The previous section has described the methods that were used to collect various types of data from both primary and secondary sources. According to Overton and Diermen (2003), once the researcher had collected the data several techniques can be used to analyse and present the data. Therefore, this section provides explicit perception about the methods that were utilised in this research to analyse and present both quantitative and qualitative data.
The researcher kept the “so what” question in his mind when reviewing the data and this led to some of the data not being used. In addition, analysing the data was aimed to explain and describe the current situation of the Municipality and the new zoning regulations. The researcher reviewed certain research to have an idea about the most appropriate statistical techniques that can be used in this research to analyse the data and how to present the data in different format. Therefore, as Creswell (2003) mentioned the researcher decided to employ descriptive and inferential analysis to analyse the quantitative data and description and thematic text to analyse qualitative data. The analysis strategy in this research is divided as follows (Figure 3.7):
12 The researcher was asked to bring a letter from the university specifying in detail what maps are
needed.
13
Very unusually for Saudi Arabia to have a woman who is the head of department that includes men in the Municipality. Actually Jeddah Municipality is the first Municipality in the Kingdom which promoted a woman to this kind of position.
110
Figure 3.7: Different techniques of analysis. First: Quantitative data analysis
The researcher used tables and different types of graphs (linear and vertical and horizontal bar charts) to illustrate the findings. In addition, quantitative analysis divided into two sections, the first section discusses the analysis techniques for interviews (presented in Chapters 7 and 9) and the second section analyses the empirical data (presented in Chapters 5 and 6), as follow:
Interview analysis
The raw data of interviews was entered as Microsoft Excel for creating charts and SPSS (Statistical Package for Social Science) for analysis. In addition, the researcher rectified the coding for some questions that were asked in a negative voice. A five-point Likert scale was used, where one is „strongly disagree‟ all the way to five which is „strongly agree‟ were fused to a three-point scale (1. disagree, 2. neutral and 3. agree). Similarly, five-point Likert scale was used to gather information about residential satisfaction, where one is „very dissatisfied‟
Qualitative Data Analyses Quantitative Data Analyses Empirical Data Descriptive Statistics Populatio n projection UGI model Structure Interview Inferentia l Statistics Meetings Case Studies Description Thematic Text Description Thematic Text Central Tendency Frequenc y and percentag e Standard Deviation Correlation coefficient ANOVA F test Chi- square Kruskal- Wallis H test
111
all the way to five which is „very satisfied‟, the final analysis uses a three-point scale (1. dissatisfied, 2. neutral and 3. satisfied). This process of data checking and entry for three interviews took time. Though the researcher had personal experience with Microsoft Excel, it was very helpful to learn SPSS software. Once the data entry was finished, analysis and producing charts and tables were easy to present. However, the challenge was which type of analysis method for quantitative data should be used and what useful forms of results would serve the purpose of the research.
The researcher employed descriptive and inferential statistical measurements to analyse and represent the quantitative data. The descriptive statistical measurements included; central tendency (mean and median), frequency and distribution that show the frequencies and percentages of the variables, dispersion measurements (standard deviation) (McCormack and Hill, 1997; Pallant, 2005; Acton et al., 2009). However, in the context of this research, looking at each variable alone, which is known as univariate data, would provide a limited amount of information and would be not sufficient for the research objectives and questions (Alshafiei, 2007). Therefore, the researcher used bivariate data and employed inferential statistical measurements. This included bivariate cross tabulation to provide frequencies and percentages and to examine whether or not there is a significant association between two variables through chi-square test, which provide the probability (P) value (less than 0.001 is high significance, 0.01 to 0.05 is moderate significance and above that is less significance) if P value is not significant then the variables are statistically independent.
In addition, the researcher used a non-parametric test known as Kruskal-Wallis or H test was used to compare the mean rank for each group, in another word, which variable has ranked the highest and lowest mean. Correlation coefficient is another inferential statistical measurement that was used to determine whether or not there is a significant correlation between two variables (see Chapter 9). If the direction is positive (as one variable increase the other one does as well) or negative (as on variable increase the other one decrease) and the strength of the relationship by using Pearson (r) measurement (McCormack and Hill, 1997; Pallant, 2005; Acton et al., 2009). The researcher tried to use regression analysis
112
to measure the level of residents‟ satisfaction. However, if one variable has no value in data, the whole case must be excluded. Thus, the researcher employed equations that were used in Mohit et al. (2010) work to measure the overall residential satisfaction with their dwellings and neighbourhoods (see appendix 5b).
Analysing the empirical data
Statistical data that was collected was used to estimate the projection of the future population of Jeddah and housing demand, where the researcher used three robust techniques to estimate the future population. In the first method, the researcher used a constant annual growth rate, the second method is the linear model and the third technique is the exponential curve projection model (see Chapter 5). Data was also collected to assess the performance of Jeddah Municipality through employing the UN-HABITAT “Urban Governance Index” (UGI) model (see Chapter 6).
According to UN-HABITAT (2004) two alternative sets of indicators have been suggested for the UGI model. The first alternative consists of 18 sub-indicators. The second alternative included indicators with high and moderate ranking, which consists of 25 sub-indicators. According to Narang (2005) the UGI may not always be the best mode of judging the quality of governance given the diversity of contexts. Although, altering the indicators to suit a specific context can lead to a loss of universality and reduce comparability of data, the UGI was not a best fit to cities in Saudi Arabia, because the UGI model looks at the whole governance and this study focuses on spatial planning. Therefore, driven by the desire to obtain a better understanding of context and looking for a robust ways of being able to assess the quality of the local government, the model might need to be altered in this study.
Most of the issues that needed alteration were concentrated in the “Effectiveness” indicator, where one of the main issues that I faced was in understanding three sub-indicators: local government revenue per capita, ratio of actual recurrent and capital budget, and local government revenue transfer. After discussion with the second supervisor (Dr. Graham Tipple) I was advised to write to the UN-
113
HABITAT (see appendix 4). However, no response was forthcoming, therefore, Dr. Tipple and I, agreed to re-identify the terms and modify the equations (Table 3.5). The researcher modified the “Effectiveness” indicator to draw on consumer price index (CPI) to measure the inflation over time and calculate more accurate figures.
Table 3.5: Re-identifying some of the terms and equations in the Effectiveness indicator.
Code
Equation
According to the UN Modification according to the Saudi context
Ratio of recurrent and capital budget (RRC)
RRC = R/C RRC= R/T
Local government revenue transfer (LGT)
LGT= T/R*100 LGT= T/ (TB)*100 According to the UN; R= Recurrent budget includes income derived on a regular basis, C= Capital including fixed income, that is derived after allocation of funds from internal or external sources, T= Income originating from higher levels of government and R= Total local government revenue (transfers and non-transfers. According to the Saudi context; R= Municipality revenue, T= Transfer income or Budget from the Central Government to the Municipality and TB= The Municipality total budget.
Looking at the indicators from the lens and context of Jeddah as a place, some indicators would seem to be more important if they moved and changed to give a sharper picture about what is happening. However, the difficulty was to identify what and which of the sub-indicators or alternative should be used for the model. The researcher recast the UGI looking only at 20 sub-indicators that matter most to planning and subtracting those factors which do not and then reweighting the indicators that remained (Table 3.6) (see Chapter 6).
114 Table 3.6: UGI, selected indicators for three alternatives.
Index Alternative 1: Only high ranking Alternative 2: High and selected moderate ranking Alternative 3: Selected high and moderate ranking
E
ff
ec
tiv
eness
1. Local government revenue per capita
2. Local Government transfers 3. Ratio of mandates to actual tax
collection
4. Published performance standards
1. Local government revenue per capita 2. Ratio of actual recurrent and capital budget 3. Local Government transfers
4. Ratio of mandates to actual tax collection 5. Predictability of transfers
6. Published performance standards 7. Customer satisfaction survey 8. Vision statement
1. Local government revenue per capita 2. Local Government transfers
3. Ratio of mandates to actual tax collection 4. Predictability of transfers E qu it y 4. Citizens charter
5. Proportion of women councillors 6. Proportion of women in key
positions
7. Pro-poor pricing policy
9. Citizens charter.
10. Proportion of women councillors 11. Proportion of women in key positions 12. Pro-poor pricing policy
13. Street vending
5. Citizens charter.
6. Proportion of women councillors 7. Proportion of women in key positions
8. Existence of pro-poor policy for violation and fees P a rt icipa tio n 8. Elected council 9. Election of Mayor 10. Voter turnout 11. People‟s Forum
12. Civic Associations (per 10,000)
14. Elected council 15. Election of Mayor 16. Voter turnout 17. People‟s Forum
18. Civic Associations (per 10,000)
9. Elected council 10. Election of Mayor 11. Voter turnout 12. People‟s Forum 13. Vision statement Acc o un ta bil it y
14. Formal publication contracts, tenders, budget and accounts 15. Control by higher levels of
government
16. Anti-corruption commission 17. Disclosure of personal income and
assets
18. Regular independent audit
19. Formal publication of contracts
20. Control by higher levels of government. 21. Codes of conduct
22. Facility to receive complaints 23. Anti-corruption commission
24. Disclosure of personal income and assets 25. Regular independent audit
14. Formal publication of contracts
15. Control by higher levels of government. 16. Facility to receive complaints
17. Customer satisfaction survey 18. Anti-corruption commission
19. Disclosure of personal income and assets 20. Regular independent audit
115
The score value of the UGI indicators ranged between 0 to 1, where 0 means poor performance and 1 means excellent performance. In addition, all the dichotomous variables (yes or no) transformed into quantitative data and expressed as „No‟= 0 and „Yes‟= 1. Each sub-indicator has a detailed formula to calculate the final score of the sub-indicators. The final score is converted from ratio to percentage. The UGI is conducted through the following formula (UN-HABITAT, 2004: 83):
Urban Governance Index = Average of (Effectiveness sub-index + Equity sub-index + Participation sub-index + Accountability sub-index).
The researcher applied the UGI formulation for both the 25 sub-indicators and the recast 20 indicators. The results of the original and recast sub-indicators of Jeddah Municipality were then compared with 24 selected cities from both developing and developed countries that were analysed by the UN-HABITAT. It should be noted that the results for the 24 selected cities were not recalculated during this research.
Second: Qualitative analysis
Grounded theory (inductive approach) was used to identify the key themes of the research (Bryman, 2004). According to Bryman (2004: 398) grounded theory “is probably the most prominent of the general approaches to qualitative data analysis”. However, one of the difficulties of qualitative research is managing the richness of the data, which includes interviews, meetings and case studies (Bryman, 2004). Therefore, the researcher started by transcribing the recorded tapes and notes after the interviews, meetings and case studies, that were conducted in Arabic into Arabic text then translated them into English. The researcher was cautious during the translation not to change the meaning. Following the translation and transcribing stages the researcher applied a thematic analysis approach and developed a coding scheme to classify and organised the data into specific relevant.
Sophisticated text management software packages such as NVivo are available to do code-based analysis, categorise the qualitative data into themes, to tag, retrieve data efficiently and maintain complex codebooks (Ryan, 2004; Richards, 2009; Ahmed, 2011). However, owing to the lack of time to learn the software, where
116
the researcher had already spent a lot of time on quantitative data, it was necessary to search for an alternative, thus, 'Microsoft Word‟ was used to generate a database to manage and categorise the qualitative data (Sawadsri, 2010 and Ahmed, 2011).