Exa Examining ining Boun oundari daries of Elec of Electoral - - PDF document

Exa Examining ining Boun

• undari

daries of Elec

• f Electoral

toral Wards in ards in Kwa Kwa-Zul ulu Nat u Natal: al: Delim elimit itati ation

• n Re

Requirem uiremen ents and Sp and Spati atial Com al Complianc liance Ntobeko Masondo _______________________________________________________________________________ Abstrac tract Spatial dimensions of electoral wards should match population growth and its spatial

• distribution. A key requirement of a ward delimitation process is that it should create wards that

contain approximately equal numbers of people, areas that are contiguous and geographically

• compact. Currently, the number and size of South African wards is solely based on the voters
• roll. In order to examine electoral ward boundaries against these requirements spatial analyses

were done in the province of KwaZulu-Natal (KZN) to, firstly assess the compliance of electoral wards in terms of voter parity, population equity and eligible voter equity; and secondly to measure the degree to which electoral wards fragment communities. The main finding is that ward boundaries of KZN are not fully compliant with the ward delimitation requirements.

• 1. Introduct

Introduction ion The three critical characteristics of electoral wards are population equity, contiguity and geographic compactness.1 In other words, electoral wards should contain approximately equal numbers of people, areas that are contiguous and geographically compact. Compliance with these three critical characteristics creates wards that facilitate fair representation of citizens and it also enhances participatory democracy.2 The South African ward delimitation criteria, laid out in the Local Government: Municipal Structures Act of 1998, prescribes ward delimitation which creates wards that have the three critical characteristics. Schedule 4, section 4(a) and (b) of the Municipal Structures Act stipulate that wards must contain approximately equal numbers of registered voters. Moreover, wards must not fragment communities.3 A study was undertaken to measure the degree to which electoral wards of KwaZulu-Natal are adhering to the delimitation requirements as laid out in the Municipal Structures Act. This paper is a concise report of that study. The hypothesis of this paper is that the geographic structure and composition of electoral wards in KwaZulu-Natal does not reflect the ideal process or factors that need to be considered, as prescribed in the ward delimitation criteria. The study presented herein will shed some light on

2

the process for determining boundaries of electoral wards in KwaZulu-Natal. The province of KZN was chosen as a case study because KZN has the most number of wards in the country (828 wards) followed by Eastern Cape with 715 wards. Moreover, KZN consists of diverse settlements, as all settlement types in the country are represented in the province (National Urban Development Framework Steering Committee 2009). This paper is based on two interrelated research activities. The first activity was an extensive literature study aimed at identifying the theoretical, legal and technical aspects pertaining to the delimitation of electoral ward

• boundaries. The second research activity was a technical exercise aimed at assessing ward

boundaries of KZN against the legal requirements. The main aim of the study was to examine the ward delimitation process, to assess KZN wards against the delimitation criteria, and to investigate the degree of fragmentation of human settlements caused by wards. To achieve this aim the first objective of the study was to critically discuss the South African ward delimitation process against the backdrop of international norms and standards. Secondly, technical methods and procedures for determining electoral boundaries were explored as part of the literature study. Thirdly, Geographic Information Systems (GIS) and statistical techniques were applied to test whether KZN wards consist of approximately equal numbers of people, registered and eligible voters. This was done through three GIS-based

• scenarios. Scenario 1 is an assessment of wards for compliance with voter parity. Scenarios 2

and 3 are assessments of wards for compliance with population equity and eligible voter equity

• respectively. Lastly, GIS techniques were applied to quantify the degree of the fragmentation of

human settlements caused by electoral ward boundaries. The theoretical, legal and technical requirements for electoral ward delimitation are contained in the literature study. Details of the South African ward delimitation process are discussed under section two. Section three contains methods employed to achieve the objectives of the study. Data used in the study as well as the relevant spatial and statistical analyses are discussed. The first component of the spatial analysis focused on measuring whether KZN wards contain approximately equal numbers of registered voters, as stipulated in Schedule 4, section 4(a) of the Municipal Structures Act. Further exploratory analyses were conducted to measure whether wards contain approximately equal populations and numbers of eligible voters. A correlation and regression analysis was performed to complement the findings of the first component of the

• analysis. The second component of the spatial analysis focused on measuring the degree to

which electoral wards comply with the second requirement of the ward delimitation criteria. This

3

requirement is contained under Schedule 4, section 4(b) of the Municipal Structures Act of 1998, and it stipulates that ward boundaries must not fragment communities. Non-fragmented communities are an indication that wards comprise contiguous and geographically compact

• areas. Section four contains results of the spatial analyses, and a discussion of such results.

Policy implications of the findings, the limitations of the study and recommendations for further research are summarised in the conclusion.

• 2. Lit

. Literature Study erature Study An electoral ward is the basis for political representation for citizens in a democracy.4 Elected representatives are accountable to people in a spatially defined area, i.e. ward as a geographic entity.5 Thus, wards are an important component of the electoral process. Democratically elected councillors and ward committees allow citizens an opportunity to be represented in local government.6 Local government uses wards as a basis to plan and implement service delivery in response to concerns raised by citizens at ward level.7 Electoral wards serve the purpose of

• rganising voting during elections, the appointment and functioning of ward committees, and

the planning, implementation and monitoring of service delivery. 8 Furthermore, wards offer political parties the means to maximise their control over municipal

• councils. Winning the most wards usually means greater control of the municipal council. In the

South African context, wards are the smallest geographic entity as far as administrative boundaries are concerned. Unlike the aggregated national and provincial election outcomes, ward-based election results indicate variations of support across different territories. Outcomes

• f ward-based local government elections offer political parties the means to gauge their support

in different parts of the country. Wards that consistently have low voter turnout and frequent service delivery protests indicate citizens’ dissatisfaction.9 For politicians and their respective parties, the purpose of having wards is to maximise political control. On the other hand, for citizens, wards are for choosing representatives who are expected to voice citizens’ concerns and interests in local government.10 Thus, any ward delimitation process needs to create wards that will ensure that both purposes are equally served. Ward boundary changes that might negatively affect the advancement of the citizens’ interests usually result in protest action. For instance, during the month of June 2015,

4

residents of Ward 93 in Durban protested the incorporation of a new area into their ward.11 Moreover, political parties do clash whenever proposed wards appear to be detrimental to their political interests. For instance, the first local government elections in the democratic South Africa were postponed to June 1996 due to ward delimitation disputes in Western Cape and

• KZN. The disputes were between opposition-led provincial governments and the ANC-led national

government over the demarcation of wards in the Cape and Durban metropolitan areas.12 Further afield, local government elections in the city of Bengaluru, India, were postponed after the state government discovered that there was non-compliance with the ward delimitation guidelines.13 In the United States of America (USA), litigation over electoral boundaries is not uncommon, as courts always intervene in electoral boundary disputes.14 Overall, these incidences indicate that the re-determination and delimitation of electoral boundaries is a contentious and highly contested process.15 Despite the disagreements, electoral boundaries have to change over time. Population change is the main reason for changes in electoral.16 Population change refers to population shifts across settlements due to migration and population growth. These changes bring about the need to delimit electoral boundaries.17 The practice of changing electoral boundaries is a worldwide

• phenomenon. Electoral legislation in the USA stipulates that electoral boundaries have to be

reviewed as soon as the population census data become available.18 India also bases its ward delimitation process on population census data.19 It must be noted that unlike in the USA, South African legislation does not explicitly state that the review of electoral boundaries must be triggered by new population census data.20 21 22 Electoral ward delimitation processes vary across democratic nations due to varying legislative requirements and electoral systems. Thus, the criteria used to delimit these boundaries will vary. Despite all variations in the criteria used, there are three critical characteristics of electoral geographic entities, i.e. population equity, contiguity and geographical compactness.23 In democratic nations throughout the world the process of determining electoral boundaries is aimed at producing electoral geographic entities with the aforementioned characteristics. The

• verall objective is to achieve fair representation of citizens and to enhance participatory

democracy.24

5

Thus, the electoral geographic entities, such as wards, should contain approximately equal populations, and be contiguous and geographically compact.25The first critical characteristic of electoral geographic entities, which will now be discussed, is population equity. Population equity is approached in different ways. Some countries, such as the USA and Australia, strive to achieve equal numbers of people across constituencies. In those countries the electoral boundary delimitation process is largely based on population data obtained from population censuses. Thus, the US census tracts or blocks are neatly nested and completely within the boundaries of electoral precincts.26 In Australia the census collection districts are neatly nested and completely within the boundaries of voting districts (VDs).27The same applies in Canada. The main difference is that the Canadian delimitation process seeks to achieve equity in the distribution of eligible voters.28 In South Africa, a critical characteristic of electoral wards is voter parity. The ward delimitation process seeks to ensure that wards contain approximately equal numbers of registered.29 The second critical characteristic of electoral geographic entities is contiguity. Basically this means that there must not be any spaces between electoral geographic entities. Secondly, no portion of one entity should be completely within another entity. Thus, electoral geographic entities should consist of connected subunits.30 In South Africa, electoral ward contiguity is maintained by merging voting districts to form wards.31 In other words, voting districts are smaller building blocks for electoral wards. Voting districts are determined by the Independent Electoral Commission (IEC) as provided for by the Electoral Act of 1998.32 Key factors considered by the IEC when determining voting districts are mainly around the logistics of conducting an election. This includes the availability of a suitable voting venue, transport and Information Communication Technology (ICT) infrastructure, and the number of registered voters.33 Overall, a critical requirement pertaining to the contiguity of wards is that no voting district of ward x should be fully enclosed by voting districts of ward y, and vice versa. The third critical characteristic of electoral geographic entities is geographic compactness. Ideally, electoral geographic entities should have round, block-like or compact shapes. Elongated, snakelike, or twisted shapes are an indication of political manipulation.34 Spatial statistical techniques have been used in Canadian cities to enforce geographic compactness based on a given criterion.35 In South Africa the ward delimitation criteria, contained in Schedule

6

1, section 4, of the Municipal Structures Act, stipulates that care should be taken to ensure that

• bvious groupings of communities, villages, townships or suburbs are not split during the

delimitation process.36 This is meant to enforce the geographic compactness of electoral wards. The three critical characteristics of electoral geographic entities (population equity, contiguity, and compactness) are essential to avoid malapportionment and gerrymandering.37 Malapportionment is when there are substantial variations in the size of the constituency, either measured by the number of registered voters or population. This creates a bias in the election results as one party might have a huge following in smaller wards, whilst other parties scramble for domination in the highly contested larger wards.38 Gerrymandering is the process of adjusting electoral boundaries to maximise the advantage of a particular political party or candidate.39 At this stage the discussion turns to a closer look at the South African ward delimitation process. When South Africa was transitioning into democracy, administrative boundaries were not uniform across the country, as there were numerous interim transitional local authorities that varied from one province to the next.40The Local Government: Municipal Demarcation Act and the Local Government: Municipal Structures Act were passed in 1998 to resolve this situation. These laws are based on the South African Constitution and they form the basis upon which all administrative boundaries are determined.41 The Municipal Demarcation Act provides for the creation of the MDB and it also gives the MDB legal powers to determine or re-determine metropolitan, district, local municipal, and ward boundaries. The Municipal Structures Act provides the legal basis, the process, and the criteria for the delimitating wards.42 The MDB undertakes ward delimitation in-between local government elections. The process must be finalised, and wards handed over to the IEC at least six months prior the local government elections.43 Figure 1 provides an overview of the MDB process pertaining to the review of municipal and wards’ boundaries.

7

Fi Figu gure re 1: MD : MDB p proc roces esses es from from 2 2011 to to 201644

44

The review of outer municipal boundaries must be finalised before the ward delimitation process can commence.45 The ward delimitation process itself consists of four phases, viz. preparations, public consultations, compliance with legal requirements and handover to the IEC.46 The first major milestone in the preparation phase is the implementation of technical working sessions with municipalities. The main aim of these sessions is for MDB staff to work with municipal GIS

• fficials, planners or other technical persons to create draft municipal wards. Thereafter, draft

wards are mapped for use in the second phase, i.e. public consultations.47 The second phase starts with a national launch of the ward delimitation consultative process. The launch is aimed at creating awareness, and to stimulate public participation in ward

• delimitation. The MDB regards stakeholder consultations as the most important phase, as the

success of the entire ward delimitation process depends on stakeholder consultations. In all the formal public consultative meetings, the MDB is represented by at least one Board member, accompanied by two staff members.48 The purpose of the stakeholder consultative meetings is to identify and discuss ward boundary proposals that are acceptable. Draft ward maps are presented, and alternative proposal are handed over to MDB representatives for consideration when finalising wards. The phase of public meetings lasts for three months, and the MDB continues accepting written alternative proposals up to ten days after the meetings.49

8

Phase three is mainly a legal process. In this phase, the MDB updates draft wards, taking into account inputs from the consultative process as well as the legal prescripts.50 These wards are published in provincial gazettes, as directed by the Municipal Structures Act. Portable Document File (PDF) versions of the ward maps are made available on the MDB website. There is a 14-day period, following the publication of the provincial gazettes, within which members of the public

• r affected organisations can object to the newly delimited wards.51 Objections are considered

for a period of two months.52 Thereafter, the MDB will make a decision and publish finalised

• wards. The ward delimitation process effectively ends after the official handover to the IEC, i.e.

phase four.53 Thereafter, the IEC will commence with logistical and other arrangements in preparation for the next local government elections.54 A critical component of the four-phased broad process discussed so far is the actual process of drawing up electoral wards, i.e. ward delimitation. The technical methods for delimiting wards will now be discussed. In South Africa, the technical process of drawing wards is done in the municipal technical working sessions.55 Four key data-sets need to be in place in order to delimit ward boundaries, viz. number of required councillors per municipality, number of registered voters allowed per ward, voting districts, and registered voters per voting district.56 Firstly, the number of required councillors per municipality needs to be determined. For this to happen, the IEC divides the national voters roll into municipal segments and hands over to the Minister of Cooperative Governance and Traditional Affairs. Based on the number of registered voters contained in the segmented voters roll, the Minister publishes formulae for determining the number of required councillors in different municipal categories. Table 1 below contains the formulae used to prepare for the 2011 local government elections.

9

Tab able le 1: Year : Year 2011 For Formulae for Determ ulae for Determini ining ng the N the Number of R er of Required Counci equired Councillor in llor in D Differen ifferent t Mu Munic nicip ipal Cat al Categories egories In the formulae above: y = number of required councillors x = the number of registered voters on the municipality’s segment

• f the voters roll

Category A refers to a metropolitan municipality. Categories B and C refer to local municipalities and district municipalities respectively.57 Provincial Members of the Executive Committee (MECs) responsible for local government apply the formulae to determine the number of required councillors per municipality.58 Thereafter the MDB divides the number of required councillors by 2 to get the number of required wards per municipality.59 Fractions are rounded upwards.60 For instance, a municipality that requires 11 councillors should have 6 wards, i.e. 11 ÷ 2 = 5.5 (0.5 rounded up to make 6). A provision of two councillors per ward is made because in South Africa, half of the councillors are elected based on the proportional representation (PR) system, based

• n party lists. The other half is elected using the first-past-the-post system. In terms of the PR

system the party with the greatest proportion of votes wins the ward, thereby gaining a seat in the municipal council. In terms of the first-past-the-post system, ballots are cast to elect ward councillors from a list of party representatives and independent candidates.61 The second data requirement for drawing ward boundaries is the number of registered voters allowed per ward. This is achieved by dividing the number of registered voters in a municipality by the number of required wards in that municipality. The resulting value is referred to as the norm, i.e. average number of registered voters allowed per ward.62 The Minimum (fifteen per

Municipal category Formula Category A municipality y = (x ÷ 10 000) + 60 Category B municipality with less than 7 501 registered voters y = 5 Category B municipality that has between 7 500 and 100 001 registered voters (x ÷ 1 682) + 1 Category B municipality that has more than 100 000 registered voters y= (x ÷ 8 333)+ 48 Category C municipality that has less than 100 001 registered voters y= (x ÷ 9 500)+ 9 Category C municipality that has more than 100 000 registered voters y= (x ÷ 12 000)+ 12

10

cent below the norm) and maximum (fifteen per cent above the norm) number of voters allowed per ward also have to be determined. For instance in a municipality with 50 000 registered voters and 10 wards the norm will be 5 000, i.e. 50000 ÷ 10 = 5000. Therefore, the maximum number

• f voters allowed per ward will be 5 750, i.e. 5000 + 15% of 5000. The minimum number of

voters allowed per ward then becomes 4 250, i.e. 5000 – 15% of 5000. The third data requirement is a GIS file containing the names and numbers of voting districts in each municipality. These data are obtained from the IEC and form the basis for creating ward boundaries, since voting districts are clustered to form wards. An important control variable for the clustering of VDs is the number of registered voters in each voting district, which is the fourth data requirement. This variable is important since the clustering of VDs must result in wards that are within the 15% deviation from the norm. Most importantly, the clustering should create contiguous wards.63 The aforementioned four key data sets form a fundamental basis for delimiting wards, and the process itself is guided by the ward delimitation criteria as stipulated by the Municipal Structures

• Act. Essentially, the criteria contain five main requirements.64 Firstly, wards must comply with

the minimum and maximum deviations from the norm. Wards should be as close to the norm as possible and no wards should be below or above the fifteen per cent deviation.65 Secondly, wards must not fragment or split communities.66 Thirdly, ward boundaries should follow identifiable natural and man-made features, e.g. rivers, ridgelines, roads, etc.67Fourthly, the shape and size

• f a ward should make it easy and convenient for citizens to interact regularly, thereby enhancing

participatory democracy. Therefore, ward delimitation should consider distances, topography and other physical characteristics.68 Lastly, voting stations should be easily accessible. Once again, roads, topography and other physical characteristics must be considered when delimiting wards.69

11

3. . MET METHO HODOLOGY OLOGY A quantitative positivistic research methodology was followed to conduct the empirical study. The positivist metatheory is more inclined to the application of quantitative research methods and techniques.70 This includes the use of measurement scales and indices, as well as quantitative statistical techniques of data analysis.71GIS-based analyses and statistical methods were applied in two interrelated components of the study. Analysis was done to firstly assess the compliance of electoral wards in terms of voter parity, population equity and eligible voter equity; and secondly to measure the degree to which electoral wards fragment human settlements. As part of the methodology the following aspects will be discussed. Firstly the data and data sources used are summarised. Thereafter the spatial analysis process is discussed, i.e. similarity checks, preparing the wards for scenarios, assessing compliance with the norm, correlation and regression analysis and measuring the fragmentation of settlements. 3.1 .1 Data ata and D and Data ata Sources Sources All spatial data used in the study are listed in Table 2 with their format, year of capture and the source of the data Table 2: Sum Summary ary of th

• f the Da

e Data a ta and Da nd Data S ta Source

• urces

ArcGIS 10.2 was used to handle all spatial data. The shapefiles were converted into geodatabase feature classes and incorporated into a file geodatabase created for the analysis. The feature classes were projected to Africa Albers Equal Area Conic. The aforementioned projection was

Data Format Year Source Ward boundaries Shapefile 2011 Municipal Demarcation Board 2016 draft ward boundaries Shapefile 2015 Municipal Demarcation Board District municipalities Shapefile 2010 Municipal Demarcation Board Local municipalities Shapefile 2010 Municipal Demarcation Board Main places Shapefile 2013 Statistics South Africa Sub-places Shapefile 2013 Statistics South Africa Population by KZN Wards Excel spreadsheet 2011 Statistics South Africa Voting Age Population (VAP) Excel spreadsheet 2011 Statistics South Africa Number of registered voters by Voting District (VD) Excel spreadsheet 2011 Independent Electoral Commission Number of registered voters by electoral ward Excel spreadsheet 2011 Independent Electoral Commission

12

chosen as it can precisely calculate area. Area calculations were done on wards and population densities were calculated by dividing population by area. Excel spreadsheet containing population figures per ward were extracted from the Census 2011 database using SuperCross Version 8.0.2.32, i.e. a cross tabulation software. The IEC provided a spreadsheet containing voter registration figures. The tabular data were linked to the “WARDS” polygon feature classes within ArcGIS. Table 3 shows the database schema of the “WARDS” feature class with the population and registered voter columns added on. Tab able le 3: : Data atabase ase Sc Schem hema for a for the W the Ward F ard Featur eature Cla e Class The “REG_VOTERS” field was essential in measuring voter parity between wards. The “POPULATION” and “EligibleVoters” columns were essential for measuring whether there is an equitable distribution of the population and eligible voters across electoral wards. Overall, the voter registration data from the IEC and the Stats SA data were essential in preparing the “WARDS” feature class for the first part of the spatial analysis. Data needed for the second part of the analysis was obtained from StatsSA, namely two GIS shapefiles, one containing main places and the other one containing sub-places of KZN. The main place feature class contains 3 121 main places and the sub-place feature class contains 4 198 polygons, which is the number of sub-places in KZN. Placename feature classes (containing both main places and sub-places) contain contiguous polygons representing boundaries of towns, suburbs, townships and villages. Main places are a higher level category comprising sub-places. The position of main and sub places in the broader geographic hierarchy

• f South Africa is illustrated in Figure 2.

Column Description Sample data PROVINCE Name of province KwaZulu-Natal MUNICNAME Name of municipality Msinga Local Municipality WARDNO Commonly used ward number 10 WARD_ID Unique ward ID in databases 52404010 SQ_KM Area in square kilometres 48 POP_DENSITY Population density 183 REG_VOTERS Registered voters 3928 POPULATION Population in 2011 8838 EligibleVoters Voting-age population, i.e. aged 18 and above (Year 2011) 4446

13

Fi Figu gure re 2: : Geogra eographic hic Hierarchy Hierarchy of

• f South A

South Afric frica Figure 2 indicates that placenames are a level below the local and metropolitan municipal level. Electoral wards are also a level below the municipal level. Placenames and wards are demarcated independently of each other, since the demarcation of wards is the sole responsibility of the MDB, whilst Stats SA is responsible for placenames. Meanwhile, the ward delimitation criteria stipulate that wards must not cut across communities. Thus, the degree of

• verlap between electoral wards and human settlements (represented herein by the placename

feature classes) was measured in the second component of the spatial analysis. 3.2 .2 Sp Spati atial A al Analy nalysis The first part of the analysis was aimed at assessing electoral wards in terms of their compliance with the norm as stipulated in the ward delimitation criteria. The second part of the spatial analysis focused on measuring the degree to which electoral wards comply with the second requirement stipulating that ward boundaries must not fragment communities. 3.2 .2.1 .1 Similarity Similarity C Checks hecks Similarity checks were done to check whether there is any proposed change on ward boundaries, as is the case with local municipal boundaries. The Compare Features Tool (an ArcGIS tool) was run on the 2011 wards and the draft 2016 feature classes. The feature classes were compared in terms of feature geometry similarities, i.e. whether the shape and size of ward polygons are identical in both layers. The Compare Features Tool report indicated that both wards layers are spatially identical. The Municipal Demarcation Board confirmed that spatial files of the 2011

14

wards and the draft 2016 wards are identical because wards used in the previous election automatically become a draft version for the next election as soon as the ward delimitation process is launched.72 It is envisaged that there will be significant changes in the final 2016 wards as some wards will be split, i.e. the number of wards is expected to increase from 828 to 940 in KZN.73 74 Currently however, the only difference between 2011 and 2016 wards is that the 2011 wards had fields referring to voting results. For the purposes of the study the 2011 wards layer was chosen to carry out the spatial analysis as it is the most complete and officially gazetted dataset. 3.2 .2.2 .2 Pre Preparing aring Wards f ards for Sc

• r Scen

enario arios The first component of the spatial analysis was aimed at assessing the compliance of electoral wards in terms of voter parity, population equity and eligible voter equity. The actual compliance assessment for the wards was done in three GIS-based scenarios. In the first scenario, the distribution of registered voters across wards was assessed. In the second scenario, population distribution across wards was assessed. In the third scenario, equity was assessed in terms of the distribution of eligible voters. During the preparation phase, data contained in the attribute tables of the wards feature classes were checked for accuracy and completeness before subjecting the data to further analysis. Table 4 contains the database schema for the 2011 wards after data preparation.

15

Tab able le 4: 4: Data atabase ase Sc Schem hema for a for the 2 the 2011 W Wards ards 3.2 .2.3 .3 Asses essing ng Com Complianc liance w e wit ith the N h the Nor

• rm

GIS analyses focused on ascertaining whether the geographic structure and composition of KZN wards is adhering to the legal requirement of wards consisting approximately equal number of registered voters. A further exploratory analysis was performed to check whether wards contain approximately equal numbers of people and eligible voters. The rationale for including population and eligible voters was to measure the relevance and appropriateness of electoral wards as a means of fair and balanced representation for citizens. Ensuring that wards contain approximately equal populations and eligible voters is also in line with international norms as discussed in the literature. The first component of the spatial analysis was aimed at assessing the spatial distribution of three variables across wards, i.e. registered voters, eligible voters and the ward population. ArcGIS queries were run on the attribute table of the wards feature classes to identify wards in which values of the three variables are greater than the maximum number allowed per ward. For those wards the “Compliance Status” column was populated with “Above Max Norm”. Another query was run to identify wards in which values of the three variables are below the minimum number allowed per ward. For those wards, the “Compliance Status” column was populated with “Below Min Norm”. The statements “Above Max Norm” and “Below Min Norm” indicate that those wards are not complying with the equity requirement. Wards in which values of the three

Column Description Sample data PROVINCE Name of the province KwaZulu-Natal CAT_B Unique identifier for the municipality ETH MUNICNAME Name of the municipality eThekwini Metropolitan Municipality WARDNO Ward number 20 WARD_ID Unique ward identifier 59500020 SQ_KM Area in square kilometres 4 HOUSEHOLD S Number of households (Census 2011) 6613 POP_DENSIT Y Population density 5596 POPULATION Number of people (Census 2011) 23536 REG_VOTERS Number of registered voters as on 06 March 2011 16631 NonVoters Number of people below the age of 18 (Census 2011) 7066 EligibleVoter Number of people aged 18 and above(Census 2011) 16471 PercNonVoter Percentage of non-voters 30 PercElgible Percentage of eligible voters 70 Norm Average number of voters allowed per ward 23246 NormMin Minimum number of voters allowed per ward 19759 NormMax Maximum number of voters allowed per ward 26733 DEVIATION 15% deviation from the norm 3487 ComplStatus Compliance status of the ward Below Min Norm

16

variables are within the permissible 15% deviation were given the status of “Within The Norm”. These are complying wards. Results of the first scenarios demonstrated the degree to which electoral wards of KZN comply with the first requirement of the ward delimitation criteria, i.e. voter parity. The second and the third scenarios indicated the degree to which wards of KZN contain approximately equal numbers of people and eligible voters, i.e. population equity and eligible voter equity. 3.2 .2.4 Correlati Correlation an

• n and re

d regres gression analy ion analysis Statistical techniques were run on the data to complement findings of the GIS-based scenarios, in that these techniques helped to identify a variable that could be used as a key determinant of the geographic size of wards. The statistical software package that was used is IBM SPSS Version

• 22. The purpose of performing correlation and regression analysis was two-pronged. Firstly, it

was aimed at measuring the strength and direction of the association between the 2011 ward area, and the densities of registered voters; eligible voters; and total population. A Pearson correlation coefficient was applied to measure the relationship between the variables. Secondly, the aim was to identify if any of latter three variables (registered voters, eligible voters and population) is the key determinant of ward size. A regression analysis was performed to this effect. 3. 3.2.5 Mea Measuring uring the F the Frag ragmen entat tation of Settleme ion of Settlements nts The second component of the spatial analysis was aimed at measuring the degree to which electoral wards fragmented human settlements. Three key spatial data sets were used in this component of the analysis, viz. KZN main places, KZN sub-places, and KZN 2011 wards. A spatial analysis was performed to quantify overlaps between placename boundaries and ward

• boundaries. The placename data are a spatial representation of human settlements. Thus, the

Union Tool, an ArcGIS tool, was applied to quantify settlement fragmentation caused by ward

• boundaries. Choropleth maps showing results of the analysis were created.

4. . Re Results ults and D and Discus ussion ion This section of the paper contains findings of the study. Under section 4.1, and its subsections, results of the GIS-based scenarios created to assess the compliance of wards with voter parity, population equity and eligible voter equity are discussed. Firstly, the voter parity scenario is

17

• discussed. Secondly, the population equity and eligible voter scenarios are discussed. Lastly,

section 4.2 contains results and a discussion of the spatial analysis performed to measure the degree of human settlement fragmentation caused by ward boundaries in KZN. 4.1 .1 Voter

• ter Parit

Parity, Pop , Population Eq ulation Equity uity and Eligib and Eligible Vot le Voter er Eq Equity uity 4.1 .1.1 .1 Re Results ults fro from Sc Scen enario 1 ario 1: : Asses essing ing Voter Pa

• ter Parit

rity The map in Figure 3 shows that 86% (715 out of 828, green areas on the map) of the 2011 wards complied with the voter parity requirement. The number of registered voters in these wards was within the maximum and minimum number of voters allowed per ward, in their respective municipalities. In other words, the complying wards had approximately equal numbers

• f registered voters. However, Figure 3 shows that the number of registered voters in 57 wards

(red areas) exceeded the maximum number of voters allowed per ward. Moreover, 56 wards (blue areas) had numbers of registered voters that were below the minimum number of voters allowed per ward. In total, 14% (113 wards, blue and red areas) did not comply with the legal

• requirement. Nevertheless, in a large percentage (86%) of KZN wards, numbers of registered

voters were within the prescribed legal limits. Fi Figu gure re 3: : Voter

• ter Parit

Parity Sc Scen enario A ario Asses essment R nt Res esult ults ( (Sc Scen enario ario 1, 2 , 2011) The presence of the 113 wards that are non-compliant indicates that using the number of registered voters to determine the number and sizes of wards has some limitations. The root cause of this limitation is the variable used to determine the number of required councillors per

18

municipality, i.e. registered voters. The voter registration figure is a key parameter in the formulae for determining the number of required councillors. Ultimately, the number of required councillors determines the number of required wards. A fundamental flaw is that ward delimitation is largely informed by a snapshot of the voters’ roll, at the time the minister publishes the formulae for determining the number of required councillors, which happens many months before the actual election. For instance, for the 2011 elections, the number of required wards was finalised in July 2009.75 Between July 2009 and March 2011, the voters’ roll continued to grow, as more people were registering to vote.76 In preparing for the 2011 local government election, the last voter registration weekend was that of 05th and 06th March 2011.77 The election was then held on 19th May 2011.78There was a time lag of 18 months from the point when the required number and size of wards were determined to the point when voter registration figures were finalised. Adjustments to ward boundaries based on the final voter registration figures do not happen, because the MDB concludes ward delimitation before the voters’ roll is closed. During elections some wards are therefore more likely to end up being non-compliant with the voter parity

• requirement. The voter parity scenarios highlight this limitation, because the voter registration

figures used in the scenario are as on 06th March 2011, whilst the number and size of wards were determined on the basis of July 2009 voter registration figures. 4.1 .1.2 .2 Re Results ults fro from Sc Scen enario 2 ario 2: : Asses essing ing Pop Population Eq ulation Equity uity The map in Figure 4 shows that more than half (56%, green areas) of the 2011 wards consist of approximately equal numbers of people. The other 44% are wards that are either too small or abnormally large. The 44% is split almost 50/50 between wards with relatively smaller populations and those with larger populations, i.e. 184 large wards and 186 small wards. Splitting the 184 (22%) abnormally large wards will certainly increase the number of wards. However, merging the remaining 186 (22%) to create larger wards, might theoretically offset that

• increase. Moreover, population equity is already achieved in most electoral wards of

municipalities such as Msinga, Ingwe, Jozini, Umhlabuyalingana, Okhahlamba and Nqutu. These findings support the contention that total population can be used as the key determinant of the number and size of electoral wards.

19

Fi Figu gure re 4: : Po Population ulation Eq Equity uity A Assessmen ent Re t Results ults (Sc Scen enario ario 2, 2 , 2011) Population equity across electoral wards ensures fair political representation and enhances participatory democracy.79 Abnormally large wards, in terms of population and geography, might limit the participation of ordinary citizens in ward-based decision-making processes, e.g. in the identification of priority projects to be included in the municipal Integrated Development Plan (IDP). Inevitably, less participation by citizens leads to the formation of cabals representing interests of a few politically connected elites. Moreover, such cabals might also successfully

• ppose ward boundary changes that might dismantle their power base. It is imperative that ward

delimitation remains an independent and non-partisan process, aimed at ensuring that wards consist of approximately equal populations. 4.1 .1.3 .3 Re Results ults fro from Sc Scen enario 3 ario 3: : Asses essing ing Eli Eligib gible Vot le Voter Eq er Equity uity The results of the scenario in Figure 5 are similar to the population equity scenario. Proportions

• f eligible voters in more than half of the wards (53%) are approximately equal. The non-

compliant wards almost break even, i.e. 22% of the wards have a high concentration of eligible voters and 25% of the wards have a low concentration of eligible voters. Generally, voting-age population increases or decreases proportionally to the total population. Thus, voting-age population can be used, as a second option to total population, to determine the number of electoral wards and the size thereof. Municipalities comprising high concentration of non- compliant wards need to be prioritised during ward delimitation, if such an approach to ward delimitation is to be followed. Wards in which there are abnormally low concentrations of eligible

20

voters need to be merged with other wards to eliminate redundancy in the allocation of electoral resources, e.g. voting stations, and other IEC resources. Wards in which there is an abnormally high concentration of eligible voters need to be split to ensure that the voting-age population is evenly spread throughout the municipality. Fi Figu gure re 5: : Eli Eligib gible le Voter E

• ter Equity

uity A Asses essmen ent t Re Result ults ( (Sc Scen enario ario 3, 2 , 2011) 4.1 .1.4 .4 Fi Finding ndings of the Correlat

• f the Correlation and

ion and Re Regre gression A ion Analy nalysis is Table 5 indicates that there is a statistically significant relationship between ward area, population density, eligible voter density and registered voter density. The relationship is more pronounced between ward area and population density, i.e. the r value is -0.277. Thus it becomes essential to test whether population density is the main predictor of ward geographic

• area. A regression was done on the two variables, and the results are presented in Table 6.

21

Tab able le 5: : Re Result ults Of T Of The Correlation A he Correlation Analy nalysis is AREA SQ_KM POPULATION DENSITY ELIGIBLE VOTER DENSITY REGISTERED VOTER DENSITY AREA_SQ_KM Pearson Correlation 1

• .277**
• .264**
• .258**
• Sig. (2-tailed)

.000 .000 .000 N 828 828 828 828 POPULATION_ DENSITY Pearson Correlation

• .277**

1 .993** .969**

• Sig. (2-tailed)

.000 .000 .000 N 828 828 828 828 ELIGIBLE_VOTE R_DENSITY Pearson Correlation

• .264**

.993** 1 .975**

• Sig. (2-tailed)

.000 .000 .000 N 828 828 828 828 REGISTERED_ VOTER_DENSITY Pearson Correlation

• .258**

.969** .975** 1

• Sig. (2-tailed)

.000 .000 .000 N 828 828 828 828 **. Correlati . Correlation is

• n is signific

ignificant at ant at the 0 the 0.0 .01 level ( level (2-tai tailed) led). Tab able le 6: : Re Result ults Of T Of The R he Regres gression A ion Analy nalysis Done is Done Betw etwee een Pop n Population ulation D Den ensit ity A And Wa nd Ward A d Area rea Model R R Square Adjusted R Square

• Std. Error of the Estimate

1 .277a .077 .076 184.67802122099400

• a. Predictors: (Constant), POPULATION_DENSITY
• b. Dependent Variable: AREA_SQ_KM

22

In the regression analysis population density was placed as an independent variable, and ward area was placed as the dependent variable. This simple linear regression revealed that population density has very little influence on ward size. The regression coefficient (R Square value of 0.077) suggests that the independent variable predicts 7% of what happens in the dependant variable. The R Squared value is too close to 0. The ward geographic area is not mainly a function of population density. However, the population variable still outperforms the

• ther two variables, i.e. eligible voters and registered voters. The total number of people per unit

area has a stronger influence on ward area than the number of registered or eligible voters. 4.2 .2 Fragm Fragmen entat tation ion of Hu

• f Human Settle

an Settlemen ents ts The union tool created 8 892 fragments representing intersections of main place and ward

• boundaries. The intersection of sub-place and ward boundaries created 11 309 fragments. In

Figure 6, fragments of the land area of KwaZulu-Natal created by overlaps between placenames and ward boundaries are mapped. Map A shows results of the union feature class comprising main places and ward boundaries. Map B shows results of the union feature class comprising sub-places and ward boundaries. The fragments are categorised to indicate the degree of settlement fragmentation. Fi Figu gure re 6: : Ma Maps Sho Showing the Fragm ing the Fragmen entat tation of H ion of Human Settleme an Settlements nts in in KZN KZN The 0% to 20% class (red areas) represents a high degree of settlement fragmentation. Generally in these areas one might find boundaries of more than four electoral wards splitting a single settlement into numerous fragments. Fifty-eight per cent of the main place fragments fall within

23

this category. The sub-place fragments constitute 56% of the fragments in the 0% to 20% class. In the 20% to 40% class (orange areas), the degree of fragmentation is moderate to high. In these areas one finds between two and four wards fragmenting a single settlement. Six per cent and five per cent of main place and sub-place fragments respectively fall within this class. Occurrence of fragments in the moderate zone is generally lower than in the 0% to 20%, and in the 80% to 100% zone. In the latter class, one finds a concentration of fragments that scored 100%. These fragments represent human settlements that are wholly within a single electoral ward. It must be highlighted that fragment categories do not represent classes of settlements. Instead, they represent the degree of overlap between settlements and electoral wards. Ideally, all fragments must score 100% to indicate that all settlements are neatly nested and wholly contained by electoral wards. Any fragment score below 100% denotes settlement fragmentation due to ward boundary alignment. The two maps in Figure 7 illustrate the concept of settlement

• fragmentation. Both maps show the fragments that resulted from the intersection of one ward

with three sub-places. An aerial image was added in the map on the right-hand side to give a visual representation of settlement fragmentation on the ground. The interpretation of the map is basically that 39% of Manzana SP; 26% of Jakkalspan SP; and 22% of Osizweni A fall within electoral ward number 30. The percentages are not contributions of the settlements to the geographic area of the ward. It must be remembered that the aim was to quantify the degree of the fragmentation of human settlements caused by electoral ward boundaries. Fi Figu gure re 7: : Sett Settlem lemen ent t Fra Fragmen entat tation at t ion at the Ward he Ward Sc Scale ale

24

Settlement fragmentation in KZN is mapped out in Figure 8. Non-fragmented settlements are highlighted in yellow, whilst the fragmented ones are greyed out. Map A shows main places, and Map B shows sub-places. 77% of both main places and sub places are fragmented. This means

• nly 23% are non-fragmented. In terms of the geographic area, fragmented main places

constitute 97% of the 93 332 square kilometres, which is the entire geographic area of KZN. Fragmented sub-places constitute 96% of the entire geographic area of KZN. There is no significant reduction in settlement fragmentation from main place to sub-place level, i.e. 1% difference. Fi Figu gure re 8: : Ma Maps Sho Showing the Fragm ing the Fragmen entat tation and Non ion and Non-Fragm Fragmen ented S ted Sett ettlem lemen ents in ts in KZN KZN Furthermore, there are no specific patterns in the distribution of fragmented and non-fragmented

• settlements. There is no discernible association between the compliance statuses of the wards

in terms of voter parity, and the fragmentation of settlements. In other words, there is no evidence to suggest that the non-compliant wards are causing settlement fragmentation. In fact, ward number 30 (Figure 7) is compliant with the voter parity requirement, i.e. the number of registered voters in that ward is within the acceptable limits. The compliant wards are also not necessarily associated with settlement fragmentation or non-fragmentation. Settlement fragmentation appears to be a widely spread random phenomenon. The main finding here is that the electoral ward boundaries of KwaZulu-Natal are not compliant with the second requirement

• f the delimitation criteria, which stipulates that ward boundaries must not fragment human

settlements.

25

Conclu Conclusion ion This paper has managed to shed some light on the electoral ward delimitation processes in South Africa and abroad. From the literature it can be concluded that there is growing recognition that for electoral geographic entities (such as wards) to be legitimate, care should be taken to ensure that they contain approximately equal numbers of people; they are contiguous; and geographically compact. In South Africa, stakeholder consultations and adherence to the legal prescripts are central to the success of the entire ward delimitation process. The Municipal Structures Act of 1998 outlines the ward delimitation criteria. The two key requirements in the criteria are voter parity and the non-fragmentation of communities. This means that South African wards should contain approximately equal numbers of registered voters. Moreover, wards should not fragment human settlements. Compliance of KZN wards with these two legal requirements was the subject of study in the GIS-based analyses. There were two main components of the GIS analysis. The first component of the spatial analysis was aimed at examining the wards of KZN with regard to compliance with voter parity, population equity and eligible voter equity. The second component measured the degree of human settlement fragmentation caused by ward boundaries in KZN. For the first component three scenarios were developed. In all three scenarios the concept of equal distribution of persons across electoral wards was tested. The voter parity scenario highlighted the limitation of using the number of registered voters to determine the number and size of wards. It is a conclusion of this study that 113 wards of KwaZulu-Natal are not compliant with the first requirement of the delimitation criteria, i.e. voter parity. The population equity and eligible voter scenarios are exploratory, because population equity and eligible voter equity are not yet a legal requirement for South African electoral wards. Both scenarios present opportunities to overcome shortcomings identified in the voter parity scenario. In many parts of the world electoral boundary changes are based on the changes in the total population.80 81 82 South Africa needs to follow this approach. The main merit of using total population as a basis for determining electoral boundaries is the fact that the population figure is inclusive. Registered and eligible voter figures are subsets of total population. Electoral ward boundaries will not be affected by fluctuations in the latter two variables if such ward boundaries were initially based on the population size. Therefore, a ward delimitation process in which total population is a key determinant of the number and size of electoral wards is recommended. The policy implication here is that population data should substitute the voters’ roll as a key variable used to determine the number of required councillors and, subsequently, the number of required wards.

26

The second component of the spatial analysis measured the degree of human settlement fragmentation caused by ward boundaries in KZN. GIS tools were used to measure and map fragmentation caused by overlaps between electoral wards and KZN placenames. High levels of the fragmentation of human settlements caused by electoral ward boundaries were discovered. This means that the electoral ward boundaries of KwaZulu-Natal are not compliant with the second requirement of the delimitation criteria, which stipulates that ward boundaries must not fragment human settlements. A feasible option to remedy the situation going forward is to delimit wards that serve as both electoral and service delivery units. The implication of this is that wards should be demarcated as part of a broader hierarchical national geographic frame. This calls for inter-departmental collaboration in the determination of administrative boundaries. A notable limitation of this study is the fact that the 2016 wards have not been finalised. Moreover, population data at ward level are only available for the year 2011, i.e. following the population census of the same year.83 Therefore, a follow up-study to measure voter parity and settlement fragmentation resulting from the finalised 2016 electoral wards is recommended. Further research can also focus on municipal-level studies on the causes and nature of ward boundary related disputes. Simulated GIS projects whereby boundaries of census Enumeration Areas (EAs) form the basis for creating contiguous and geographically compact electoral wards, that contain approximately equal numbers of people, are also recommended. The merits of using population data as a key determinant of the number of wards and the size thereof can be explored in such research undertakings. Overall, the paper presented herein has shown that the geographic structure and composition of the 2011 electoral wards of KwaZulu-Natal does not reflect the ideal process or factors that need to be considered as prescribed in the ward delimitation criteria. Notes

• tes and

and Re Referen ferences es

1Gopalan, R.,Kimbrough, S.O., Murphy, F.H., Quintus, N., 2013. The Philadelphia Districting

Contest: Designing Territories for City Council Based Upon the 2010 Census. Interfaces, vol 41, no 6, p477– 489.

2 Forest, B., 2004. Information Sovereignty and GIS: the evolution of “communities of interest’

in political redistricting. Political Geography, vol 23, p425 – 451.

27

3 Republic of South Africa, 1998. Local Government: Municipal Structures Act, Act 117 of

• 1998. Government Gazette 19614, 3.7.1998.

4 Du Plessis, N.W., 2002. An Electoral System for South Africa: Various Options. Available at

www.elections.org.za [Accessed 08 June 2015].

5 Mills, G., 1976. The determination of local government electoral boundaries. OR, vol 18,no

3,p243 – 255.

6 Gottmann, J., 1980. Spatial partitioning and the politician's wisdom. International Political

Science Review, vol 1,no 4,p432-455.

7 Napier, C.J., 2007. Delimitation of local government ward boundaries in the Tshwane

municipal area: the challenge of achieving fair political representation. Politeia, vol 26, no2, p179 -191.

8 Visser, J., Steytler, N. & Mettler, J., 2000. Electing Councillors: A guide to municipal elections.

Cape Town, University of the Western Cape: Community Law Centre.

9 Kersting, N., 2012. Local government and reforms in South Africa. Politeia, vol 31, no 1, p5 –

21.

10 Montanaro, L., 2012. The democratic legitimacy of self-appointed representatives. The

Journal of Politics, vol 74,no 4,p1094 - 1107.

11 Stolley, J., 2015. Ward boundary changes at the heart of Durban protest. News 24,04 June

2015, Available at http://www.news24.com/SouthAfrica/News/WardboundarychangesattheheartofDurbanprotest 2015064 [Accessed on 04 June 2015].

28

12 Lupton, M. & Mather, C. 1996. The anti-politics machine: GIS and the reconstruction of the

Johannesburg local state. Political Geography, vol 16, no 17, p565 – 580.

13 Raghunandan, P.M., 2015. Glaring errors in Palike ward delimitation report. Deccan Herald

26 February Available at http://www.deccanherald.com/content/462281/glaring-errors-palike- ward-delimitation.html [Accessed 02 June 2015].

14 Forest, B., 2012. Electoral redistricting and minority political representation in Canada and

the United States. The Canadians Geographer, vol 56, no 3, p318 – 338.

15 Chou, C.I. & Li, S.P., 2006. Taming the gerrymander: Statistical physics approach to political

districting problem. Physica A, vol 369, p799 – 808.

16 Bozkaya, B., Erkut, E., Haight, D., & Gilbert, L., 2011. Designing New Electoral Districts for

the City of Edmonton. Interfaces, vol 41, no 6, p534–547.

17 Gopalan, et al, 2013, p477– 489. 18 Ibid. 19 Raghunandan, P.M., 2015. 20 Republic of South Africa, 1998. Electoral Act, Act 73 of 1998. Government Gazette 19018,

1.10.1998

21 Republic of South Africa, 1998b. Local Government: Municipal Demarcation Act, Act 27 of

• 1998. Government Gazette 19020, 18.12.1998.

22 Republic of South Africa,1998c. Local Government: Municipal Structures Act, Act 117 of

• 1998. Government Gazette 19614, 3.7.1998.

23 Gopalan et al, 2013, p477– 489. 24 Sivaramakrishnan, K.C, 2001. Constituencies delimitation: Deep freeze again?. Economic

and Political Weekly, vol 36, no 51, p4694 – 4696.

29

25 Gopalan et al, 2013, p477– 489. 26 Forest ,B., 2004. Information Sovereignty and GIS: the evolution of “communities of interest’

in political redistricting. Political Geography, vol 23, p425 – 451.

27 Eagleson, S., Escobar, F., Williamson, I., 2002. Hierarchical spatial reasoning theory and GIS

technology applied to the automated delineation of administrative boundaries. Computers, Environment and Urban Systems, vol 26, p185 – 200.

28 Hojati, M., 1996. Optimal political districting. Computers Operational Research, vol 23,

no12, p1147 – 1161.

29 Sutcliffe, M., 2002. The South African demarcation process. Proceedings of the French-

South Africa meeting on territorial innovation held Januray 2002. Grenoble-Avignon, France: HAL

30 Bozkaya, B., Erkut, E. & Larporte, G., 2003. A tabu search heuristic and adaptive memory

procedure for political districting. European Journal of Operational Research, vol 144, p12 – 26.

31 Municipal Demarcation Board, 2014c. Delimitation of ward boundaries for the 2016 local

elections: Awareness and public participation programme.Circular 2/2014. Pretoria: Municipal Demarcation Board. Available from www.demarcation.org.za [Accessed 15 March 2015].

32 Republic of South Africa, 1998a. Electoral Act, Act 73 of 1998. Government Gazette, 19018,

1.10.1998

33 Visser et al, 2000. 34 Bozkaya, B., Erkut, E., Haight, D. & Gilbert, L., 2011. Designing New Electoral Districts for the

City of Edmonton. Interfaces, vol 41, no 6, p534–547.

35 Bozkaya et al., 2003,

30

36 Republic of South Africa, 1998c. 37 Bozkaya et al., 2011. 38 Johnston, R., 2002. Manipulating maps and winning elections: measuring the impact of

malapportionment and gerrymandering. Political Geography, vol 21, p1 – 31.

39 Chou, C.I., & Li, S.P., 2006. Taming the gerrymander: Statistical physics approach to political

districting problem. Physica A, vol 369, p799 – 808.

40 Giraut, F. & Maharaj, B., 2002. Contested terrains: Cities and hinterlands in post-apartheid

boundary delimitations. GeoJournal, vol 57, p39 – 51.

41 Sutcliffe, M., 2002. 42 Municipal Demarcation Board, 2012b. Strategic plan for the fiscal years 2012 to 2017.

Pretoria: Municipal Demarcation Board. Available from www.demarcation.org.za [Accessed 02 June 2015].

43 Sutcliffe, M., 2002. 44 Municipal Demarcation Board, 2015b. Municipal Demarcation Board (MDB) outlines ward

delimitation process priorities. Pretoria: Municipal Demarcation Board. Available from www.demarcation.org.za [Accessed 15 March 2015].

45 Municipal Demarcation Board, 2011. Redetermination of municipal boundaries. Circular

1/2011. Pretoria: Municipal Demarcation Board. Available from www.demarcation.org.za [Accessed 14 February 2015].

46 Municipal Demarcation Board, 2014b. Delimitation of wards during 2014/15 for the 2016

local elections. Circular 4/2014. Pretoria: Municipal Demarcation Board. Available from www.demarcation.org.za [Accessed 15 March 2015].

47 Ibid. 48 Municipal Demarcation Board, 2014c. Delimitation of ward boundaries for the 2016 local

elections: Awareness and public participation programme. Circular 2/2014. Pretoria: Municipal Demarcation Board. Available from www.demarcation.org.za [Accessed 15 March 2015].

31

49 Ibid. 50 Municipal Demarcation Board 2014a. Delimitation of wards during 2014/15 for the 2016

local elections [online].Circular 3/2014. Pretoria: Municipal Demarcation Board. Available from www.demarcation.org.za [Accessed 15 March 2015].

51 Municipal Demarcation Board, 2015c. Ward delimitation: Public meetings. Circular 6/2015.

Pretoria: Municipal Demarcation Board. Available from www.demarcation.org.za [Accessed 02 June 2015].

52 Ibid. 53 Ibid. 54 Ibid. 55 Municipal Demarcation Board, 2014a. Delimitation of wards during 2014/15 for the 2016

local elections .Circular 3/2014. Pretoria: Municipal Demarcation Board. Available from www.demarcation.org.za [Accessed 15 March 2015].

56 Ibid. 57 Sutcliffe, M., 2002. 58 Municipal Demarcation Board, 2014b. 59 Ibid. 60 Ibid. 61 Du Plessis, N.W., 2002. 62 Republic of South Africa, 1998c. 63 Municipal Demarcation Board, 2014a. Delimitation of wards during 2014/15 for the 2016

local elections. Circular 3/2014. Pretoria: Municipal Demarcation Board. Available from www.demarcation.org.za [Accessed 15 March 2015].

64 Republic of South Africa, 1998c. 65 Ibid. 66 Ibid.

32

67 Ibid. 68 Ibid. 69 Ibid. 70 Mouton, J., 2001. Theory, Metatheory and Methodology. In Cotzee, J.K., Graaff, J., Hendricks,

• F. & Wood, G. (eds) Development Theory, Policy and Practice. Cape Town: Oxford University

Press, p10 – 25.

71 Barnard, W.S., 2001. Conceptions of Geography. University of Stellenbosch: Centre for

Geographical Analysis.

72 Dlamini, S., 2015. GIS Specialist, Municipal Demarcation Board, Pretoria. E-mail on 23

September about Ward Delimitation.

73 Municipal Demarcation Board, 2015a. Final decision in respect of the delimitation of ward in

terms of schedule 1 to the Local Government: Municipal Structures Act, 1998. Pretoria: Municipal Demarcation Board. Available from www.demarcation.org.za [Accessed 17 September 2015].

74 Nkomo, S., 2015. Regional Coordinator, Independent Electoral Commission, KwaZulu-Natal.

E-mail on 29 September about Voter Registration.

75 Independent Electoral Commission, 2011. Municipal Elections Report: 2011. Cape Town:

Independent Electoral Commission. Available from www.elections.org.za [Accessed 27 July 2015].

76 Ibid. 77 Ibid. 78 Ibid 79 Kersting, N., 2012, p5 – 21. 80 Sivaramakrishnan, K.C, 2001, p4694 – 4696. 81 Bozkaya, et al. 2003. 82 Gopalan et al, 2013, p477– 489.

33

83 Iturralde, D., 2015. Executive Manager: Demography, Statistics South Africa, Pretoria. E-mail

• n 2 July about Population Estimates.