inside politics

Here I explain what a turnout track is and how it works, with a view to producing a range of these graphs for the Resources Page. Their most powerful insight is comparative, allowing us to see a turnout differential between two parties.

Developing a turnout track, using Johannesburg

By: Gareth van Onselen
Follow @GvanOnselen
7 February 2026

This essay is the 14th in an on-going series on Election 2026/7, for all other editions of this series, please click here: Election 2026/7

Introduction

I am going to be adding a new set of indicators to the Resources Page: turnout tracks. I will do these nationally, for each province and metro, and I am going to use this page to explain how they work.

We will use Johannesburg as an illustration, and the ANC and DA as the parties we will focus on. In 2021, the ANC got 33.22% and the DA 25.45% on the PR ballot, in the metro. We can break those performances down into performance percentage points and wards in more systematic fashion.

In other words, using the DA, we could plot how many wards the DA got 1% or more in, how many wards it got 2% or more in, 3% or more in, 4% or more in, and so on. Up to 100%. Naturally, the number of wards for each percentage point will drop to zero as you approach 100%, as no party has ever achieved 100% in a ward in JHB.

Once we have all the relevant groups of wards for each performance percentage point, we can work out the DA’s turnout (the number of valid DA votes in that group, divided by the total registered population of that group), and put a dot on a graph.

If we run turnout percentage (0-100) on the vertical axis, and performance percentage points (also 0-100) on the horizontal axis, when we have a dot for every performance percentage point we can join them up and see the turnout track – a line that shows us the relationship between how many DA voters voted by how well the DA did.

Here is what it looks like for JHB in 2021:

So, what can we tell from that? Well, for one, we can immediately see that the DA’s highest performing ward in JHB in 2021 delivered 72.57% (ward 72, as it so happens) – the point where the turnout track line stops. As the DA did not secure any vote share higher than that in any ward, there can be no more DA ward groups beyond that dot, and the line stops there.

I haven’t included labels for every dot or the graph would be become too cluttered; regardless, they don’t really matter, it is trends we are interested in. So I won’t be using them on the Resource Page graphs. That said, there are a few labels here just to illustrate that DA turnout increases the better it does. Which makes sense.

There is, however, important context missing. The number of wards that correlate with each performance percentage point. They are hidden. But we can make them appear, we just need to use a secondary axis on the right (1-135, the number of JHB wards in 2021). Here is what that looks like:

Now we have both primary indicators, and it becomes far easier to understand what we are looking at. DA turnout grows the better it does; but, as it does better, so the number of wards in which it performs well drops. That final dot (72.57%) represents just one ward. But 0% represents all 135 wards (which makes sense, the DA did better than 0% in every single JHB ward).

There is one final core indicator we need though, to make complete sense of this graph: the average turnout, against which we can compare the DA’s turnout level. This is calculated by taking the same ward groupings for each performance percentage point, and working out turnout for all parties in that group, not just the DA (total valid votes divided by total registered population).

This allows us to see what share of total turnout DA voters comprised, for each performance percentage point. Here is what that looks like:

Here we can see that as the DA performs better, so its voters increasingly dominate the average turnout share – the gap between the two lines being big at 0% and much smaller at 72.57%.

You might argue this all makes good sense but is not generally that helpful. This stuff is fairly obvious. You would be right. The true value of these graphs is comparative. Let us go back to one line – the DA’s turnout track – but include the ANC’s turnout track on the same graph:

Now we have something very interesting. What you are looking at is a turnout differential, and you can immediately see why the DA does so well in local government elections. From around 3% onwards, it dominates the ANC and as its support grows, the gap between the two widens. Not only that, but its line runs deeper than the ANC. The ANC’s track stops at 66.4%. The DA’s track runs to 72.25%.

Let’s add the ward numbers for each party:

Now we can see that ANC turnout is flat, but broad: it might have a lower differential turnout, but it makes up for it in absolute numbers, as each performance percentage dot generally comprises a far greater number of wards than the DA. And that, despite the DA’s performance, is how the ANC ended on 33.2% and the DA on 25.5%.

Perhaps the most critical insight on this graph, from a DA perspective, is the gap between the two lines. Let’s call it an “advantage window”. It means that there are a (admittedly small) number of wards (5 in the case of 2021), where the DA’s turnout advantage over the ANC is absolute. That is the sign of a more motivated party (because these dots are at the end of the track they comprise the wards with the most DA support – the party’s strongholds).

2021 has a broader context, however. Overall, turnout fell through the floor (it was 41.6% on the PR ballot, compared to 56% in 2016). The previous LGE, 2016, was by contrast the highest turnout has ever been in a JHB LGE, so it makes for a good point of comparison for the same set of indicators. Here is what those look like:

Some profound differences between those two graphs. Higher turnout across the board in 2016, meant both parties’ tracks started higher up (ANC at 25% turnout, compared to 14%; DA at 21% compared to 10%), and thus both lines run higher. The DA even breaches the 50% line. Together with the ward count, the ANC is again flat but broad, but the DA’s line in 2016 doesn’t drop anywhere near as steeply as it does in 2021, meaning many more wards are covered per performance percentage point.

Perhaps the biggest difference is the advantage window. The DA runs longer than the ANC again (the first time it had ever done that), as well as longer than its own performance in 2021, but between those two lines are 17 wards, compared to five in 2021. That is a huge number, enough to drive the DA up to an all-time high of 38.5%.

There is much more that can be done with these types of graphs.

Here are the big four parties in JHB in 2021:

Here are the DA’s turnout tracks for each JHB LGE since 2000:

Each of them have something to offer. I will try fit as much as I can into the coming graphs. I will start uploading the metros this week and work on finishing everything from there over the coming month. When they come, I hope the above is enough to make sense of them.

This essay is the 14th in an on-going series on Election 2026/7, for all other editions of this series, please click here: Election 2026/7