Dot-maps of racial distribution in South African cities

Inspired by Bill Rankin's "Chicago Boundaries", and having finally obtained a copy of the small-area data from Census 2011, I decided to draw some similar dot-maps showing how the population is distributed in South African cities. The primary lesson from these is that the legacy of apartheid is still very clearly visible. I suppose that was to be expected.

UPDATE: click here for my new post with comparable maps from the 2001 census.

UPDATE 2: click here for a zoomable, scrollable map of the whole country.

In each map, one dot represents 50 people. Here's Cape Town (with all of these, click for a full-size version):

 Johannesburg:

 Durban:

Durban

I've also made maps for Bloemfontein, East London, Port Elizabeth and Pretoria.

Comments

That is so awesome!

Adrian, this is fantastic work. I'm doing a lot of work with political maps (ward boundaries and eventually voting districts) and there is a clear correlation between 'yellow' (ANC) or 'blue' (DA) wards and your work. I look forward to more.

Paul, you should get Adrian to co-host a masterclass or demo session at Hacks/Hackers in both Jozi & Cape Town.

Adrian, on behalf of Future Cape Town (futurecapetown.com) may I just say that this is outstanding work. Politically and analytically meaningful, graphically clear, nice and clean. Well done. - Brett

Hi Adrian,
These are really nice maps, I'll be sending my students to look at them.
A quick question though as you may be able to help us. We haven't been able to access the small area statistics (I have been trying to get hold of the data for Makana Municipality to map Grahamstown's racial distributions). Where in the Statssa website did you locate them?

Roddy Fox (R.Fox@ru.ac.za)

Tel: (012) 310 8600
Fax: (012) 310 8944
info@statssa.gov.za

One of our local offices can deliver the CD's to you.

Kind regards
Ashwell

You can order a product of 3 DVDs from Stats SA with all the census variables for all the levels of geography. Contact helenev@statssa.gov.za to assist

Nice idea Adrian and interesting stuff. How freely accessible is the raw data? I'm just wondering if it's qorth expanding on this project: getting earlier census data and creating time comparative stuff, or just doing some analysis on it for clustering and densities etc. Perhaps most of this is done already and available somewhere? Thank, cheers.

The raw data is available from Stats SA. For previous censuses as far as I know only 2001 has data at a similar level of detail.

Alex,

We (www.gripcompany.co.za) have integrated an array of different datasets into our location analytic offerings. Our geo-demographic data is updated on an annual basis (through primary research and statistical adaptations), which allows our data to visualise demographic trends over time, for example. If you would like more information on what we are busy with, drop me a mail - michael@gripcompany.co.za.

I am surprised, but encouraged by how integrated middle-class neighbourhoods appear in these maps. Thanks for doing this.

great charts, well done

"The primary lesson from these is that the legacy of apartheid is still very clearly visible." That sounds like knee-jerk PC rhetoric. I think the astonishing thing is how many blue dots there are amongst the pink after only two decades of progressive economic engineering.

Fascinating material. Thank you!

Also interesting to note the concentration of the colours of the different groups.. no surprise either though

Hi from reddit,

Could you have a bigger picture of the Gauteng area? I am specifically interested in the East Rand.

Good job and thanks!

Great stuff.

Would be very interesting to overlay that with maps showing life expectancy, number of children per adult, number of schools per child, number of health care facilities, doctors and nurses per capita, prevalence of TB, HIV, per capita income, electricity, water, sewage, gas, service availablity etc. etc. etc.

You've opened a can of worms. Have you tried d3.js?

Regards,
Andrew

First, I think credit must go to Stats SA for the great work they are doing in providing this data to us so that we can produce these wonderful maps. What is of interest to me is the greater extent of integration in the former "white" suburbs while the townships remain fairly homogeneous from a race perspective. It is also interesting to see how the CBD areas have become dominated by black/African people. What will be interesting is to do a comparison with the 1996 census to see if and how these patterns have changed. I will be creating a lifestyle segmentation of the 2011 census and will compare it to a lifestyle segmentation that the HSRC produced for the 1996 census to see how the demography of our cities and rural areas have changed. PS Stats SA - how about a socio-economic atlas of the 2011 census?

Thanks for this. The next step would be for someone to make an interactive map using assorted API's and adding other statistics.

That's one of the things in my future plans.

Great work. I have already integrated this into a course I am doing on mapping at WITS. I am also surprised at the level of integration shown in some of the maps (especially in Johannesburg) and also how this appears to be linked to density.

Hi, great maps. Possible to overlay suburb names or highlight the highways etc for easy navigation?

Adrian,

Great visualisation Adrian. We (www.gripcompany.co.za) are currently working on an array of different location analytical solutions for businesses in South Africa. Our SpatialKey offering (www.spatialkey.com) has been launched this month, with full internal database integration capabilities.

I am currently busy developing solutions for the retail and real estate sectors in South Africa. If you would like more information on what we are doing, drop me a mail - michael@gripcompany.co.za.

The combination of race and density makes for very powerful images. Well done. If you included the whole municipal area you would reveal the particularly stark situation facing isolated townships like Botshabelo, Soshanguwe, Atlantis, Orange Farm and Mdantsane, all of which are currently excluded by your narrower boundary. The single most important additional variable to add would be whether or not the adults are employed. This would demonstrate the problems of concentrated poverty in our isolated townships and informal settlements. It would also be useful to add a few key place names for ease of orientation.

More people are moving to the cities and they are capable of maintaining the City lifestyle. I would be interested in seeing the life expectancy of a rural person in 1996 compared to 2011.

"The primary lesson from these is that the legacy of apartheid is still very clearly visible. I suppose that was to be expected"

FANTASTIC map - how ever, I doubt the legitimacy of "the dot" and that it represents 50 people. In some instances yes, I think it is correctly so - such as Alex & hillbrow (JHB) etc, but when I look at the area that I live in, there are many "incorrect" dots I see. Such as that there are definitely not 50 black people living where the dot has been placed (i know this because i live in a small boomed off community where everyone "knows" everyone). Which brings me to the question, what is the census based on? Is it people that reside there (including domestics / temporary staff) or people that work there or people that are just moving through?

This is great - excellent to do this work. Take a look at a new tool we're developing: http://www.housingfinanceafrica.org/blog/presenting-the-new-citymark-das... Would love to talk more about what you're doing and see if there's a potential for collaboration?

Even without apartheid, it seems like people with similar cultures and ethnicity tend to group together anyway, but this graphic shows it to be pretty extreme here. Here in the states there are regions of cities that tend to be dominated by one particular ethnicity as well. South Los Angeles tends to be black, but central Orange County tends to be Hispanic and Vietnamese.

- RM Dude

Add new comment

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
By submitting this form, you accept the Mollom privacy policy.