**2. Methodology**

This section provides a discussion of the methodologies followed in data acquisition for the two components of quantitative research completed. The results are presented for each component in Section 3. As previously mentioned, this paper presents a refinement of and elaboration on the study conducted by Cilliers and Cilliers (2015) [12] investigating the proximity principle in the case study of Potchefstroom using updated municipal valuations in its analysis (See Section 2.1). The paper tested the proximity principle in the local context in 2019, and compared 2019 findings with 2015 findings, to identify if the proximity principle would still be rejected, as in the 2015 study. The 2019 study was further substantiated with supporting quantitative data collected through two structured questionnaires capturing perceptions amongs<sup>t</sup> a sample of Potchefstroom residents pertaining to green space values; and a questionnaire distributed amongs<sup>t</sup> Professional Planners in South Africa regarding green space planning in the local context (See Section 2.2). Although these questionnaires were limited in sample size, and should be expanded on in future research endeavours, results provide preliminary insights into community and professional perspectives on UGSs in support of the importance of context and community input discussed in Section 1.4.

#### *2.1. Data Acquisition Part A: Proximity Principle in Potchefstroom*

The data were collected in Potchefstroom (26.7145◦ S, 27.0970◦ E), situated in the North-West province of South Africa. Potchefstroom was chosen as the local case study area as international literature were divergent from the results of a study conducted in Potchefstroom by Cilliers and Cilliers in 2015 [12]. This paper refined the previous study by Cilliers and Cilliers, employing the same methods of data collection and data analysis to verify the results study four years later. As such, the same residential areas and respective properties within each residential area included in the 2015 study by Cilliers and Cilliers [12] were reselected in 2019 to determine the impact of green spaces on property value, using 2019 municipal property valuations. These residential areas were originally selected based on their proximity and accessibility to UGSs, and included the following, as seen in Figure 1:


**Figure 1.** The five selected residential areas in Potchefstroom.

Within each residential area, the selected properties were divided into three zones to test the proximity principle, as illustrated in Figure 2. Zone 1 bordered a UGS, whereas zone 3 was located furthest away from the green the same UGS. Firstly, the property price per square meter was determined by dividing the municipal property values of June 2019 with the total area (in square meters). Thereafter, the mean square meter value per zone in each selected residential area was calculated. Therefore, in each selected residential area, zone 1, zone 2 and zone 3 were assigned a mean ZAR per square meter value that was used to statistically analyse the data. As verified in the international literature, the proximity principle suggests, that zone 1 should present the highest property value [12]. The data obtained from the 2019 Potchefstroom case study were accordingly analysed using three analytical methods including the Analysis of Variance (ANOVA), the Kruskal–Wallis analysis and the Dependent T-test.

**Figure 2.** Zone 1 to 3 within area B, Van der Hoff Park.

#### *2.2. Data Acquisition Part B: Community Survey and Professional Planner Survey*

As a refinement on the study conducted by Cilliers and Cilliers [12] and gain further insights on the findings derived from the quantitative investigation, two additional surveys were conducted. The data were collected through two structured questionnaires employing a Likert scale (1 = Fully agree, 2 = Agree, 3 = Not sure, 4 = Disagree and 5 = Fully disagree). The first questionnaire, referred to as the Potchefstroom resident survey, focussed on a sample of Potchefstroom residents and their perceptions and appreciation of green space benefits in Potchefstroom. A total of 74 residents completed the survey. The aim was to understand how communities perceive open spaces in Potchefstroom, as the primary users of these spaces. Data were statistically analysed using Cramer's V test and cross-tabulations of selected questions. The second questionnaire focussed on the perspectives of a sample of Professional Planners relating to UGS planning in the context of South Africa. The Professional Planner questionnaire focused on green space aspects including the available financial resources (local budgeting) in green space management, community engagemen<sup>t</sup> in green space planning, environmental considerations in practice and green space typologies. A total of 26 planners completed the survey where 17 were Professional Planners, 8 were Candidate Planners and 1 was not registered as a planner, but did work in planning practice. Both questionnaires were distributed electronically, and consent was granted by virtue of completion of a questionnaire.

#### **3. Data Analysis and Results**

#### *3.1. Data Results Part A: Proximity Principle in Potchefstroom*

A verdict on the rejection or acceptance of the proximity principle can be made solely based on the mean ZAR per square meter values obtained for the 3 zones within each residential area; however, the statistical evidence provides credence. In observing the 2019 mean ZAR per square metre for each zone in each area the following could be identified. The mean ZAR per square meter values in four of the five residential areas (Grimbeek Park, Van der Hoff Park, Potchefstroom Dam Area and Heilige Akker) indicated that zone 1 represented the lowest value in ZAR per square meter compared to zone 2 and zone 3. Thus, zone 1 that is located closest to the UGS presented the lowest mean ZAR per square meter value instead of the highest value as suggested by the proximity principle. Zone 3 displayed the highest mean ZAR per square meter value in three of the five residential areas (Grimbeek Park, Van der Hoff Park and Potchefstroom Dam area), the zone furthest away from the green space.

By using the municipal property values of 2019, the mean ZAR per square meter values of each zone within each selected residential area were calculated as mentioned in the previous section. The latter values were used to complete the ANOVA (mean1 – mean 2/max SD, also referred to as the "standardized difference between the means" and Kruskal–Wallis analyses (Z/sqrt(N). The effect sizes presented by the ANOVA and Kruskal–Wallis analyses were used to verify the significance of the results, as the sample sizes did not provide enough power to test for normality, requiring both parametric and the non-parametric tests. Where results differ, the non-parametric test was preferred.

The effect sizes obtained determined whether a practically significant difference was present between zone 1 and zone 2, as well as between zone 1 and zone 3 within each residential area. The effect sizes obtained from analyses were interpreted. A small effect size of 0.2 indicated no practically significant difference, whereas a medium effect size of 0.5 indicated a practically visible difference. A large effect size of 0.8 indicated a practically significant difference [39].

The results showed that the comparison between zone 1 and zone 2 in the respective residential areas delivered an overall medium effect size, thus, a practically visible difference between the mean ZAR per square meter values (≈0.5), as seen in Table 3. Four of the five residential areas indicated a practically visible difference (≈0.5) (Van der Hoff Park, Potchefstroom Dam Area, Heilige Akker and Oewersig), while only one residential area indicated a practically significant difference (≈0.8) (Grimbeek Park). The effect sizes produced by comparing zone 1 and zone 3 in the respective residential areas delivered an overall visible (≈0.5) to a significant difference (≈0.8) (Table 3). Two residential areas presented a practically significant difference (≈0.8) (Grimbeek Park and Van der Hoff Park) and another two residential areas delivered a practically visible difference (≈0.5) (Potchefstroom Dam Area and Oewersig). Only one residential area presented no practically significant difference between zone 1 and zone 3 (≈0.2) (Heilige Akker).



The effect sizes indicated an overall medium to large effect size (≈0.5/≈0.8), indicating a visible difference to practical significant difference between the mean ZAR per square meter values of zone 1 and the zones further away from the green space. Zone 1 also presented values lower than the zones further away from the green space, refuting much of the existing research on the positive impact of UGS on property value. The proximity principle was thus rejected, with the statistical analyses supporting the data showing a medium to large differences between values. Thus, in the case study a more proximate location to a UGS did not indicate an increase in property value, but showed an inverse effect.

The data from Cilliers and Cilliers (2015) [12] were statistically compared to the data obtained in the 2019 Potchefstroom case study, as captured in Table 4. The effect sizes obtained from the Dependent T-test, once again, indicated whether a practically significant difference existed between the data. The effect sizes presented a practically significant difference between the mean ZAR per square meter values (effect size of ≈ 0.8/0.5). The latter was clear by observing the old and new mean ZAR per square meter values in Table 4. Property values increased between 2015 and 2019 by 68% in Grimbeek Park, 35% in Van der Hoff Park, 55% in the Potchefstroom Dam area, 46% in Heilige Akker and 45% in Oewersig, thus presenting an aggregate increase on average of almost 50%. Various potential factors could have led to the increase in property values identified from 2015 to 2019. Inflation was a potential factor as the Consumer Price Index (CPI) increased from 5.78% in 2013 to 6.59% in 2016 and decreased, thereafter, to 4.25% in 2019 [40]. Other factors influencing property value in South Africa, according to the Absa residential property market database, include migration trends, security issues, income levels, employment, monetary and fiscal policies, investment returns, the condition of the property and foreign property buying of South African properties [41].


**Table 4.** Dependent T-test statistical analysis.

#### *3.2. Data Results Part B: Community Survey and Professional Planner Survey*

#### 3.2.1. Potchefstroom Resident Survey

The results obtained from the survey indicated that the sample of Potchefstroom residents recognise the social, environmental and economic value of green spaces; however, fewer residents recognise the economic value of green spaces. Question 4 focused on the residents' perceptions of green spaces in Potchefstroom, referring to safety matters. A total of 52% of respondents agreed that green spaces in Potchefstroom are perceived as crime hotspots, thus, contributing to unsafe neighbourhoods and indicating a related ecosystem disservice (cross-reference to Section 2.2). A total of 60% of respondents agreed that they would pay more for a property that is located next to a green space in Question 5; however, many residents were unsure or disagreed. Interestingly, some residents

who perceived green spaces as crime hotspots in Question 4, still agreed that they would be willing to pay more for a property that is located next to green space (Figure 3).

**Figure 3.** Potchefstroom resident survey results. Question 1: Urban green spaces have environmental value; Question 2: Urban green spaces have social value; Question 3: Urban green spaces have economic value; Question 4: Urban green spaces are perceived as crime hot spots in Potchefstroom; Question 5: I would pay more for a property because it is located next to an urban green space.

As a result of the answers to Question 4 and 5, in recognition of the ecosystem disservices linked to the South African and Potchefstroom contexts, a cross-tabulation were conducted as part of the analysis, in an attempt to further clarify findings: Question 4 (Urban green spaces are perceived as crime hot spots) was cross-tabulated with Question 5 (I would pay more for a property because it is located next to an urban green space). The cross-tabulation was completed with 74 valid cases and a medium practical significant difference as the Cramer's V test value was V = 0.287 (V≈0.3) was presented. Thus, findings supported the observation that although certain residents perceived UGSs as crime hotspots (ecosystem disservices), they residents are still willing to pay more for a property is located next to an UGS.

#### 3.2.2. Professional Planner Survey

The results of Question 1 and Question 2, as shown in Figure 4, both indicated that 88% of the planners agreed that unattractive green spaces are due to a lack of maintenance by local authorities and a lack of community engagement. Question 3 delivered interesting results as 50% of respondent planners agreed that environmental considerations are not prioritised in the planning process; however, the other 50% were either unsure or disagreed that environmental considerations are not prioritised in the planning process Question 4 of the survey focussed on local budgeting for green space planning and indicated that only 62% of respondents agreed with the statement that green spaces are not prioritised in local budgeting. A total of 38% of the planners were either unsure or disagreed that green spaces are not prioritised in local budgeting. The majority of planner respondents reported being familiar with green space typologies (92%).

**Figure 4.** Professional Planner survey results. Question 1: Unattractive urban green spaces are the result of a lack of maintenance by local authorities; Question 2: Unattractive urban green spaces are the result of a lack of community engagement; Question 3: Environmental considerations are not prioritised in the planning process; Question 4: Environmental considerations are not prioritised in local budgeting; Question 5: I am familiar with green space typologies.
