You are on page 1of 10

Suman, Bolia, Tiwari

USER PROFILE, TRIP PROFILE, AND PERCEPTION OF BUS COMMUTERS IN DELHI

Hemant K. Suman1, Nomesh B. Bolia2, Geetam Tiwari3

ABSTRACT
A large number of working people travel by bus in Delhi. At present 41% of total vehicular trips are made by bus;
however, the mode share of bus trip has reduced from the 60% observed in 2001. If high share of bus trips has to be retained
in future, specific policy interventions may be required based on the needs and perceptions of commuters. The
socioeconomic profile, trip profile, and perception of current bus commuters regarding bus fare, safety, security, directness,
accessibility, crowdedness, and punctuality can significantly help in framing the future guidelines for bus systems. In this
study, a survey of current bus commuters is conducted and revealed that 18%, 38%, and 9% belong to a car, motorcycle,
and bicycle owing household respectively, and 47% do not own any vehicle at household. Multiple linear regression is
performed and results show that student commuters travel shorter distances compared to others. Responses with strong
statistical significance reveal that bus fare, safety, security, directness, and accessibility are attributes that do not inhibit the
use of buses. Responses also reveal with a strong statistical significance that crowdedness is an attribute that inhibits them
to use buses. Further, a Multinomial Logistic regression based analysis reveals that the buses are relatively more safe &
secure for male commuters as compared to female counter part.

KEY WORDS: User profile, Trip profile, Perception of captive users, Sign test, Multiple linear regression, Multinomial
Logistic regression, Guidelines for future bus system.

1. INTRODUCTION
Bus transportation system accounts for a major share of total vehicular trips made in India. The mode share of bus trips
varies from city to city but there is no doubt regarding the importance it plays in terms of the number of passengers it
carries in Indian cities. However, the number of cars and two wheelers has been rapidly growing in all cities in the last
decade. Motorised two wheelers have been growing at a higher rate than cars in most cities. In Delhi, cars have been
growing with much higher rates as compared with two wheelers (1). In the national capital territory (NCT) of Delhi,
vehicular growth rate has increased from 4.72% in 1999-2000 to 7.27% in 2011-12. Delhi has the highest number of
registered motorised vehicles (7.35 million in 2011-12) as compared to other cities in India. Vehicle ownership is also high
with 30% people owing two-wheelers and 16% owing cars (2) & (3). The bus mode share has been decreasing in Delhi in
the last decade. Recent studies show the number of bus trips has increased by 22% but bus mode share has decreased by
17% from 2001 to 2008 (4) & (5). Delhi government has taken several measures to improve the bus system. These include
augmenting bus fleets, introduction of low floor and AC buses, ramps on both sides of the bus queue shelters, and intelligent
transport systems (ITS) (6) & (7). However, no detailed data is available to understand the perceptions of current bus users.
This paper presents the captive users socioeconomic profile, trip profile, and perception regarding bus fare, safety, security,
directness, accessibility, crowdedness, and punctuality. To the best of our knowledge, this is the first such comprehensive
study over bus passenger in Delhi. The findings can significantly help in framing future guidelines for bus systems.

This paper is structured as follows: Section 2 includes design of survey instrument and data collection
methodology. Section 3 includes socioeconomic profile of bus commuters. Section 4 includes trip profile of bus commuters.
Section 5 includes the perception of bus commuters. Section 6 includes discussion of results and section 7 concludes the
paper with recommended guidelines for future bus system and provides the direction for possible extension of this work.

1
Department of Mechanical Engineering, Indian Institute of Technology Delhi, Hauz Khas, New Delhi, India-110016
Tel: +918447632535; Fax: +91-11-2658-2053; Email: hemantsmn@gmail.com
2
Department of Mechanical Engineering, Indian Institute of Technology Delhi, Hauz Khas, New Delhi, India-110016
Tel: +91-11-2659-1141; Fax: +91-11-2658-2053; Email: nomesh@gmail.com
3
Department of Civil Engineering, Indian Institute of Technology Delhi, Hauz Khas, New Delhi, India-110016
Tel: +91-11-2659-1047; Fax: +91-11-2685-8703; Email: geetamt@gmail.com

TRB 2015 Annual Meeting Paper revised from original submittal.


Suman, Bolia, Tiwari
2. DESIGN OF SURVEY INSTRUMENT AND DATA COLLECTION METHODOLOGY
Final survey instrument was designed based on the experience of a pilot survey conducted on four bus routes in NCT of
Delhi. The pilot survey had three sections; section one corresponded to user profile, section two corresponded to responses
related to perception about bus transport and section three corresponded to trip profile of current bus users (8). Rastogi and
Rao (9) suggest that the questions related to user profile should be kept at last, because most non-responses in pilot survey
were observed for user profile. For collecting the perception of bus commuters, about seven constructs (i.e. bus fare, safety,
security, directness, accessibility, crowdedness, and punctuality) that are major issues related to bus system have been
included. Every construct was a function of more than one attribute, and two to four important attributes corresponding to
each construct were considered for the pilot survey. Pilot survey was conducted in December 2013 and 86 responses
collected on four routes. The most important attribute was selected in each construct based on the highest average score in
pilot survey and that was considered for the main survey. The respondents in pilot survey were observed to lose interest in
answering questions after three minutes. Therefore, survey format was modified such that it could be completed within
three minutes. As pointed out by Peter Stopher (10), the problems of non-responses, unwillingness towards survey, and
misinterpreting the questions are common during survey. The problem of non-responses is resolved by designing a short
and compact survey instrument. The problem of unwillingness towards survey is taken care by providing an incentive in
the form of a chocolate, and including the term ‘study’ in place of ‘survey’ in the final survey instrument as recommended
by Peter Stoper (10). The outcome of providing an incentive is remarkable: every bus commuter responded to the survey.
Whereas the non-responses rate in the pilot survey conducted without the incentive was greater than 50%. The problem of
misinterpreting the questions is resolved through filling the questionnaire by surveyors during face-to-face interview,
among captive users by going to bus stops and bus routes all over Delhi, and making all closed questions (10). Therefore,
every respondent interpreted the questions in the same manner.

3. SOCIOECONOMIC PROFILE OF CURRENT BUS COMMUTERS


Out of the total bus, commuters 18%, 38%, and 9% belong to a car-owing, motorcycle-owing and bicycle-owing household
respectively, and 47% belong to no vehicle-owing household. There is a significant difference in the profile of frequent
and infrequent bus commuters. Among frequent commuters, 16%, 35%, and 8% belong to a car-owing, motorcycle-owing
and bicycle-owing household respectively (50% do not have any vehicle at household) and among infrequent commuters,
23%, 42%, and 11% belong to a car-owing, motorcycle-owing and bicycle-owing household respectively (40% do not have
any vehicle at household). Similarly, the percentages of the commuters fall in different categories are shown in Table 1.

Table 1: Percentage of Frequent and Infrequent commuters in each category


Categories Frequent Commuters (%) Infrequent commuters (%)
Female 13 13
Gender
Male 87 87
≤ 25 yrs 45 41
25 to 40 yrs 32 39
Age
41 to 55 yrs 16 13
≥ 55 yrs 7 7
≤ Rs10,000 35 33
Rs10,000 to Rs30,000 42 37
Income
Rs30,000 to Rs50,000 14 16
≥ Rs50,000 9 14
Student 31 26
Occupation Service 43 36
Others 26 38
Car 16 23
Motorcycle 35 42
Vehicle ownership
Bicycle 8 11
None 50 40

Dargay (11) has observed that the income and vehicle ownership are highly correlated and our analysis obtained
the same. The Pearson correlation coefficient between income and vehicle ownership is 0.446. Further, in the linear
regression model of vehicle ownership vs. income the coefficient of income is statistically significant (p-value = 0.000) in
the model. Thus, it can be concluded that, there is a relationship between income and types of vehicle owned. The
relationship to find out, what are the likelihood of type of vehicle with income category is shown in Table 2. The commuters

TRB 2015 Annual Meeting Paper revised from original submittal.


Suman, Bolia, Tiwari
with household income Rs 10,000-30,000, Rs 30,000-50,000, and more than Rs 50,000 are 5 times, 26.6 times, and 44
times respectively, more likely to have car with respect to non-motorised vehicle (NMV) owners when compared with the
income segment of less than Rs 10,000 from Table 2, (values of Exp(B) for car are 44.13, 26.59, and 4.973). Similarly, the
likelihood of having motorcycle with respect to NMV are 3.3 times, 3.7 times, and 4.6 times more likely for commuters
having income Rs 10,000-30,000, Rs 30,000-50,000, and more than Rs 50,000 respectively when compared with
commuters having income less than Rs 10,000 (values of Exp(B) for motorcycle are 4.648, 3.736, and 3.216).

Table 2: Likelihood of Vehicle Owned With Income Category


Ownershipa B Std. Error Wald dfc Sig. Exp(B)
Car Intercept -3.12 0.264 140.21 1 0
Income (more than Rs 50000/month) 3.787 0.338 125.63 1 0 44.13
Income (Rs 30000 to 50000/month) 3.28 0.318 106.5 1 0 26.59
Income (Rs 10000 to 30000/month) 1.604 0.302 28.161 1 0 4.973
Income (less than Rs 10000/month) 0b . . 0 . .
Motorcycle Intercept -1.43 0.123 134.26 1 0
Income (more than Rs 50000/month) 1.536 0.266 33.314 1 0 4.648
Income (Rs 30000 to 50000/month) 1.318 0.226 34.094 1 0 3.736
Income (Rs 10000 to 30000/month) 1.168 0.155 56.668 1 0 3.216
Income (less than Rs 10000/month) 0b . . 0 . .
a
The reference category is: Non-motorised vehicle (NMV)
b
This parameter is set to zero because it is redundant.
c
Degree of freedom

4. TRIP PROFILE OF CURRENT BUS COMMUTERS


There is not a significant difference in trip profile of bus commuters owing car, motorcycle, and cycle at household.
However, the trip length of student commuters having bicycle was 5.8 km that is shorter than overall trip length i.e. 13.1
km. Overall, the average trip length of student and young commuters are shorter as compared to others. The result of the
Multiple Linear Regression shows that there is a relationship between trip length and user profile. We have used Multiple
Linear Regression because our dependent variable is continuous and independent variables are categorical (12) & (13).
The details of the explanatory variables are shown in Table 3.

Table 3: Details of Dependent and Independent Variables Used in Multiple Linear Regression
S.N. Name of Type of Nature of Coding of variables
variables Variables variables
0. Trip length Dependent Continuous Not applicable
1. Gender Independent Categorical ‘0’ for female & ‘1’ for male
2. Age Independent Categorical ‘0’ for > 40 yrs & ‘1’for ≤ 40 yrs
3. Income Independent Categorical ‘0’ for > Rs 30000 &‘1’for ≤ Rs 30000
4. Car Independent Categorical ‘0’ for No & ‘1’ for Yes
5. Student Independent Categorical ‘0’ for No & ‘1’ for Yes
6. Service Independent Categorical ‘0’ for No & ‘1’ for Yes

Table 4 shows that the coefficient of at least one of independent variable is non-zero because p-value is less than
0.05 (overall model) and the null hypothesis (i.e. the coefficient of all independent variables are zero) is rejected. Thus, the
trip length is a function of independent variables.

Table 4: Overall fit of the model


ANOVA
Model Sum of Squares Degree of Freedom Mean Square F Sig.
1 Regression 3416.275 6 569.379 6.617 .000
Residual 96118.878 1117 86.051
Total 99535.153 1123
Dependent Variable: Trip length
Predictors: (Constant), Gender (1), Age (1), Income (1), Car (1), Student (1), Service (1)

The coefficients of independent variables are shown in Table 5 and trip length is defined by the equation1:
Trip length = 14.05 + 0.31*Gender + 0.23*Age + 0.95*Income + 0.55*Car - 4.47*Student – 1.91*Service (1)

TRB 2015 Annual Meeting Paper revised from original submittal.


Suman, Bolia, Tiwari
Table 5: Estimates of the coefficients of independent variables
Unstandardized Coefficients Standardized Coefficients
Model B Std. Error Beta t Sig.
1 (Constant) 14.05 1.2 11.61 0
Gender (1) 0.31 0.84 0.011 0.36 0.717
Age (1) 0.23 0.69 0.01 0.33 0.741
Income (1) 0.95 0.71 0.044 1.33 0.183
Car (1) 0.55 0.8 0.023 0.69 0.489
Student (1) -4.47 0.78 -0.21 -5.72 0
Service (1) -1.91 0.66 -0.1 -2.88 0.004

The coefficients corresponding to student and commuters employed in formal sector (in-service) are significant
and negative thus, it can be concluded that the students and in-service commuters travel shorter distance as compared to
rest of the commuters. There is not much difference in the trip length by vehicle ownership. Figure 1 revealed that the 52
% of bus commuters are travelling more than 10 km, and it varies slightly among different categories. Car owners are
travelling shorter distances compared to others, may be because direct bus service is not available for long distances and
comfort is an issue may be more severe for long distances. The commuters are using buses for shorter trip as well for which
bicycles can be potentially used.

Overall 25% of bus commuters are travelling less than 5 km (Figure 1) and their trips can potentially be shifted to
bicycles. Since, the average income of bus commuters is Rs. 20,965 per month, bicycles can be easily owned. Table 6
shows the trip lengthwise cost of commute by bus in Delhi and estimated household income. The percentage of household
income spent on transportation varies from 8% to 15%, and increased with income (14). Estimated household income
shown in Table 6 (Column C, D, and E) is calculated by assuming 10% of the income spent on travel. Column ‘A’ of Table
6 shows the trip lengthwise one-way trip fare.

Car Motorcycle None Total


100
Cummulative % of commuters →

75

50

25

0
0 5 10
15 20 25 30 40
Average trip length →
Figure 1: Trip length distribution of captive users

Table 6: Cost of Bus Commute and Estimated Household Income Per-Month


A B C D E
Expected household income per month (Rs)d
Fare Monthly fare 1 Commuter 2 Commuters/ 3 Commuters/
Distance (km)
(Rs) (Rs) /household household household
Up to 4 km 5 220 (5*2*22) 2200 4400 6600
4 - 10 km 10 440 (10*2*22) 4400 8800 13200
Above 10 km 15 660 (15*2*22) 6600 13200 19800
Maximum (Daily pass) 40 880 (40*22) 8800 17600 26400
d
If, 10% is spend for commute.

TRB 2015 Annual Meeting Paper revised from original submittal.


Suman, Bolia, Tiwari
Column ‘B’ shows the monthly fare of bus, if commuters travel 22 days in a month, column ‘C’, ‘D’, and ‘E’
shows the trip length wise expected monthly household income for 1 commuter, 2 commuters, and 3 commuters per
household respectively. The maximum expected income required to travel by bus is Rs 19,800 as shown in table 3
(corresponding to Row ‘above 10 km’ & column ‘E’). However, the Average household income of bus commuters is Rs
20965, which is less than maximum expected income corresponding to 3-commuters per household. Hence, up to three
commuters per household can travel more than 10 km by bus without spending more than 10% of their income. Therefore,
the low-income commuters (77% frequent bus commuters) are captive users of public transport even though 16% have cars
and 38% have motorised vehicles.

5. PERCEPTION OF CURRENT BUS COMMUTERS


The questions regarding the perception of commuters in the survey has three options i.e. ‘Yes’ (Positive perception), ‘No’
(Negative perception), and ‘May be’ (Neutral perception) for all attributes. The summary of the responses for different
categories of commuters are shown in Table 7, i.e. why people are using public transport. Based on overall responses of
frequent bus commuters (Table 8), the main reasons of using public transport, specifically buses are as follows:
• Fare is low: Public transport is affordable.
• Travel in buses as well as access way to bus stops is safe and secure.

Further, the detailed analysis of survey responses is explained in section 5.1.

Table 7: Different categories wise percentage of Positive Responses of frequent commuters (Travelled more than 3
days/week by bus)
Gender Bus Fare Safety Security Directness Accessibility Crowdedness Punctuality
Female 83 78 79 75 68 33 66
Male 88 86 90 69 68 29 71
Age Bus Fare Safety Security Directness Accessibility Crowdedness Punctuality
Below 25 yrs 85 81 88 70 63 16 64
25 to 40 yrs 88 84 88 68 68 20 63
40 to 55 yrs 87 84 88 62 63 19 62
Above 55 yrs 90 93 87 68 70 21 71
Income Bus Fare Safety Security Directness Accessibility Crowdedness Punctuality
Less than Rs 10,000 83 81 86 67 58 11 64
Rs 10,000 to Rs 30,000 87 83 89 65 65 22 62
Rs 30,000 to Rs 50,000 90 85 89 74 75 15 69
More than Rs 50,000 94 86 89 76 78 26 63
Occupation Bus Fare Safety Security Directness Accessibility Crowdedness Punctuality
Student 89 76 87 71 69 15 61
Service 85 85 88 65 67 20 59
Others 81 95 95 55 56 20 78
Vehicle Ownership Bus Fare Safety Security Directness Accessibility Crowdedness Punctuality
Car 88 89 89 73 77 27 75
Motorcycle 82 78 85 71 67 22 62
MV 84 82 86 72 71 24 67
NMV 88 84 89 65 61 13 61

Table 8: Overall responses of frequent bus commuters


Bus fare Safety Security Directness Accessibility Crowdedness Punctuality
Positive Responses 709 673 706 556 525 132 458
Total Responses 847 841 826 826 835 819 793

5.1. Detailed analysis of perception of bus system among commuters:


To test the statistical significance of the responses, non-parametric statistical analysis (sign test) is used. Commuters
responses for all attributes (bus fare, safety, security, directness, accessibility, and punctuality) are more than 50% positive
except for crowdedness, where it is 20%. Therefore, if null hypothesis is rejected for particular attribute then it encourages
use of buses but inhabits in case of crowdedness. The responses are binomial random variable with sample size 875 and
probability of 0.5. Sample size is good enough hence; it is approximated as normal random variable with mean 437.5 and
standard deviation 14.79. Further, the mean and standard deviation are varies slightly amongst the attributes as sample size

TRB 2015 Annual Meeting Paper revised from original submittal.


Suman, Bolia, Tiwari
is not same for all attributes (Table 8). Commuters perceptions for various attributes (bus fare, safety, security, directness,
accessibility, crowdedness, and punctuality) of buses are captured by asking them if a given attribute inhibits/encourages
them to use buses. Responses reveal with strong statistical significance that bus fare, safety, security, directness, and
accessibility are attributes that do not inhibit the use of buses. The null hypothesis of commuters being indifferent to these
attributes is rejected with p-value very close to zero. The alternative hypothesis that attributes do not inhibit is therefore
convincingly accepted. Responses also reveal with a strong statistical significance that crowdedness is an attribute that
inhibits their use of buses. They also reveal that punctuality is the other attribute that inhibits their use of buses, albeit with
a lower statistical significance. These results are summarised in Table 9:
Ho: p = 0.5 [Users are indifferent about various attributes]
Ha: p ≥ 0.5

Table 9: Results of sign test and power of test


Attributes → Bus Fare Safety Security Directness Accessibility Crowdedness Punctuality
p-value ~0 ~0 ~0 ~0 ~0 ~0 0.0778
Power (α=0.0003) 0.9922 0.992 0.9906 0.9906 0.9913 0.9899 0.9872

5.2. Relationship between perception and socioeconomic characteristics of bus users


Multinomial Logistic regression is used to find out the relationship between perception and socioeconomic characteristics
of bus commuters. Multinomial logistic regression compares the groups through the combination of binary logistic
regression (Simon P. Washington et al., 2003). In our data set we have three categorical dependent variables (negative
responses, may be responses (neutral), and positive responses towards the existing bus system in Delhi) and five categorical
independent variables. Categorical independent variables were age, gender, income, vehicle ownership, and occupation.
The reference category for dependent variable is negative responses, as interest is to know relative importance of the
attributes that encourage/inhibits the use buses, which is measured as the likelihood of predicting the positive response
with respect to negative responses. Similarly, the reference category for gender as male, for age as old age (age more than
55 years), for income as more than Rs 50,000, for vehicle ownership as NMV, and for occupation as others (except student
and commuters employed in formal-sector) corresponding to independent variables. The section 5.2.1. tests the fitness of
model and section 5.2.2. likelihood of perceptions commuters at categorical level.

5.2.1. Test of Fitness of Multinomial Logit Model


Chi-Square Statistics
The p-value of the overall model corresponding to chi-square statistic that is the difference in -2 log-likelihood between
the final model and the reduced model (with and without entering the independent variable into logistic model) is less than
the significance level of 0.1 for all attributes i.e. bus fare, safety, security, directness, accessibility, crowdedness, and
punctuality. Hence, the null hypothesis i.e. no difference between the model with and without independence variables is
rejected. Therefore, at least one of the independent variable is significant in the model.

Multicollinearity
Multicollinearity in Multinomial Logistic regression is measured by standard error of ‘B’ coefficients. The standard errors
corresponding to all attributes are less than two. In fact all values are less than 0.5 therefore, there is no problem associated
with multicollinearity (10).

Fitness of Model for Independent Variables


Corresponding to each independent variable, at least one of the categorical variable has significant difference in the
perception with respect to other categorical variables. The coefficients of significant independent variables for each
construct are as follows:
 For Bus fare: Only the coefficient of income is non-zero at significance level of 0.05.
 For Safety: The coefficients of age, gender, and occupation are non-zero at significance of 0.05 and coefficient
of vehicle ownership is non-zero at significance level of 0.1.
 For Security: The coefficient of gender is non-zero at significance of 0.05 and coefficient of vehicle ownership
is non-zero at significance level of 0.1.
 For Directness: Only the coefficient of the gender is non-zero at significance of 0.05.
 For Accessibility: The coefficients of income, vehicle ownership, and occupation are non-zero at significance of
0.05 and coefficient of age is non-zero at significance level of 0.1.

TRB 2015 Annual Meeting Paper revised from original submittal.


Suman, Bolia, Tiwari
 For Crowdedness: The coefficient of vehicle ownership is non-zero at significance of 0.05 and coefficient of
gender is non-zero at significance level of 0.1.
 For Punctuality: The coefficients of vehicle ownership and occupation are non-zero at significance of 0.05 and
coefficient of income is non-zero at significance level of 0.1.

Thus, there is no problem regarding the fitness of model at the variable level. The next section explains the likelihood of
relative perception about each construct.

5.2.2. Likelihood of Perception of Commuters at Categorical Level


Table 10 shows the significant coefficients of positive responses with respect to negative responses corresponding to bus
fare, safety, security, directness, accessibility, crowdedness, and punctuality. The odds ratio explains the relative
importance of one category over another (Exp(B)) as shown in Table 10. In Table 10, results of ‘maybe’ (neutral) responses
are not included, because the purpose is to know the relative perceptions of commuter about different attributes that
encourage/inhibits use of buses in Delhi. Neutral responses are just inputs to the model to keep it unbiased but their analysis
is not required. The meaningful insights emerged out from the analysis corresponding to all attributes are shown in Table
10:

Table 10: Significant coefficients of positive responses with respect to negative responses for all attributes
Parameter Estimates
Positive responses w.r.t. Negative responses B Std. Error Wald Sig. Exp(B)
Bus fare Income (Less than Rs10,000) -0.642 0.347 3.43 0.064 0.526
Gender (Female) -0.796 0.229 12.141 0 0.451
Age (≤ 25 years) 0.778 0.364 4.578 0.032 2.177
Safety
Age (25 to 40 years) 0.658 0.326 4.062 0.044 1.931
Income (Less than Rs10,000) -0.617 0.382 2.599 0.107 0.54
Gender (Female) -0.669 0.215 9.674 0.002 0.512
Security Ownership (Motorcycle) -0.388 0.186 4.363 0.037 0.679
Occupation (Student) -1.07 0.276 15.056 0 0.343
Income (Less than Rs10,000) -0.634 0.257 6.072 0.014 0.53
Ownership (Car) 1.078 0.299 13.017 0 2.939
Accessibility Ownership (Motorcycle) 0.306 0.151 4.103 0.043 1.358
Occupation (Student) 0.656 0.207 10.085 0.001 1.927
Occupation (Service) 0.438 0.157 7.828 0.005 1.55
Gender (Female) 0.428 0.21 4.131 0.042 1.534
Crowdedness Ownership (Car) 1.019 0.26 15.409 0 2.772
Ownership (Motorcycle) 0.506 0.179 8.007 0.005 1.659
Income (Less than Rs10,000) 0.516 0.254 4.123 0.042 1.675
Ownership (Car) 0.606 0.265 5.245 0.022 1.833
Punctuality
Occupation (Student) -0.509 0.225 5.129 0.024 0.601
Occupation (Service) -0.576 0.166 12.059 0.001 0.562

Further, with the help of multinomial logistic regression the following results are stand out at categorical level that, the
perception of commuters vary with their socioeconomic profile. The following are the conclusions of the analysis according
to the socioeconomic profile of the commuters for all attributes:
1. Bus fare: Overall, the fare of the public transport is not high but commuters belong to low-income (income less
than Rs 10,000) household, relatively perceive that bus fare is more as compared to commuters belong to high-
income (income more than Rs 50,000) household (from value of Exp(B) is 0.526).
2. Safety: Public transport is perceived to be more safe for male, young (age less than 40 years), and high-income
commuters as compared to female, old (age above 55 years), and low-income commuters respectively (value of
Exp(B) in rows corresponding to safety from Table 10).
3. Security: Public transport is more secure for male, student, and motorcycle owing commuters as compared to
female, formally unemployed, NMV owning commuters respectively (value of Exp(B) in rows corresponding to
security from Table 10).

TRB 2015 Annual Meeting Paper revised from original submittal.


Suman, Bolia, Tiwari
4. Accessibility: Public transport is perceived to be more accessible for students, formally employed commuters, car
& motorcycle owning Commuters, and high household-income commuters when compared to formally
unemployed commuters, NMV owning commuters, and low household-income commuters respectively (value of
Exp(B) in rows corresponding to accessibility from Table 10).
5. Crowdedness: Public transport is perceived to be more crowded by female commuters and car & motorcycle
owning Commuters as compared to male commuters and NMV commuters (value of Exp(B) in rows
corresponding to crowdedness from Table 10).
6. Punctuality: Public transport is perceived to be more punctual by car owing and low-income household commuters
whereas it is perceived to be less punctual by students and formally employed commuters (value of Exp(B) in
rows corresponding to punctuality from Table 10).
Note: All categories of bus commuters have same perception about the Directness of travel.

6. DISCUSSIONS
The mode share of motorcycles and cars are 35% and 16% amongst frequent bus commuters, which is similar as reported
in earlier studies (2), (3), & (15). According to the RITES (4) report, the average trip length of cars, two wheelers, and
buses was 11.2 km, 7.4 km, and 10.6 km respectively. Whereas, the average trip length of current bus commuters in
February 2014 is 13.1 km. Among the current bus commuters, the average trip length of motorcycle owners are 12.7 km
and car owners are 11.7 km. The percentage share of current bus commuters having bicycle is 9% in Delhi.

Bus transport is cheaper as compared to private modes of transport and only the mode of transport that is affordable
for a substantial proportion of the population. Most current bus transport users are captive users, who use bus transport not
because of choice but due to unaffordability of other options. The average income of the current bus commuters is Rs
20,965. Presently, 16% of the frequent commuters of bus transport have cars and 77% of have income less than Rs 30,000.

As the quality of public transport is not good, forcing the commuters to use private transport (16). The share of
total vehicular trips by bus has reduced from 60% in 2001 to 41.5% in 2008. The buses in Delhi are over-crowded.
According to data shared by Delhi Integrated Multi-Modal Transit System Ltd. (DIMTS) (as well as observed during
survey), the current buses are over-crowded. Buses are designed to carry 40 (sitting) to 50 (standing) passengers. However,
on an average they carry double the capacity during peak hours. One of the reasons of overcrowding of buses is low
frequency as observed from the survey. The important concerns of bus transport were overcrowding, punctuality,
connectivity, time, and safety as observed by Badami & Haider (1), Grdzelishvili & Sathre (17), and Jain (18). Our analysis
revealed that the bus fare, safety, security, directness, and accessibility are the attributes that do not inhabits the use of
buses in Delhi but only attribute that inhabits the use of buses is crowdedness and little bit of punctuality. It is possible that
the commuters who perceive these attributes inhabits them to use buses are not using it. Hence, to capture their perceptions,
we have planned another more study for non-captive users.

7. CONCLUSIONS
This paper studied the user profile, trip profile, and perception of bus commuters about the bus system in Delhi. Following
are the important results stand out from this study:
• Sixty percentage of frequent bus commuters belongs to vehicle owing household but still they are using buses.
Amongst the frequent commuters 16% belong to car-owing household still they are captive users of buses, because
average household income of bus commuters is Rs20,965 and public transport costs less than 10% of their
household income. The reason is that the mostly commuters belong to low income segment. Therefore, they cannot
afford to travel frequently by private vehicle. However, they can easily afford to travel by bus.
• The percentage share of car owners are 18%, motorcycle owners are 38%, bicycle owners are 9% and 47% do not
have any vehicle at home.
• Percentage share of motorcycle owners is 42% and 35% among frequent and infrequent commuters respectively
and they perceive safety as a major concern as compared to others.
• Income and vehicle ownership are highly correlated higher income commuters are more likely to belong car owing
and motorcycles owing household as compared to low-income commuters.
• Males that do not have any vehicle at home travelled longer distances as compared to others.
• Male commuters feel that public transport is more safe & secure as compared to females.
• Female commuters feel that buses are more crowded as compared to male counterpart.

TRB 2015 Annual Meeting Paper revised from original submittal.


Suman, Bolia, Tiwari
• Low-income household commuters feel that public transport is more punctual.
• Captive users perceive that public transport is cheaper, safe & secure. but major problem is overcrowding and
second is punctuality

Since the busload factor is already high, if we want to retain the current bus users, the guidelines for future bus systems
need to be framed while keeping the above conclusions in mind. With rising income levels in the country, it will be an
important challenge for all cities to maintain or enhance shares of public transport. To increase the use of bus transport
services by choice, it will be necessary to ensure that the use of bus transport is comfortable, accessible, frequently
available, and should follows its schedule.

7.1. Future Scope


This study helped in understanding the profile of current bus users and their perception about bus transport. This will help
the planners in framing the guidelines for future bus systems, to retain the current bus users. It does not provide any
information about the profile and perception of the non-captive users of bus systems. Hence, we have planned another
study for non-bus users. That would help in providing the guidelines to attract new users’ i.e. non-captive users of public
transport become captive. The combined results of the both studies would help in increasing the mode share of future bus
transport in Delhi. Possible interventions to accomplish this could include the following:
• Optimising the frequency of buses
• Modeling for bus-bunching
• Demand management through congestion charging, staggered office hours, preferential lane for buses, integrating
ticketing system
• Currently conducting a survey among non-captive users of buses

REFERENCES
1. Badami, Madhav G., and Murtaza Haider. "An analysis of public bus transit performance in Indian cities."
Transportation Research Part A: Policy and Practice 41, no. 10 (2007): 961-981.
2. MORTH. “Road transport year book (2007–2009) (Vol. I). Ministry of Road Transport and Highway,
Government of India”. (2011)
3. MORTH. “Road transport year book (20011–2012). Ministry of Road Transport and Highway, Government of
India”. (2013)
4. RITES. “Transport demand forecast study and development of an integrated road cum multi-modal transport
network for NCT of Delhi. Final Report”. (2010)
5. Sahai, Sanjiv N., and Simon Bishop. "Bus System Reform in Delhi." (2008).
http://www.dimts.in/pdf/Bus_System_Reform_in_Delhi.pdf. Accessed June 27, 2014.
6. DIMTS, Delhi Integrated Multi-Modal Transit System Ltd.
http://www.dimts.in/Services_Transportation_Intelligent_Transport_System.aspx. Accessed July 1, 2014.
7. DTC, “Govt. Of NCT of Delhi”. Delhi Transport Corporation.
http://www.delhi.gov.in/wps/wcm/connect/doit_dtc/DTC/Home/New+Initiatives. Accessed July 1, 2014.
8. Link, Heike, and John Polak. "Acceptability of transport pricing measures among public and professionals in
Europe." Transportation Research Record: Journal of the Transportation Research Board 1839, no. 1 (2003):
34-44.
9. Rastogi, Rajat, and KV Krishna Rao. "Survey design for studying transit access behavior in Mumbai City, India."
Journal of transportation engineering 128, no. 1 (2002): 68-79.
10. Stopher, Peter. Collecting, managing, and assessing data using sample surveys. Cambridge University Press,
2012.
11. Dargay, Joyce M. "The effect of income on car ownership: evidence of asymmetry." Transportation Research
Part A: Policy and Practice 35, no. 9 (2001): 807-821.
12. Washington, Simon P., Matthew G. Karlaftis, and Fred L. Mannering. Statistical and econometric methods for
transportation data analysis. CRC press, 2010.
13. Park, Hyeoun. "An introduction to logistic regression: from basic concepts to interpretation with particular
attention to nursing domain." Journal of Korean Academy of Nursing 43, no. 2 (2013): 154-164.
14. Diaz Olvera, Lourdes, Didier Plat, and Pascal Pochet. "Household transport expenditure in Sub-Saharan African
cities: measurement and analysis." Journal of Transport Geography 16, no. 1 (2008): 1-13.

TRB 2015 Annual Meeting Paper revised from original submittal.


Suman, Bolia, Tiwari
15. Pucher, John, Nisha Korattyswaropam, Neha Mittal, and Neenu Ittyerah. "Urban transport crisis in India."
Transport Policy 12, no. 3 (2005): 185-198.
16. Singh, Sanjay K. "Review of urban transportation in India." Journal of Public Transportation 8, no. 1 (2005):
79-97.
17. Grdzelishvili, Inga, and Roger Sathre. "Understanding the urban travel attitudes and behavior of Tbilisi
residents." Transport Policy 18, no. 1 (2011): 38-45.
18. Jain, Suresh, Preeti Aggarwal, Prashant Kumar, Shaleen Singhal, and Prateek Sharma. "Identifying public
preferences using multi-criteria decision making for assessing the shift of urban commuters from private to
public transport: A case study of Delhi." Transportation Research Part F: Traffic Psychology and Behaviour 24
(2014): 60-70.

TRB 2015 Annual Meeting Paper revised from original submittal.

You might also like