Page 102 - Kỷ yếu hội thảo khoa học quốc tế - Ứng dụng công nghệ mới trong công trình xanh , lần thứ 8

th
HỘI THẢO KHOA HỌC QUỐC TẾ ATiGB LẦN THỨ TÁM - The 8 ATiGB 2023 85

The correlation between the independent variables choosing motorcycles. Motorbikes are still the
have been examined. Most variables have Spearman preferred mode of transportation for students.
correlation coefficients lower than 0.50, indicating - The travel time of pedestrians and buses is
weak correlations. However, the Spearman statistically significant with a coefficient 'p' less than
correlation coefficient between the 'Dis' variable and 0.05. The travel time variable coefficients indicate
the 'Time' variable is 0.79 as well as the Spearman that as time increases, the probability of students
correlation coefficient between the 'Dis' variable and choosing to walk, bike decreases by 0.35 and 0.09
the 'Cost' variable is 0.90 with a significance value respectively. However, in this context, choosing the
less than 0.05, suggesting the possibility of bus is still a positive option for students. It is clearly
multicollinearity. Therefore, the Variance Inflation that the longer the comute time, the more likely
Factor (VIF) was employed to assess the extent of students are to choose the bus.
multicollinearity. The results reveal that the VIF of
the 'Dis' variable and the 'Cost' variable is 5.8 and 6.8 The confidence interval that is used in this study is
respectively, which is greater than the VIF of the 95%. The model shows significant results at a
'Time' variable (VIF = 3.35), leading to exclude the confidence interval of 95% and insignificant results as
'Dis' variable and the 'Cost' variable. shown in Table 6. The insignificant results are kept
for comparison, as described later.
Table 4 - Model Fitting Information provides
statistics used to assess the overall model fit. The The location of the school, Gender, and
'Final' row presents information about whether all Motorcycle ownership are included in the model,
model coefficients are equal to 0 (i.e., whether any where each one of them has a different impact on
coefficients have statistical significance). McFadden transport modes.
demonstrates that the model created possesses high - The location of the school has an impact on the
predictive capabilities, with a McFadden R2 value of transport mode choice, UTE students typically prefer
0.67 (McFadden 1974). The pseudo R-squared value to use all three modes of transportation mentioned
is nearly similar to the squared R in linear regression. here rather than using motorbikes, with walking being
It can be seen that the p-value is < 0.001 representing the most commonly chosen mode, especially with the
a degree of freedom of 42, which means that the full highest coefficient of 5.04. However, because UTE
model's predictions have statistical significance in students account for the majority of the survey results,
predicting the dependent variable better than the this result needs to be verified.
intercept-only model.
- In general, the results of the similarity analysis
The likelihood ratio tests of every variable’s effect for female students show that they prefer to use the
on the developed model was presented in Table 5. It remaining 3 means of transport (-4.6, -3.76 and -3.5,
can be observed that most variables have statistical respectively) rather than motorcycles, especially
significance at a confidence interval of 95% (meaning walking.
they have an impact on the dependent variable),
except for the "Rider License" (p = 0.831 > 0.05) and - As predicted, motorcycle ownership holds great
the "Part-time Jobs" variable (p = 0.309 > 0.05) significance. Students owning motorcycles are much
more likely to commute by motorcycle than by
The developed model's characteristics can be walking or cycling to school (3.46 and 2.65
found in Table 6, which provides estimations for each respectively). It is not surprising that the likelihood of
parameter coefficient associated with various using personal transportation is significantly higher
transportation modes. The reference mode is for students who own motorcycles. In fact, in
motorcycle mode, with the second column denoting developing countries, many people heavily rely on
the parameter's name and "estimate" indicating the motorcycles due to their mobility, survival capability,
parameter's coefficient. In the third column, the and accessibility.
standard error (Std. Error) was shown, "Sig." signifies
the significance value, commonly known as the p- - Students living with family are less likely to
value. The research yields several conclusions as walk to school than students living in dormitories or
follows: rented houses. Meanwhile, students with low living
costs choose to walk more than their peers with
- In the case of intercept only (i.e., without higher costs.
considering the independent variables), the analysis
results indicate that the coefficients related to the - Students living with family are more likely to
choice of walking, bicycles and buses are both use motorbikes to get to school than students living in
negative (-11.08, -17.55 and -11.13 respectively). dormitories or rented houses. Meanwhile, students
This implies that the probability of students choosing with low living costs choose to walk more than their
walking, bicycles and buses is lower compared to peers with higher costs.

ISBN: 978-604-80-9122-4