Development of census output areas in South Africa.
Loading...
Date
2015
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The use of the same geographical unit for both census data collection and dissemination is
common in many countries across the world, especially in developing countries. This poses
some serious concerns. Firstly, this practice has caused various difficulties for the census data
users as the ideal characteristics of an area to facilitate efficient census data collection differ
considerably from those which aid analysis and interpretation of the published data. Secondly,
some Enumeration Area (EA) populations fell below the census confidentiality limits,
requiring the data to be combined with those of a nearby EA. Thirdly, the design of EAs
before census data collection does not take into account local social divisions in boundary
placement. Lastly, the shape compactness of areas is often ignored. In order to address these
four concerns, the advanced techniques of automated zone design methods, such as
Automated Zone-design Tool (AZTool), are required for the development of suitable output
areas in South Africa that would address the four concerns as much as possible. Therefore, the
overall aim of this study was to develop optimized census output areas using AZTool program
in South Africa. In order to achieve this aim, among others, the following research objectives
had been developed; firstly, the creation of output areas using AZTool program with the 2001
census EAs as building blocks in South Africa. Subsequently, the determination of the
statistical qualities of the AZTool generated output areas with regard to population target
mean, minimum population threshold, social homogeneity and shape compactness was
explored. In addition, the comparison of the newly created output areas with existing census
small areas was also considered. The study area comprised of two of the nine provinces of
South Africa. These included the Free State (representing rural settings) and Gauteng
(representing urban areas). This study employed EAs from the 2001 census estimates (HSRC,
2005) as building blocks for creating new census output areas in South Africa. The 2001
census SubPlaces, the 2001 census Small Area Layers (SALs) and 2011 census SALs data
were also explored for further evaluation the AZTool program. In order to validate results
from the AZTool program, some analyses such Analysis of Variance (ANOVA), Shapirowilk
test, paired t-test, and Kolmogorov sminov test were performed using Statistical Package
for Social Sciences (SPSS). Results showed that the primary criterion of minimum population
threshold of 500 people (which is the official minimum population threshold used by
Statistics South Africa) was kept and not breached throughout all the AZTool newly created
output areas at different geographical levels as well as in both rural and urban areas.
Furthermore, the Intra-Area Correlation (IAC) of 0.62 for the two provinces (Free State and
Gauteng) combined indicated that the selected homogeneity variables (geotype and dwelling
type) were good indicators of social homogeneity for creating optimised output areas in South
Africa. It was also found that the newly AZTool generated census output areas out-performed
the existing official SALs and SubPlaces, non-zone design developed geographies. This was
proven by the fact that AZTool output areas effectively satisfied minimum and target
population thresholds, while the population distributions were much narrower in range than
those of the existing SALs and SubPlaces. However, the AZTool created output areas were
less compact in shape than the SALs and SubPlaces in all geographical regions. In general,
there was statistically significant (p < 0.05) difference in Perimeter Squared per Area (P2A)
means between the output areas and the SALs. The LSD post-hoc test revealed that difference
between the P2A means for the AZTool output areas and the SALs was not statistically
significant (p > 0.05). Therefore, it was concluded that there is potential in application of
automated zone design methods, particularly AZTool program, in the creation of optimized
census output areas in South Africa. It was also concluded that findings from this study
contribute to the research in general and to the potential applications of automated zone
design methods in developing countries. One of the main recommendations is that further
research and general work should evaluate the application of automated zone design methods,
such as AZTool computer program, in the creation of census output areas across the entire
country. In addition, data should be made accessible at lower geographical level such as EA
or household levels even if it is under secure conditions to allow robust developments of
optimized census output areas using automated zone design techniques.
Description
Doctor of Philosophy in Environmental science. University of KwaZulu-Natal, Durban 2015.
Keywords
Census districts--South Africa., Population--Statistics., South Africa--Population., Theses--Environmental science., Census output areas.