potential voters did not vote (compare with Schulz-Herzenberg and Mattes 2023). Nevertheless the data does have the important advantage of being mappable to ward level within Gauteng.

Spatial variation

This map of the month (Figure 1) presents the Quality of Life survey data spatially as a percentage of potential voters in each ward who did not vote. In those wards with dark shading, more than 38% of potential voters in the ward did not vote in the 2019 elections. These include wards in downtown Johannesburg (H on the map) extending south to Turffontein, downtown Pretoria (B) and surrounding wards, some wards in townships such as parts of Soshanguve (A), Diepsloot (D), Tembisa (E), Daveyton (G), Braamfischerville (J), Soweto (K), Orange Farm (L), Sebokeng (M), and some rural wards. In those wards with light shading, voter participation was higher; less than 20% of potential voters in the ward did not vote in the 2019 elections. These include many wards dominated by more affluent households, such as Randburg (F) and Pretoria east. Large parts of the West Rand around Randfontein (I) and some rural wards also have high participation rates. A similar situation is also evident in the eastern parts of Tshwane. Importantly, townships are not uniformly low in participation – many wards in townships have lighter shading and therefore have high participation.

Figure 1: The percentage of potential voters in each ward who did not vote in the 2019 national and provincial elections. [Download a high resolution map here]


Variation by population group, gender, age, income and housing type

While there do seem to be some spatial patterns, the mapping of these results does not by itself explain the patterns we see. The analysis below slices up the data according to other characteristics: population group, age, gender, income and housing type. The dashed vertical reference line marks the average participation level in Gauteng. Where the green line is shorter, voter participation is below average and where it is longer, it is above average. This shows that lower proportions of black African and coloured respondents reported that they voted compared to Indian or white respondents. Voter participation is higher amongst older age cohorts. Men were less likely to have voted than women. Voter participation increases with income. Residents of informal accommodation were less likely to vote than residents of formal accommodation.

Figure 2: Disaggregating voter participation by race, gender, age, income and housing type.


Mapping component data 1: Potential voters who are not registered

The data used in Figure 1 was derived from two separate questions, which allows us to disaggregate the data further. The first question asks respondents whether they were registered to vote. 17% of potential voters said that they were not registered to vote. Figure 3 is a map specifically of the proportion of people in each ward who were not registered to vote.

Figure 3: The percentage of potential voters in each ward who were not registered to vote.


Mapping component data 2: Potential voters who were registered but did not vote

A follow up question in the Quality of Life survey asked only those who are registered to vote whether they did in fact vote in the 2019 elections. These results showed us that 11% of all potential voters were registered voters but did not vote in the 2019 elections. Figure 4 shows the proportion of each ward who were registered to vote but did not vote.

Figure 4: The percentage of potential voters in each ward who were registered but did not vote in the 2019 election.


Reasons given for not voting by those who were registered and did not vote

A further question asked only those who were registered but did not vote in the 2019 election (11% of potential voters in the province) why they did not vote. There is a mix of practical barriers and reasons that we might call voter disaffection (also see Mkhize, Naidoo, Götz, Seedat 2021). Practical reasons include: was not registered in current voting district (14%), respondent or family member sick during the election (6%), no ID (5%), respondent was working (4%), respondent was out of the province or country (2%). Reasons that indicate voter disaffection were more prominent and include: did not think his/her vote will make any difference (26%), do not like politics, broken politics, waste of time (16%), did not know who to vote for (10%), do not care (9%).

Figure 4: Reasons given for not voting by the 11% of the potential voters in the province who were registered but did not vote.


Interpreting voter participation levels

These results confirm other studies on voter participation. Political scientists have shown that support for the African National Congress has been declining over time (Braun 2022, Schulz-Herzenberg and Mattes 2023). The conventional expectation is that voters who are frustrated with a governing party switch their support to alternative parties who they feel will better meet their expectations. However when voters have doubts about the ability of alternative parties to meet their expectations, they become more inclined to not vote at all. Low voter turnout also reflects voter perceptions of the effectiveness of the state based on its performance. Mahangu and Schulz-Herzenberg (2022) conclude, based on survey analysis, that "years of state-based failures have eroded citizen perceptions of political system responsiveness which has, in turn, undermined how citizens perceive the utility of their vote’’. Braun (2022) shows that voters feel that they have no reason to vote if they have not received services and other development promises from the state.

More generally, the issue of voter disaffection has been observed in many parts of the world. One interpretation of this phenomenon connects voter participation to changing forms of party politics (Clarke 2023). In the mid 20th century, liberal democracies such as the US and UK had mass parties that engendered strong support from their constituencies. Parties organised a mass base through branches, who would participate in key decisions. By the later parts of the 20th century, mass parties had given way to what are dubbed ‘cartel parties’. Rather than organizing members, they turned to the media to mobilise support and they relied on state and other funding rather than membership funding. Furthermore, their attempts to appeal to many different constituencies means that they do not appear to be very distinct from their competition. Above all, voters are alienated by the way in which cartel parties are more interested in winning power for its own sake than social change.

Note on data

The wording of the questions used is below, and those who answered 'No' to each of these questions were combined into the composite map of Figure 1.

Q7.1 - Are you a registered voter?

Q7.2 - Did you vote in the 2019 National elections? It does not matter who you voted for, just whether or not you voted.?

The Quality of Life 6 dataset contains a sample of 13616 respondents, with a minimum of 20 respondents in each of Gauteng's 529 wards. The questionaire asks respondents which country they were born in, but not their current citizenship. We excluded all Gauteng residents born outside of South Africa from the analysis, recognizing the limitation of doing this – that some of these respondents have become citizens and would be able to vote.

We also excluded all 18 and 19 year olds from our analysis since they would have been below voting age at the time of the election and should not be considered potential voters for the purposes of this analysis. After excluding those born outside South Africa and 18 to 19 year olds, we had a sample of 11 890 potential voters.

The 2019 general elections were held on 8 May 2019. However this survey was run from September 2020-May 2021. As a result respondents were recalling their participation between 1.5 and 2 years after the election. This may account for the higher reported participation in this survey than in other data. Some respondents may also have said they were registered and or voted because they felt this was the right answer to give.


Braun, Michael (2022) ‘We vote for good things’: The moral economy of voting in South Africa, 2009-2019. PHD dissertation. Political Science, University of Toronto.

Clarke, Nick. (2023) “Parties: The Rise and Fall of Mass Party Politics.” In The Routledge Handbook of Social Change, edited by Richard Ballard and Clive Barnett, 207–18. London: Routledge.

Mahlangu, Thoko & Collette Schulz-Herzenberg (2022) The Influence of Political Efficacy on Voter Turnout in South Africa. Politikon. 49(2): 158-174

Schulz-Herzenberg, Collette and Robert Mattes (2023) South Africa’s ruling party is performing dismally, but a flawed opposition keeps it in power. The Conversation. 18 June.

Mkhize, Thembani, Laven Naidoo, Graeme Götz and Rashid Seedat (2021) Voting patterns in the 2021 local government elections. GCRO Map of the Month. December.

Inputs and edits:

Graeme Götz, Christian Hamann, Gillian Maree


The GCRO sends out regular news to update subscribers on our research and events.