**5. The statistical analysis**

The statistical analysis examines the extent to which the indicators of the physical environment are able to explain the variance among the walking outcomes of

*Approaching Urban Design through the Analysis of Structural Differences within Three… DOI: http://dx.doi.org/10.5772/intechopen.87221*

the individuals. The measured attributes of the physical environment were tested in terms of their predictability for the walking minutes. Thus, their variables are considered predictors (X), which need to be tested in terms of their predictability for the walking minutes (Y). In other words, this analysis tests whether the walking outcome variables can be individually predicted by the measured attributes of the physical environment. For such a purpose, the hierarchal regression analysis was chosen because of its flexibility to enter predictors in a split block with extra predictors. Moreover, the *p*-value (<0.05) indicates the significance of the models, while the determination coefficient (*R*<sup>2</sup> ) explain the potential of the predictors to explain the variance of the outcomes. The hierarchical regression analyses were run to test the predictability of the physical environment indicators, which were individually tested with the walking minutes in one model for each indicator to determine the effect significance of the predictor on the outcome variable (*p*-value) and the (*R*<sup>2</sup> ). Moreover, the SPSS software was utilized for the conducting all the required statistical analysis in this study.

#### **5.1 Findings**

*Sustainability in Urban Planning and Design*

**380**

**Figure 7.**

**Figure 6.**

**Figure 8.**

*Al-Mugawlen edges assessment.*

*Al-Saymmar edges assessment.*

*Al-Abassya edges assessment.*

(**Table 1**). The variables required to apply the indicator are the width of the street and the heights of the adjacent buildings, which were measured directly in this study. Then, the function of the indicator was applied to the sections of all the sampled streets segments. In terms of the main streets (EnRBS1M), the three case studies have a broadly similar value of enclosure ratios, at 2.9, 2.7, and 2.7, for Al-Saymmar, Al-Mugawlen, and Al-Abassya, respectively. In terms of the Connecting streets with the green level of betweenness (EnRBS1C), the highest score was noted with the Al-Mugawlen neighborhood (1.7); Al-Abassya showed a moderate enclosure ratio level (1.3), and the lowest level was noticed within Al-Saymmar (1). In terms of the local streets and the blue level of betweenness (EnRBS1CS), the highest score was noted within the Al-Mugawlen neighborhood (1.7); Al-Saymmar showed a moderate enclosure ratio level (1.1), and the lowest

The statistical analysis examines the extent to which the indicators of the physical environment are able to explain the variance among the walking outcomes of

level was noticed with Al-Saymmar (0.93) (**Table 2**).

**5. The statistical analysis**

The Block density, the predictability of the model was significant (*F*(173, 1) = 16.989, *p* < 0.001, *R*<sup>2</sup> = 0.089). Thus, the singular model was able to explain 8.9% of the variance of the total walking minutes. Also, the test shows that the higher walking behavior outcome score was associated with higher block densities (*b* = 11.817, *p* < 0.001). In term of Housing Units density, the predictability of the model was significant (*F*(173, 1) = 26.231, *p* < 0.001, *R*<sup>2</sup> = 0.132). Thus, the model was able to explain 13.2% of the variances of the total walking minutes. It was showed that the higher scores for walking behavior outcomes were associated with higher housing units densities (*b* = 0.040, *p* < 0.001).

Diversity of all commercial land use on a 400-m radius scale; the predictability of the model was significant (*F*(173, 1) = 11.145, *p* < 0.001, *R*<sup>2</sup> = 0.061). Thus, the singular model was able to explain 6.1% of the variances of the total walking minutes. Furthermore, the higher walking behavior outcome scores were associated with a higher diversity of all commercial land uses (*b* = 2.478, *p* < 0.001). Regarding the commercial land use without parking, wholesale, and workshops variable on a 400-m radius scale, the predictability of the model was significant (*F*(173, 1) = 20.176, *p* < .001, *R*<sup>2</sup> = 0.104). Thus, the model was able to explain 10.4% of the variances of the total walking minutes. Furthermore, the higher scores of walking behavior outcomes were associated with the higher diversity of commercial land use without parking, wholesale, and workshops (*b* = 2.390, *p* < 0.001). In term of the diversity of the non-residential land use on a 400-m radius scale, the predictability of the model was significant with the total walking minutes (*F*(173, 1) = 13.947, *p* < 0.001, *R*<sup>2</sup> = 0.075). Thus, the model was able to explain 7.5% of the variances of the total walking minutes. Also, the higher of walking behavior outcomes scores were associated with a lower diversity of nonresidential land use (*b* = −3.390, *p* < 0.001).

Regarding the connectivity indicators; the node (streets intersections) intensity on both scales 400-m and 600-m radius, the predictability of the model was significant (*F*(173, 1) = 26.940, *p* < 0.001, *R*<sup>2</sup> = 0.135) and (*F*(173, 1) = 18.678, *p* < 0.001, *R*<sup>2</sup> = 0.097), respectively. Thus, the models were able to explain 13.5 and 9.7% of the variances of the total walking minutes on the two scales, respectively. However, the higher walking behavior outcome scores were associated with the higher node intensities only on the 400-m radius scale in terms of the total walking minutes (*b* = 0.573, *p* < 0.001). In term of the street intensity on 400-m and 600-m radius scales, the predictability of the two models was significant for the total

walking minutes (*F*(173, 1) = 27.071, *p* < 0.001, *R*<sup>2</sup> = 0.135) and (*F*(173, 1) = 13.331, *p* < 0.001, *R*<sup>2</sup> = 0.072), respectively. Also, the models were able to explain, 13.5 and 7.2% of the variances of the total walking minutes, respectively. Moreover, the higher walking behavior outcome scores were associated with a higher street intensity on a 400-m radius scale (*b* = 0.009, *p* < 0.001) and on a 600-m radius scale (*b* = 0.005, *p* = 0.001).

The Pedestrian Catchment Area (PCA) on a 400-m radius scale; The predictability of the model was significant with the total walking minutes (*F*(173, 1) = 14.914, *p* < 0.001, *R*<sup>2</sup> = 0.079). Also, the singular model was able to explain 7.9% of the variances of the total walking minutes. Furthermore, the higher walking behavior outcome scores were associated with the lower value of the Pedestrian Catchment Area (*b* = −0.059, *p* < 0.001). Also, Pedestrian Route Directness Ratio (PRDRS1 and 2) on 400-m and 600-m radius scale; the predictability of the two models was nonsignificant for the total walking minutes were (*F*(173, 1) = 0.301, *p* > 0.05, *R*<sup>2</sup> = 0.002) and (*F*(173, 1) = 0.142, *p* > 0.05, *R*<sup>2</sup> = 0.001). Similarly, the models were inconsiderably explained the variances of the total walking minutes, at 0.2 and 0.1%, respectively. Also, the higher walking behavior outcome scores were not associated with the Pedestrian Route Directness Ratio (*b* = 1.231, *p* > 0.05). Regarding the Clustering coefficient of destinations on a 400-m and 600-m radius scales, the predictability of the two models were nonsignificant for the total walking minutes (*F*(173, 1) = 0.142, *p* > 0.05, *R*<sup>2</sup> = 0.001) and (*F*(173, 1) = 2.288, *p* > 0.05, *R*2 = 0.019). Also, the models were able to explain, 0.1 and 1.9% on the two scales, respectively.

Frontage quality index of the streets on the 400-m radius scale: the predictability of the three models of the main, Connecting and col-de-sac streets was significant for the total walking minutes (*F*(173, 1) = 26.427, *p* < 0.001, *R*<sup>2</sup> = 0.133), (*F*(173, 1) = 26.586, *p* < 0.001, *R*<sup>2</sup> = 0.133) and (*F*(164, 9) = 4.374, *p* > 0.05, *R*<sup>2</sup> = 0.211), respectively. However, the higher walking minutes were marginally associated with the lower frontage quality index of the main, the connecting and the col-de-sac streets (*b* = −0.110, *p* < 0.001), (*b* = −0.145, *p* < 0.001) and (*b* = −0.097, *p* < 0.001), respectively. Regarding the enclosure ratio of the streets on the 400-m radius scale, the predictability of the three models of the main and Connecting streets was significant for the total walking minutes (*F*(173, 1) = 20.840, *p* < 0.001, *R*<sup>2</sup> = 0.108), (*F*(173, 1) = 49.636, *p* < 0.001, *R*<sup>2</sup> = 0.223), respectively. While, in term Enclosure ratio of the col-de-sac street on the 400-m radius scale, the predictability of the model was not significant for the total walking minutes (*F*(173, 1) = 0.428, *p* > 0.05, *R*2 = 0.002). However, the higher walking minutes were marginally associated with the lower frontage quality index of the main, the connecting and the col-de-sac streets (*b* = 3.720, *p* < 0.001), (*b* = −0.615, *p* < 0.05) and (*b* = 0.103, *p* > 0.05), respectively.
