Panel Paper:
Changing Racial Segregation in the New South Africa
*Names in bold indicate Presenter
Method: In our analysis, we make use of two conventional measures of spatial segregation: the Index of Dissimilarity and the Entropy statistic. Both of these fall into the dimensional classification of measuring population unevenness. We apply these statistics to enumerated census data collected for the years 1996, 2001, 2006, and 2011. We use Quantec Corporation boundary-allocated data.
Results. For 2011 pairwise dissimilarity in Gauteng alone [Figures not visible in submitted abstract] we observe that almost all segregation levels are fairly high by conventional standards. When one considers that these are Ward data (population size larger and thus expecting more heterogeneity on that basis), the figures confirm the very sharply differentiated South African population. One also observes that Black/African-vs-White [DBlkWhi] segregation attains the highest value. Further examination points to particularly high levels in all pairings. In Johannesburg and Tshwane (examined individually) the segregation values are similar for most group pairs.
Our results for all 9 South African provinces point to appreciable differences by province in the level of segregation for these selected pairs and also highlight some sharp difference in trend by region. We find that that Black/African -vs-White [DBlkWhi] segregation (as indicated by the dissimilarity index) is quite high throughout. In all provinces upwards of three quarters of Whites (or, symmetrically, Black/Africans) would have to relocate across wards (within their province) to produce an even distribution, in which each ward had the same share of both the provincial Black/African and White populations. We also see that Black/African-vs-White spatial unevenness [DBlkWhi] is highest in the Eastern Cape; lower values are found in Free State and Northern Cape. Declines in Black/African-vs-White segregation during the 1996-2011 interval are modest (less than other pairings). Proportionate decline was 10% or less in all Provinces.
This submission reports preliminary research. Our extensions to this work will include using the Entropy Index to calculate a single comprehensive measure for residential segregation and change. We will repeat both dissimilarity and entropy analysis for specific urban agglomerations. We will also provide a more analytical interpretation on these results.