Table 5.1 summarizes the estimation results of access segment. Four regression models which examining the association between BE attributes and walking distance to transit stop within four different buffers are estimated for each segment. A model that represents a buffer is selected based on the following steps including: selecting AIC, analysis of deviance table by Anova, and chiquare test. Based on this result, we can choose four model four each segment as the optimal models to use for interpretation. Note that “Male” is a dummy variable, defined to be 1 if the respondent is male and 0 is female; “Low income”, “High income”, “Middle income” are defined to be 1 if the household income of respondents are low, high, and middle respectively, and 0 otherwise; and “Motorbike driver's license” and “Car driver's license” is defined to be 1 if the respondent has a license and 0 otherwise. “Main transportation mode: motorbike” is define to be 1 if the main transportation mode of respondent is motorbike and 0 otherwise. The conditions of Perceived neighbourhood walkability are defined to be 1 if the point is less than 3 point (poor and very poor) and 0 otherwise.
The estimation results in all models consistently show that “Male” has a significantly positive association with walking distance. This is understandable because male generally have better physical attributes than female, they are able to walk a farther distance to bus stops. This is also the reason for the explanation of
“Young people” and “Senior people”. “Young people” has a positive association with walking distance in model 4, whereas “Senior people” has a negative association.
The dummy variables related to household income have different correlations with walking distance. “Low income” has negative correlation with the walking distance, whereas “Middle income” and “High income” has a positive association, which means that middle and high income people walk to the bus stop farther than low-income people. To explain this, we need to know that most low-income people here are students who come from other provinces and rent a house to live in Hanoi.
They always prefer to select areas near bus stops to facilitate going to their university.
36
Many high-income people live in new urban areas where the density of bus stops is low, and most of them do not choose buses as a major means of transportation.
Next, the estimation results in all models consistently show that “Motorbike driver's license” and “Car driver's licence” have a significantly negative association with walking distance to public transit. This is reasonable because people who have a driver's license tend to use private vehicles more than walking in general and walking to the bus stop in particular. This is also the same reason why “Main transportation mode: motorbike” and “Private vehicle available” have negative association with walking distance to transit stop.
“Total Transfers” has a negative correlation with walking distance. This reasonable because the complexity of the trip could make passengers unwilling to walk too far to make this trip. If the distance to bus stop is long and the trip has many transfer, they can give up and choose other means of transportation.
In terms of perceived neighbourhood walkability, outside of the “Step up and down”, all other dummy variables including “Conflict with other mode”, “Level road”, “Cross the street”, “Drainage”, and “Walking amenities” have the negative association with the walking distance to transit stop. These negative coefficient indicates that in good walking conditions transit users walk farther to bus stop than poor walking conditions. “Step up and down” has a positive correlation with walking distance. This is understandable because when transit users have to walk a lot to go up or down stairs (pedestrian bridge, walkway), their walking distance is also longer.
However, in general, perceived neighbourhood walkability have weak correlations with walking distance.
Finally, BE attributes have the association with walking distance. “Population density” in all four model is estimated to be significantly negative, which means in denser areas transit users walk shorter to public transit. This is reasonable in Hanoi because the public transport bus stops are distributed very much in the central areas of Hanoi, where the population density is very high. That is reason why residents in these areas could easily access the bus stops. It is worth noting that “Job density” is insignificant for walking distance of access segment. “Numbers of bus stop” and “Bus
37
frequency” is estimated to be significantly negative. This means that better accessibility to public transit gives shorter walking distance. This is similar to
“Numbers of public facilities”. Finally, “Land use mix” is estimated to be significantly negative, which suggests that transit users’ walking distance in higher level of land use mix is shorter.
38
Table 5.1: Estimation result of Poisson model for walking distance of access trip
Explanatory variables Model 1 (buffer 100) Model 2 (buffer 200) Model 3 (buffer 500) Model 4 (buffer 1000) Coefficient P > z Coefficient P > z Coefficient P > z Coefficient P > z
Intercept 7.0159*** 0.0000 7.1562*** 0.0000 7.3050*** 0.0000 6.9944*** 0.0000
Male 0.0181** 0.8364 0.0131* 0.8699 0.0344*** 0.6802 0.0002 0.9985
Young people - - - - - - 0.1454*** 0.1299
Senior people -0.1428*** 0.2272 -0.1498*** 0.1918 -0.0795*** 0.4676 - -
Low income -0.1769*** 0.0595 -0.1875*** 0.0386 -0.1813*** 0.0367 - -
Middle income - - - - - - 0.2511*** 0.0074
High income 0.1280*** 0.3784 0.0993*** 0.4720 0.0974*** 0.5002 0.3669*** 0.0197
Motorbike driver's license -0.1015*** 0.2503 -0.0749*** 0.3928 -0.0527*** 0.5254 -0.0859*** 0.3239 Car driver's licence -0.2460*** 0.1105 -0.1961*** 0.1260 -0.2213*** 0.0936 -0.1897*** 0.2286 Main transportation mode:
motorbike -0.2788*** 0.0380 -0.2039*** 0.0984 -0.2314*** 0.0706 -0.2248*** 0.0616
Total transfers -0.0933*** 0.2188 -0.1813*** 0.0177 -0.2482*** 0.0062 -0.1793*** 0.0701 Private vehicles available -0.0917*** 0.3010 -0.0997*** 0.2363 -0.0985*** 0.2281 -0.1124*** 0.1907 Conflict with other mode -0.1838*** 0.0462 -0.1580*** 0.0579 -0.1168*** 0.2269 -0.1176*** 0.2371
Level road -0.0620*** 0.7149 -0.1157*** 0.4886 -0.1562*** 0.3348 -0.1753*** 0.3282
Cross the street -0.1152*** 0.2717 -0.1232*** 0.2118 -0.0941*** 0.3513 -0.1606*** 0.1341
Drainage -0.1650*** 0.2029 -0.1158*** 0.3337 -0.1213*** 0.2780 -0.0739*** 0.5598
Step up and down 0.0509*** 0.6347 0.0532*** 0.5925 0.0586*** 0.5656 0.02203* 0.8387
Walking amenities -0.3444*** 0.0187 -0.4073*** 0.0118 -0.3970*** 0.0594 -0.3415*** 0.0839 Population density -0.0008*** 0.0122 -0.0008*** 0.0240 -0.0012*** 0.0127 -0.0022*** 0.0000
Land use mix -0.3482*** 0.0351 -0.2052*** 0.3515 - - -0.1121* 0.8426
Numbers of intersection -0.1158*** 0.3399 -0.0459*** 0.2412 -0.0160*** 0.0935 -0.0079*** 0.0049
Numbers of bus stop -0.2141*** 0.0034 - - - - - -
Bus frequency - - -0.0015*** 0.0000 -0.0006*** 0.0000 - -
Numbers of public facilities -0.0377*** 0.2954 -0.0238*** 0.1655 -0.0036*** 0.4933 - -
AIC 46461 43430 44946 48630
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
39
Table 5.2: Estimation result of Poisson model for walking distance of egress trip
Explanatory variables Model 1 (buffer 100) Model 2 (buffer 200) Model 3 (buffer 500) Model 4 (buffer 1000) Coefficient P > z Coefficient P > z Coefficient P > z Coefficient P > z
Intercept 6.1230*** 0.0000 6.2409*** 0.0000 5.9554*** 0.0000 6.6699*** 0.0000
Male 0.0957*** 0.3225 -0.0103 0.9139 0.04466*** 0.6520 0.00113 0.9912
Young people - - 0.2136*** 0.0592 0.3788*** 0.0009 - -
Senior people -0.3890*** 0.0154 - - - - -0.5050*** 0.0060
Low income 0.1367*** 0.2055 - - - - 0.1483*** 0.1861
High income 0.1483*** 0.3505 0.0076 0.9635 0.0828*** 0.6297 0.1809*** 0.2843
Motorbike driver's
license -0.0124 0.8920 -0.0324*** 0.7211 -0.0008 0.9927 -0.0034 0.9721
Main transportation
mode: motorbike 0.0985*** 0.5200 0.2007*** 0.1468 0.2190*** 0.1294 0.0855*** 0.5984
Total Transfers -0.2205*** 0.0245 -0.2021*** 0.0313 -0.1665*** 0.0901 -0.1676*** 0.1013 Private vehicles
available 0.1008*** 0.2780 0.1387*** 0.1269 0.0580*** 0.5265 0.0704*** 0.5088
Population density -0.0008*** 0.0459 -0.0006*** 0.1822 0.0003*** 0.5457 - -
Job density -0.0009*** 0.0064 - - - -
Land use mix 0.4139*** 0.0208 -0.0825*** 0.7544 - - -1.1430*** 0.0083
Numbers of intersection -0.1522*** 0.0327 -0.1090*** 0.0006 - - - -
Numbers of bus stop -0.2128*** 0.0016 -0.1563*** 0.0002 - - - -
Bus frequency - - - - -0.0007*** 0.0000 - -
Numbers of public
facilities 0.1096*** 0.0007 0.0488*** 0.0000 0.0094*** 0.0001 0.0021*** 0.0487
AIC 52897 53005 53949 60162
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘’ 1