Statistical analysis of the questionnaire data was carried out using IBM SPSS Statistics for Windows
Statistical analysis of the questionnaire data was carried out using IBM SPSS Statistics for Windows, version 20.0. Descriptive statistics and scale frequencies, response percentages, means, modes and standard deviations (SD) were computed for the urban indicators. The demographic data was analysed descriptively by computing percentages and frequencies. Internal consistency reliability was assessed via Cronbach’s alpha (?) coefficient (Cronbach, 1951), which provides a single estimate to determine internal consistency or average correlation of questionnaire items, in order to measure their reliability (Webb et al., 2006). Several studies suggested that the average coefficient of internal consistency (? = 0.70) is considered the threshold of acceptable reliability (Tavakol and Dennick, 2011; Santos et al., 2016). In this study, the average ? coefficient of all extracted components was 0.867, indicating a high internal consistency or reliability (Mourshed and Zhao, 2012).
Principal Component Analysis (PCA) was carried out as a mathematical technique on all 19 indicators to identify the underlying structure, by characterising the items into groups of correlated variables. The importance of each component was assessed by testing scree plots and the contribution of each component to the total variance (;5%). Variance Maximisation (varimax), as an orthogonal rotational strategy, was carried out using the outcomes of the PCA. Rotation can reduce the number of factors on which the variables under investigation have high loadings and makes the interpretation of the analysis easier (Mourshed and Zhao, 2012). Items that were included had factor loadings of more than 0.40. Five components were extracted: economic; cultural; safety and security; design context and housing demand. Following this, internal consistency and validity were established for the questionnaire items. Bartlett’s test of sphericity was utilised to determine significant correlations between survey items while sampling adequacy was computed with Kaiser-Meyer-Olkin (KMO) measure, which was 0.892 for this research. A KMO value greater than 0.8 is considered good and indicates that PCA is useful for the questionnaire variables (Cerny and Kaiser, 1977).