Content uploaded by Moataz Mohamed
Author content
All content in this area was uploaded by Moataz Mohamed on Feb 19, 2016
Content may be subject to copyright.
Full Terms & Conditions of access and use can be found at
http://www.tandfonline.com/action/journalInformation?journalCode=gtpt20
Download by: [McMaster University] Date: 19 February 2016, At: 07:29
Transportation Planning and Technology
ISSN: 0308-1060 (Print) 1029-0354 (Online) Journal homepage: http://www.tandfonline.com/loi/gtpt20
Measuring the influence of bus service quality on
the perception of users
Moataz Mahmoud & Julian Hine
To cite this article: Moataz Mahmoud & Julian Hine (2016): Measuring the influence of bus
service quality on the perception of users, Transportation Planning and Technology
To link to this article: http://dx.doi.org/10.1080/03081060.2016.1142224
Published online: 18 Feb 2016.
Submit your article to this journal
View related articles
View Crossmark data
Measuring the influence of bus service quality on the
perception of users
Moataz Mahmoud
a
and Julian Hine
b
a
Department of Architectural Engineering, Faculty of Engineering, Assiut University, Egypt;
b
Built
Environment Research Institute (BERI), School of the Built Environment, University of Ulster, Antrim, UK
ABSTRACT
Encouraging the use of public transport is a key policy goal in many
countries. Therefore, public transport should offer the level of
quality that accommodates the demands of current users, and
importantly, the desires of potential users. This study investigates
the influence of the perceived quality of bus service on the
perception of both current and potential users. The study draws
upon data from 512 questionnaires distributed across Belfast City
in the UK. The study utilises a binary logistic regression model to
quantify the relationships between the perceived quality of 29
bus indicators (independent) and the perceptions of users
towards the overall bus service (dependent). Eleven significant
indicators are reported to have significant influence on the
perception of users. These indicators are utilised to propose
scenarios for optimising the quality of bus service with the
perceptions of current and potential users.
ARTICLE HISTORY
Received 6 October 2014
Accepted 5 November 2015
KEYWORDS
User perception; bus service
quality; indicators; binary
logistic regression; scenario
development
Introduction
The quality of bus transit service represents a fundamental aspect for implementing inte-
grated and sustainable transit solutions that can alleviate the problems of accelerated car
dependency. However, there is a demand on decision makers to rationalise and justify
public expenditures that are directed at the enhancement of bus service quality. Therefore,
it is imperative to identify the critical attributes of bus service quality that have significant
influence on the perception of different categories of users (Beirão and Sarsfield Cabral
2007; Eboli and Mazzulla 2007; Bordagaray et al. 2013). The availability of a tool that
prioritises the required improvement of bus service quality from the user perspective is
essential in order to increase passenger patronage through tailored solutions that target
the demands of different categories of users (Bordagaray et al. 2013; de Oña et al. 2013).
However, the quality management of bus transit service is a complex process. This
complexity emerges from the multidimensional interrelationships between the quality
of the service on the one hand, and its effects on the perception of users on the other.
Accordingly, the investigation of bus quality should represent the perspective of users
as well as service providers; both are essential partners in the quality management process.
© 2016 Taylor & Francis
CONTACT Moataz Mahmoud moutaz.mohamed@eng.au.edu.eg
TRANSPORTATION PLANNING AND TECHNOLOGY, 2016
http://dx.doi.org/10.1080/03081060.2016.1142224
Downloaded by [McMaster University] at 07:29 19 February 2016
This study focuses on bus service quality from a user perspective and aims at investi-
gating the relationships between the perceived quality of 29 indicators (independent) and
the perception of users on the overall bus service (dependent). The study also addresses
several objectives that include the development of a holistic framework of bus quality,
and the development of optimisation scenarios for bus transit service that considers differ-
ent categories of users.
In this respect, the study defines the term ‘perceived quality’ as the level of quality eval-
uated by users (Transportation Research Board 1999), and the result of user assessment of
the utility of a bus service (Lai and Chen 2011). Perception is defined by Habib, Kattan,
and Islam (2011) as an abstract and physiological term that plays a major role in governing
behaviour and defining action. Both concepts are detailed in the following section.
The study is organised as follows: the next section reviews the theoretical concepts of
bus service quality and the current practises of bus quality management. The ‘Method-
ology’ section illustrates the procedures of data collection, sampling, and statistical analy-
sis. The ‘Results’ section details the measurement of perceived quality, and the regression
model. Lastly, the study concludes with the discussion of potential scenarios for enhancing
bus service quality.
Current practise of bus service quality
The term ‘service quality’ in the context of public transport is defined as the measure of
how well the delivered service matches customer expectations (Transportation Research
Board 1999; Eboli and Mazzulla 2008; Mahmoud and Hine 2013). The term represents
two different ends in the same loop. For service providers, bus quality reflects how well
the delivered quality matches the targeted quality. In contrast, users are considering
service quality as the difference between the quality perceived from the service and the
desired quality that reflects their expectations. These definitions illustrate the complex
nature of bus service quality, as different measures of quality are linked in the same
process (Eboli and Mazzulla 2011; Habib, Kattan, and Islam 2011).
The inception of a quality loop identifies four different measures of bus service quality
(Quattro 1998; European Committee for Standardisation 2002). These include delivered
quality (the quality level delivered by a service provider), targeted quality (the quality stan-
dards targeted by a service provider), perceived quality (the quality of service as perceived
by a user), and desired quality (the quality expectations from users) (European Committee
for Standardisation 2002; Nathanail 2008). These four quality measures represent the per-
spectives of both customers and service providers as illustrated in Figure 1. Nathanail
(2008) argues that the service is considered successful only if the quality loop is retained.
Accordingly, three categories of measures have emerged to investigate the quality loop that
include performance-based measures, perception-based measures, and composite per-
formance/perception-based measures. Performance-based measures are not the focus of
this study.
Perception-based quality measures are employed to identify different patterns of
quality assessment based on the perception of different categories of user (Commission
for Integrated Transport 2002; Beirão and Sarsfield Cabral 2007; Guiver 2007); further-
more, to measure the perception, attitude, and preference of users towards bus services
2 M. MAHMOUD AND J. HINE
Downloaded by [McMaster University] at 07:29 19 February 2016
(Tyrinopoulos and Antoniou 2008; Iseki and Taylor 2010; dell’Olio, Ibeas, and Cecín 2010;
de Oña et al. 2013).
In practice, perception-based quality measures have been operationalised in various
forms. These include the evaluation of preference (Stradling, Anable, and Carreno 2007;
Rojo et al. 2013), satisfaction (Eboli and Mazzulla 2007, 2008, 2010), perceived quality
(Eboli and Mazzulla 2012; de Oña et al. 2013), and desired quality (dell’Olio, Ibeas, and
Cecin 2011). Preference measures are utilised to identify the level of importance allocated
to service attributes (Transportation Research Board 1999; Stradling, Anable, and Carreno
2007) while satisfaction refers to either transaction-specific or cumulative satisfaction.
Transaction-specific satisfaction is the expectation/perceived gap towards individual indi-
cators of the service, and cumulative satisfaction is the overall level of satisfaction of the
service (Stradling, Anable, and Carreno 2007; Chen 2008). Although they have been oper-
ationalised separately in the literature, some scholars have argued that separately neither
preference nor satisfaction can provide a comprehensive evaluation of user perception,
and advocated that both should be integrated to capture the perception of users (Oliver
2010; Eboli and Mazzulla 2012; Mahmoud and Hine 2013). Measures of perceived
quality identify the gap between the level of quality delivered by service providers and
the level of quality perceived by users (Nathanail 2008; Eboli and Mazzulla 2011).
Although the measures of perceived quality and perception (satisfaction and prefer-
ence) are used interchangeably, the differences between them are distinguished in the lit-
erature. Perceived and desired quality measures represent cognitive judgments (Lai and
Chen 2011), while satisfaction and preference (perception) measures represent affective
judgments (Joewono and Kubota 2007; Stradling, Anable, and Carreno 2007; Tyrinopou-
los and Antoniou 2008; Iseki and Taylor 2010; Oliver 2010).
In this respect, several scholars have investigated the links between perceived quality
and perception. Joewono and Kubota (2007) argued that perceived quality (also addressed
as user-perceived quality) has a significant impact on satisfaction, customer loyalty, and
mode choice. Lai and Chen (2011) echoed the same conclusion and added that perceived
value has a significant impact on the behavioural intentions of users, which eventually
affect mode choice and travel behaviour. Eboli and Mazzulla (2007) have also argued
that service quality influences the expectations and demands of users of the service.
Many researchers emphasised that user and service heterogeneity should be considered
in the evaluation of perceived quality (Cirillo, Eboli, and Mazzulla
2011; Bordagaray
Figure 1. Quality loop.
TRANSPORTATION PLANNING AND TECHNOLOGY 3
Downloaded by [McMaster University] at 07:29 19 February 2016
et al. 2013), which highlights that perceived quality is evaluated differently by different cat-
egories of user.
Two remarks may be drawn from the literature review; firstly, there is a clear distinction
between measures of perception and perceived quality, and secondly, there is a strong
causal relationship between perceived quality and perception. Although there is a
debate about the causal order of this relationship, it is theoretically and empirically
evident that perceived quality influences perception (Chen 2008; Lai and Chen 2011).
Therefore, this study argues that the quality evaluation of a bus service, from a user per-
spective, should consider the perceived quality as evaluated by users (cognitive judgment)
and its influences on the perception of users on the overall service (affective judgment).
This investigation could yield a better understanding of the influence of the quality of
each service attribute on the overall perception of users, which eventually plays a major
role in the process of mode choice and travel behaviour.
Methodology
Study context
This study draws upon both current and potential users for evaluating bus service quality:
‘current users’ refer to individuals who regularly use bus services as their main travel
mode, and ‘potential users’ refer to individuals who regularly use private cars as their
main travel mode and use bus services occasionally.
This study focuses on bus services in Belfast City (UK) provided by Translink that
include Metro and Ulster Bus. Translink is a regulated public transit service operated
by Northern Ireland Transport Holding Company (branded as Translink), and supervised
by the government’s Department of Regional Development (Northern Ireland). Metro ser-
vices operate in the Belfast metropolitan area through 12 arterial corridors that establish
an integrated link with other public transit services, while Ulster Bus services connect
Belfast City with surrounding towns and rural areas.
Despite the on-going efforts for increasing public transport market share in Northern
Ireland, problems are inherent. Belfast is regarded as one of the most car-dependent cities
in the UK. In 2010, 83% of work and leisure trips were made by private modes of trans-
port, compared to 70% in the UK as a whole, while only 5.1% of trips were made by public
transport services (3.6% by bus and 1.5% by rail) (NISRA 2010). Therefore, increasing pas-
senger patronage represents an element of significant importance to local authorities and
service providers in Belfast.
Data and sampling
The data collection process draws upon two measures that include perceived quality and
perception. The perceived quality data were collected as user self-reported evaluation of
service quality. The study utilised a set of 29 indicators that were derived from a long
list of indicators using focus group discussions with current and potential users, and
expert panel analysis that considered the perspectives of service providers, local auth-
orities, and academics. [This process is detailed at length in Mahmoud, Hine, and
Kashyap (2013).] The 29 quality indicators were classified into six attributes, namely
4 M. MAHMOUD AND J. HINE
Downloaded by [McMaster University] at 07:29 19 February 2016
service design, access to service, operation, fare, information and facilities, and safety and
security – as shown in Table 1.
The perception measure was derived from a previous study by the authors as a weighted
perception index (WPI, Mahmoud and Hine 2013). In a nutshell, the WPI is the inte-
gration of the preferences of users and their level of satisfaction using the analytical hier-
archy process. It should be noted that the perceived quality and perception measures
represent the same data set.
A questionnaire survey was distributed across Belfast City. The perceived quality from
29 indicators was evaluated using a 10-point Likert scale, whereby 10 is the highest value
given to an indicator (as illustrated in Figure 2). Nine explanatory variables were included
in the questionnaire to address both socio-economic and travel characteristics. The survey
also included other measures that are beyond the scope of this study.
The study employed the sampling approach devised by Bartlett, Kotrlik, and Higgins
(2001) to ensure that the collected data reflect generalised results. They have indicated
three fundamental aspects for the minimum sample size that address an adequate
Table 1. Bus quality indicators.
Main attributes Code Indicators Code
Service design SD
.
The comfort, cleanliness, and crowding of the bus
.
Need for transfers
.
Driver attitude and helpfulness
.
Route (network area covered)
SD_COB
SD_NFT
SD_DAH
SD_NAC
Access to service AS
.
Ease of access stops (routes and infrastructure)
.
Bus stop location and distance between stops
.
Handicap access installations
.
External interface to pedestrians, cyclists, car, and taxi
.
Availability of park and ride schemes
AS_EAS
AS_BSL
AS_HAI
AS_EIP
AS_APR
Operation OP
.
Waiting and transfer time
.
Boarding and alighting time
.
Total travel time
.
Reliability of the service (arrival time)
.
Operating hours
.
Frequency (weekly, weekend, and holidays)
OI_WTT
OI_BAT
OI_TTT
OI_ROS
OI_SOH
OI_FOS
Information and
facilities
IF
.
Availability of shelters, benches, and waiting areas at stop
.
Availability of amenities (enquiries points, sanitary, refreshment) at
terminals
.
Information during travel (real-time information)
.
Availability of information at station (signs, schedule, and maps)
.
Pre-trip information (phone and web)
IF_ABW
IF_AVA
IF_IDT
IF_IAS
IF_PTI
Fare FA
.
Bus fare
.
Availability of multiple-mode tickets
.
Ease of purchasing tickets (on board, at stops, at terminals)
.
Availability of monthly discount passes
FA_BFA
FA_AMP
FA_EPT
FA_AMD
Safety and Security SS
.
Visible monitoring (CCTV)
.
Lighting, noise, vibration, speed, and temperature on bus
.
Safety during trip (day and night)
.
Absence of offensive
.
Security against crimes on bus and at stops
SS_CTV
SS_LNV
SS_SDT
SS_AOO
SS_SAC
TRANSPORTATION PLANNING AND TECHNOLOGY 5
Downloaded by [McMaster University] at 07:29 19 February 2016
proportion of the population. These include the margin of error, sampling error, and pro-
portion of the population. Following these sampling guidelines, the study utilised the data
collection process with an alpha (α) level at 0.05, a 5% sample error, and a 0.25 estimation
variance for 50% of the population as shown in the following equation:
n =
t
2
∗p∗q
d
2
, (1)
where n is the acceptable sample size, t is the corresponding value for alpha level, p is
the proportion of the population, q =(1−p), and d is the sampling error. As a result, t for α
= 0.05 is (1.95), p = (.5), q = (0.5) and d = (0.05). Therefore, the minimum sample size is
calculated as 384 cases.
The analysis utilised 512 valid questionnaires that fulfilled the validation criteria. The
socio-economic characteristics of the data highlight the balance between current and
potential users (46.7% and 53.3%, respectively), and the diversity of the socio-economic
characteristics as highlighted in Table 2. Moreover, the diversity of geographical location
(rural and urban) was taken into consideration, which ensures that different constraints
and opportunities are accounted for.
Binary logistic regression model
A bi nary logistic regression model was developed to investigate the influence of per-
ceived quality on the likelihood of a user to have an overall high/low perception
towards the service. Several research studies have frequently advocated multiple
regression analysis to address the relation ships between a single-dependent vari able
Figure 2. Partial illustration of the distributed questionnaire.
6 M. MAHMOUD AND J. HINE
Downloaded by [McMaster University] at 07:29 19 February 2016
and a set of multiple-independent variables. Multiple regression employs a linear calcu-
lation to measure the influences of unit(s) c hang e in the independent variables on the
outcome of the dependent variable (Tabachnick and Fidell 2007;Field2009;Hair,
Black, and Babin 2010). The advantages of multiple regression models are based on situations
whereby a metric linear-nature relationship is investigated. However, the review of the percep-
tion and behaviour literature indicates that users often cannot explicitly express their perceptions
and attitudes (cf. Doorn, Verhoef, and Bijmolt 2007). In addition, it has been argued that
psychological behaviour could be comprehensively represented with a non-linear relationship
(e.g.high/low,yes/no,like/dislike)(cf.Field2009; Muley, Bunker, and Ferreira 2009). Doorn
et al. have stated that linear models cannot detect possible non-linearity in the relationships
between attitude and behaviour.
Binary logistic regression analysis is a binary form of multiple regression, which has the
capabilities to address non-linear relationships (Tabachnick and Fidell 2007; Field 2009;
Hair, Black, and Babin 2010). A binary logistic model combines the advantages of multiple
regression models in a binary form, whereby the outcome of the relationship is oriented
towards the likelihood of an event to occur rather than specific numerical values. Other
scholars have also advocated the use of ordinal logistic regression model to consider the
different ordinal categories (Eboli and Mazzulla 2009). However, since the perception
value is not collected according to an ordinal scale, it is derived from two measures, it
is not suitable to apply the ordered logit model. Accordingly, a binary logistic approach
is operationalised to predict and explain a binary (1–0) categorical variable from the rel-
evant sets of independent variables (metric and/or categorical). The main difference
between binary logistic and multiple regression is that the former employs a logistic
curve (S-shape) to ensure that the outcome of the dependent variable will fall within a
defined range (1–0), while the latter employs a linear relationship (Field 2009; Katz 2011).
In the transport context, a binary logistic approach is implemented to measure the like-
lihood and probability of mode choice, satisfaction, and behavioural intentions (Transpor-
tation Research Board 1999; Muley, Bunker, and Ferreira 2009). Binary logistic regression
offers several advantages for analysing the relationships between service quality and user
perception. It identifies two groups of indicators that in fluence the perception of users
including ‘drivers’ and ‘barriers’. Drivers refer to quality indicators that have significantly
Table 2. Socio-economic and travel behaviour characteristics of respondents.
Socio-economic variables Frequency Percentage
Gender Male 240 46.9
Female 272 53.1
Occupation Employed 297 58.0
Unemployed 215 42.0
Place of living City centre 57 11.1
Urban 166 32.4
Sub-urban 150 29.4
Periphery and rural 139 27.1
Driving licence Yes 400 78.1
No 112 21.9
Car ownership Yes 332 64.8
No 180 35.2
Travel mode Frequent bus user 239 46.7
Frequent car user 273 53.3
TRANSPORTATION PLANNING AND TECHNOLOGY 7
Downloaded by [McMaster University] at 07:29 19 February 2016
positive relationships with user perception; in contrast, barriers refer to indicators that
have significantly negative relationships with user perception (Field 2009; Katz 2011).
The typical binary logistic model is carried out in four steps: model structure, math-
ematical calculations (probability, odds, and coefficient), selection of regression method,
and model validation (Tabachnick and Fidell 2007; Field 2009; Hair, Black, and Babin
2010). In this respect, the binary logistic model is structured using the WPI as a dependent
categorical variable with two tiers of perceptions: a high perception tier (WPI values are
greater than the mean of all respondents) and a low perception tier (WPI values are
lower than the mean of all respondents). The average value of perception is used as a
threshold to distinguish between the two tiers; the rationale is to model the status quo.
Therefore, the WPI values are recoded using a binary code, whereby a high perception
tier = 1 and a low perception tier = 0. As a result, variables within the range of the high
perception tier (high tier ≥ 5.437) are transformed into a dummy variable with a value
of Y = 1, while all variables with the range of the low perception tier (5.437 > low tier)
are transformed into a dummy variable with a value of Y = 0. The quality measures of
29 indicators are used as metric independent variables.
Secondly, the probabilities of being in a high or low tier are calculated using the odds in
Equation (2) as the values of P(Y = 1) and P(Y = 0). Here, the concept of odd is the ratio of
an event occurring relative to the probability of the event not occurring. Accordingly, the
probability P(Y = 1) is expressed as
P(Y = 1) =
odds (Y − 1)
1 + odds (Y = 1)
=
e
logit (y =1)
1 + e
logit (y=1)
.
(2)
Thirdly, the study selects a regression method for the binary logistic model. Generally,
the logistic regression model has three different regression methods including forced,
forward stepwise, and backward stepwise entries (Field 2009; Hair, Black, and Babin
2010). It is argued that the stepwise (forward and backward) methods are suitable for
exploring a new hypothesis, whereby the model is developed to test new relationships.
In contrast, the forced method is suitable for theory confirmation, whereby the model
is supported by a strong theoretical approach (Field 2009). Therefore, the study utilises
the forced entry regression method, whereby all variables are included in the model.
Lastly, the model’s goodness-of-fit is calculated with Equation (3) as the ratio between
−2 log likelihood (−2LL) of the null hypotheses and the −2LL of the model and is
expressed using the pseudo R
2
as detailed in Equation (3). Hair et al. (2010) identified
that the pseudo R
2
value ranges from 0 to 1, whereby the R
2
value of 1 refers to a
perfect fit of the model.
R
2
Logit
=
−2LL
null
− ( − 2LL
model
)
−2LL
null
, (3)
where −2LL
null
is the LL function of the model without independent variables and
−2LL
model
is the deviation χ
2
of the model with all independent variables.
The interpretation of the binary logistic model draws upon the value of the exponen-
tiated coefficient (Exp.B), which provides evidence on the change of the odds. The Exp.B
8 M. MAHMOUD AND J. HINE
Downloaded by [McMaster University] at 07:29 19 February 2016
minus one equals the percentage change of the odds; therefore, an Exp.B value of 1 means
that there is no change in the odds, while an Exp.B value of 0.1 means that there is a 90%
change in the odds (0.1–1). Accordingly, quality indicators with a positive significant value
of Exp.B refer to an increase in the probability P(Y = 1), and are considered as drivers of
high perception. Quality indicators with a negative significant value of Exp.B refer to a
decrease in the probability P(Y = 1), and considered as barriers to high perception.
Results
Perceived quality from bus service
The results of the perceived quality from bus service indicate that overall the service
quality is moderate, with an overall quality value of 4.83 out of 10. The modal split of
current and potential users indicates that both have almost similar evaluations of the
service with overall perceived quality values of 4.92 and 4.74, respectively. Furthermore,
the modal split shows that both categories have the same pattern of evaluation.
Several indicators are regarded with relatively high quality. These include security
against crime (5.85), information at stops or stations (5.37), and the availability of
closed-circuit television (CCTV) (5.46). Three indicators are clearly regarded as providing
relatively lower quality including the frequency of the service (3.96), bus stop location
(3.99), and the availability of monthly discounts (4.09), as illustrated in Figure 3.
These findings indicate that currently the service provides relatively high quality in
safety and security, and service design attributes, while it provides relatively moderate
quality in both operational and information and facility attributes. Fares and access to
Figure 3. Perceived quality from bus service (current and potential users).
TRANSPORTATION PLANNING AND TECHNOLOGY 9
Downloaded by [McMaster University] at 07:29 19 February 2016
service attributes have relatively lower quality levels compared to other attributes as per-
ceived by users. These findings indicate some variations in the service quality as perceived
by both current and potential users.
Quality indicators influencing user perception
The results of the binary logistic model illustrate the quality indicators that influence the
perception of users. The results of the model – omnibus tests of model coefficients – indi-
cate that the model is statistically significant at P ≤ .0001, χ
2
= 486.518, and df = 5.0. These
results indicate that the model is able to distinguish the impact of perceived quality on user
perception and classify this impact into two different tiers of perceptions (high and low).
In total, the model explains a wide range, from 60.0% (Cox and Snell R
2
) to 80.3%
(Nagelkerke R
2
), of the impact of perceived quality over user perception, and classified
91.6% of all cases. In addition, the calculation of the overall model fit (pseudo R
2
) indicates
a considerable goodness-of-fit with R
2
LOGIT = 0.666.
Furthermore, the block entry approach contributes to the model improvements for
−2LL, as illustrated in Table 3. The table shows clear improvements on the Nagelkerke
R
2
from the first to the last block (columns three and four) whereby all variables are
included. In addition, there is a clear improvement in the predictive power of the
model by adding all relevant variables (columns five and six). These results confirm the
significant impact of quality on user perception, and support the model structure for
both variable selection and block entries.
The results of the model indicate that 11 (37.93%) indicators have significant impacts
on user perception, as detailed in Table 4. These significant indicators are distributed
across all six main attributes of bus service quality and include comfort of the bus, need
for transfer, bus stop location, availability of park and ride scheme, waiting and transfer
time, reliability of the service, frequency of the service, bus fare, monthly discount, infor-
mation at the station, and security against crime during travel.
The values of the original coefficient (B) and the exponentiated coefficient (Exp.B) indi-
cate that two variables have a relatively higher influence on user perception, namely the
need for transfer (B = 0.895, Exp.B = 2.446) and the frequency of the service (B = 0.734,
Exp.B = 2.083). In contrast, both information at the stop/station and the comfort of the
bus have relatively lower, yet significant, influences on user’s perception. Further
Table 3. Model improvement; Nagelkerke R
2
and predictive power.
[1] [2] [3] [4] [5] [6]
Block ID
−2 Log
likelihood
Nagelkerke
R
2
Change on
Nagelkerke R
2
Predictive
power (%)
Change on predictive
power (%)
Block 0 (Beginning Blok) 703.198 –– 55.7 –
Block 1 (SD) 537.576 0.370 0.376 74.8 19.1
Block 2 (SD+AS) 410.473 0.583 0.207 82.2 7.40
Block 3 (SD + AS + OP) 281.096 0.752 0.169 89.3 7.10
Block 4 (SD + AS + OP
+ IF)
273.305 0.761 0.009 89.6 0.3
Block 5 (SD + AS + OP
+ FI + FA)
249.190 0.787 0.026 89.8 0.2
Block 6 (SD + AS + OP
+ FI + FA + SS)
234.680 0.803 0.016 90.8 1
10 M. MAHMOUD AND J. HINE
Downloaded by [McMaster University] at 07:29 19 February 2016
inspection of the results indicates that all significant indicators have a positive relationship
(+B value) with user perception. In other words, an increase in the quality level of these
indicators will result in an enhancement of user perception on the overall bus service,
which confirms the order setting of the relationship between perceived quality and
perception.
The probability calculation shows that the overall probability of a user to be in the high
perception tier P(Y = 1) is 67.37%. However, the modal split of both current and potential
users indicates that current users have an 85.33% probability of having a high perception,
while potential users have a 45.13% probability of having a high perception. These findings
show the influence of perceived quality on the perceptions of both current and potential
users, and it supports the arguments of previous studies (Friman 2004; Eboli and Mazzulla
2007; Chen 2008; Lai and Chen 2011).
Optimising bus quality with user perception
Investigating the quality drivers and/or barriers to user perception offers in-depth knowl-
edge that can be utilised to optimise bus service quality. It offers an opportunity to develop
a holistic assessment of bus service quality from the perspective of users that provides a
detailed portrayal of bus service quality.
In this respect, the study develops a holistic quality model of bus transit that is consti-
tuted of two main elements – the WPI and the influence of the perceived quality for each
indicator. The model includes two thresholds and four quadrants. These thresholds
include, firstly, the impact size (B value) of perceived quality that is derived from the logis-
tic model and classified into two categories – significant and insignificant; and, secondly,
the WPI value for each indicator that is classified into two categories: above average and
below average. This classification produces two thresholds: horizontal (WPI value of 5.43)
and vertical (B value is significant > B > insignificant).
The model produces four quadrants as detailed in Figure 4. Nine indicators are allo-
cated to quadrant one (upper right) including the need for transfer, the frequency of
Table 4. The impact of perceived quality on user percepti on.
Variables in equation B SE Wald df P Exp(B)
Comfort of bus QU_SD_COB 0.238 0.114 4.321 1 .038* 1.269
Need for transfer QU_SD_NFT 0.895 0.177 25.616 1 .000*** 2.446
Bus stop location QU_AS_BSL 0.359 0.130 7.627 1 .006** 1.432
Park and ride schemes QU_AS_APR 0.560 0.142 15.596 1 .000*** 1.750
Waiting and transfer time QU_OI_WTT 0.380 0.135 7.891 1 .005** 1.462
Reliability of the service QU_OI_ROS 0.524 0.126 17.246 1 .000*** 1.689
Frequency of the service QU_OI_FOS 0.734 0.126 33.775 1 .000*** 2.083
Information at stop QU_IF_IAS 0.250 0.092 7.389 1 .007** 1.284
Bus fare QU_FA_BFA 0.432 0.144 8.954 1 .003** 1.540
Availability of discounted tickets QU_FA_AMD 0.397 0.129 9.450 1 .002** 1.487
Safety at stops/stations QU_SS_SAC 0.380 0.132 8.245 1 .004** 1.462
Constant −27.874
Cox and Snell R
2
0.600
Nagelkerke R
2
0.803
R
2
L
0.666
Note: Only significant indicators are reported in the table.
***P ≤ .001.
**P ≤ .01.
*P ≤ .05.
TRANSPORTATION PLANNING AND TECHNOLOGY 11
Downloaded by [McMaster University] at 07:29 19 February 2016
the service, the availability of park and ride schemes, the reliability of the service, bus fare,
bus stop locations, security against crime at stops, the comfort of bus, and information at
station. These indicators have relatively higher perception values, and the perceived
quality from these nine indicators have a significant impact on user perceptions. There-
fore, it is argued that these nine indicators represent the first priority of intervention
for optimising bus service quality.
The results show that two indicators are allocated to quadrant two (lower right); these
are the waiting and alighting time, and the availability of a monthly discount. Although the
perceived quality values from these two indicators have a significant influence on user per-
ceptions, the WPI values of these indicators are below average. Therefore, these two indi-
cators represent the second candidates for optimising service quality. Quadrant three
(upper left) represented eight indicators with high perception values, yet no significant
impact on quality. Lastly, quadrant four (lower left) represented 12 indicators with
lower perception values and no significant impact on perceived quality.
This model differs from the current literature in many ways; yet it is argued that it pro-
vides a more comprehensive analysis of bus transit quality. The model prioritises the
required quality schemes based on the perception of users, and accounts for the internal
relationships between the perceived quality and the perception of users.
Discussion
The results of this study provide a detailed taxonomy of bus service quality from a user
perspective. It illustrates the linkages between the perceived quality from bus service
and the perception of users on the overall service. In addition, it represents an opportunity
for enhancing bus service quality in a way that considers the perceptions of both current
and potential users. The study proposes optimisation models of bus service quality that
better inform policy-makers and service providers with the impact of quality improvement
on the perception of different categories of users.
Figure 4. Integration of weighted perception and the perceived quality.
12 M. MAHMOUD AND J. HINE
Downloaded by [McMaster University] at 07:29 19 February 2016
The study developed two optimisation scenarios for bus service quality based on the
theoretical approaches of the pursuit of synergy and the removal of barriers (May and
Roberts 1995; May, Kelly, and Shepherd 2006). The pursuit of synergy approach argues
that the use of group instruments at the same time will reinforce each other. Accordingly,
in this approach, all 11 significant indicators are included, and the scenario indicates that a
10% increase in the perceived quality (one unit at the ordinal scale) of these indicators
Figure 5. [Colour online] Bus quality optimisation: the pursuit of synergy scenario.
Figure 6. [Colour online] Bus quality optimisation: the removal of barriers scenario.
TRANSPORTATION PLANNING AND TECHNOLOGY 13
Downloaded by [McMaster University] at 07:29 19 February 2016
increases the probability of current and potential users to have high perceptions by 13.32%
and 44.34%, respectively, as detailed in Figure 5.
However, financial barriers represent fundamental constraints for quality improvement
schemes. As a result, the study also employed the removal of barriers approach to mini-
mise the cost of quality improvements. The removal of barriers scenario (whereby only
eight operational indicators are included based on a simple cost analysis – frequency of
the service, reliability of the service, waiting and transfer time, security at stop or
station, the comfort of the bus, bus fare, the availability of monthly discounted tickets,
and information at the stop or station) indicates that a 10% increase in the perceived
quality of these indicators increases the probabilities of having high perceptions for
both current and potential users by 11.53% and 33.35%, respectively, as detailed in
Figure 6.
Conclusions
This study has presented a different approach to investigating the quality of bus service
from a user perspective, and offered practical implementations for evaluating and
further optimising the quality of bus service in a way that considers the perceptions of
both current users and potential users. The study has also provided measures of the per-
ceived quality of bus transit based on user’s self-evaluation of bus service quality.
The main contribution of this study emerges from the quantification of the influence of
perceived quality on user perception. The study identified 11 indicators of significant
importance if we are to enhance service quality. These indicators are frequency of the
service, reliability of the service, waiting and transfer time, security at stop/station,
comfort of the bus, availability of monthly/seasonally discounted tickets, information at
stop/station, bus fare, need for transfer, bus stop location, and the availability of park
and ride service. These findings provide an opportunity to develop scenarios for optimis-
ing the quality of bus service in a genuinely new way. These scenarios illustrate that a con-
siderable shift in the perception of users towards bus service could be implemented
through the quality improvement of a relatively small set of indicators.
Overall, the findings of this study provide policy-makers with clear indications of the
required quality improvement of bus service that would potentially lead to increases in
passenger patronage as they consider the demands of different categories of user. In
addition, further research could emerge from this model that considers the relationship
between targeted service quality and the desires of users.
Acknowledgment
The authors would like to thank the Scientific Committee of the 13th World Conference on Trans-
port Research (WCTR) for awarding the presentation of an abridged version of this article the ‘Best
Paper Presented by a Young Researcher Award’.
Disclosure statement
No potential conflict of interest was reported by the authors.
14 M. MAHMOUD AND J. HINE
Downloaded by [McMaster University] at 07:29 19 February 2016
ORCID
Moataz Mahmoud http://orcid.org/0000-0002-1345-7240
References
Bartlett, James, Joe Kotrlik, and Chadwick Higgins. 2001. “Organisational Research: Determining
Appropriate Sample Size in Survey Research.” Information Technology, Learning, and
Performance Journal 19 (1): 43–50.
Beirão, Gabriela, and J. A. Sarsfield Cabral. 2007. “Understanding Attitudes Towards Publi c
Transport and Private car: A Qualitative Study.” Transport Policy 14 (6): 478–489. doi:10.
1016/j.tranpol.2007.04.009.
Bordagaray, Maria, Luigi dell’Olio, Angel Ibeas, and Patricia Cecín. 2013. “Modelling User
Perception of bus Transit Quality Considering User and Service Heterogeneity.”
Transportmetrica A: Transport Science 10 (8): 705–721. doi:10.1080/23249935.2013.823579.
Chen, Ching-Fu. 2008. “Investigati ng Structural Relations hips Between Service Quality, Perceived
Value, Satisfaction, and Behavioral Intentions for Air Passengers: Evidence From Taiwan.”
Transportation Research Part A: Policy and Practice 42 (4): 709–717.
Cirillo, Cinzia, Laura Eboli, and Gabriella Mazzulla. 2011. “On the Asymmetric User Perception of
Transit Service Quality.” International Journal of Sustainable Transportation 5 (4): 216–232.
doi:10.1080/15568318.2010.494231.
Commission for Integrated Transport. 2002. Public Attitudes to Transport in England (MORI).
London: CfIT.
Doorn, Jenny, Peter C. Verhoef, and Tammo H. A. Bijmolt. 2007. “The Importance of Non-Linear
Relationships Between Attitude and Behaviour in Policy Research.” Journal of Consumer Policy
30 (2): 75–90. doi:10.1007/s10603-007-9028-3.
Eboli, Laura, and Gabriella Mazzulla. 2007. “Service Quality Attributes Affecting Customer
Satisfaction for bus Transit.” Journal of Public Transportation 10 (3): 21–34. doi:10.1.1.300.2043.
Eboli, Laura, and Gabriella Mazzulla. 2008
. “A Stated Preference Experiment for Measuring Service
Quality in Public Transport.” Transportation Planning and Technology 31 (5): 509–523. doi:10.
1080/03081060802364471.
Eboli, Laura, and Gabriella Mazzulla. 2009. “An Ordinal Logistic Regression Model for Analysing
Airport Passenger Satisfaction.” EuroMed Journal of Business 4 (1): 40–57. doi:10.1108/
14502190910956684.
Eboli, Laura, and Gabriella Mazzulla. 2010. “How to Capture the Passengers’ Point of View on a
Transit Service through Rating and Choice Options.” Transport Reviews 30 (4): 435–450.
doi:10.1080/01441640903068441.
Eboli, Laura, and Gabriella Mazzulla. 2011. “A Methodology for Evaluating Transit Service Quality
Based on Subjective and Objective Measures from the Passenger’s Point of View.” Tran sport
Policy 18 (1): 172–181. doi:10.1016/j.tranpol.2010.07 .007.
Eboli, Laura, and Gabriella Mazzulla. 2012. “Structural Equation Modelling for Analysing
Passengers’ Perceptions about Railway Services.” Procedia – Social and Behavioral Sciences 54:
96–106. doi:10.1016/j.sbspro.2012.09.729.
European Committee for Standardisation. 200 2. Transportation-Logistics and Servi ces-Public
Passenger Transport-Service Quality Definition, Targeting and Measurement. London: CEN.
Field, A. P. 2009. Discovering Statistics Using SPSS. 3rd ed. London: Sage.
Friman, M. 2004. “Implementing Quality Improvements in Public Transport.” Journal of Public
Transportation 7 (4): 49–65.
Guiver, J. W. 2007
. “Modal Talk: Discourse Analysis of how People Talk about Bus and Car Travel.”
Transportation Research Part A: Policy and Practice 41 (3): 233–248. doi:10.1016/j.tra.2006.05.004.
Habib, Khandker M. Nurul, Lina Kattan, and Md Tazul Islam. 2011. “Model of Personal Attitudes
Towards Transit Service Quality.” Journal of Advanced Transportation 45 (4): 271–285. doi:10.
1002/atr.106.
TRANSPORTATION PLANNING AND TECHNOLOGY 15
Downloaded by [McMaster University] at 07:29 19 February 2016
Hair, J. F., W. C. Black, and B. J. Babin. 2010. Multivariate Data Analysis: A Global Perspective.7th
ed. Upper Saddle River, NJ: Pearson Education.
Iseki, Hiroyuki, and Brian D. Taylor. 2010. “Style Versus Service? An Analysis of User Perceptions
of Transit Stops and Stations.” Journal of Public Transportation 13 (3): 23–48.
Joewono, Tri Basuki, and Hisashi Kubota. 2007. “User Satisfaction with Paratransit in Com petition
with Motorization in Indonesia: Anticipation of Future Implications.” Transportation 34 (3):
337–354. doi:10.1007/s11116-007-9119-7.
Katz, M. H. 2011. Multivariable Analysis: A Practical Guide for Clinicians and Public Health
Researchers. New York: Cambridge University Press.
Lai, Wen-Tai, and Ching-Fu Chen. 2011. “Behavioral Intentions of Public Transit Passengers: The
Roles of Service Quality, Perceived Value, Satisfaction and Involvement.” Transport Policy 18 (2):
318–325. doi:10.1016/j.tranpol.2010.09.003.
Mahmoud, Moataz, and Julian Hine. 2013. “Using AHP to Measure the Perception Gap between
Current and Potential Users of Bus Services.” Transportation Planning and Technology 36 (1):
4–23. doi:10.1080/03081060.2012.745316.
Mahmoud, Moataz, Julian Hine, and Anil Kashyap. 2013. “Stakeholders’ Perspectives Towards Bus
Quality Indicators.” Proceedings of the ICE – Transport 1–22. doi:10.1680/tr an.11.00083.
May, Anthony D., Charlotte Kelly, and Simon Shepherd. 2006. “The Principles of Integration in
Urban Transport Strategies.” Transport Policy 13 (4): 319–327. doi:10.1016/j.tranpol.2005.12.
005.
May, A. D., and M. Roberts. 1995. “The Design of Integrated Transport Strategies.
” Transport Policy
2(2):97–105. doi:10.1016/0967-070x(95)91989-w.
Muley, Deepti, Jonathan Bunker, and Luis Ferreira. 2009. “Investigation into Travel Modes of TOD
Users: Impacts of Personal and Transit Characteristics.” International Journal of ITS Research 7
(1): 3–13.
Nathanail, Eftihia. 2008. “Measuring the Quality of Service for Passengers on the Hellenic
Railways.” Transportation Research Part A: Policy and Practice 42 (1): 48–66. doi:10.1016/j.tra.
2007.06.006.
NISRA. 2010. Northern Ireland Transport Statistics 2010–11. Belfast: Department for Regional
Development.
dell’Olio, Luigi, Angel Ibeas, and Patricia Cecín. 2010. “Modelling User Perception of Bus Transit
Quality.” Transport Policy 17 (6): 388–397. doi:10.1016/j.tranpol.2010.04.006.
dell’Olio, Luigi, Angel Ibeas, and Patricia Cecin. 2011. “The Quality of Service Desired by Public
Transport Users.” Transport Policy 18 (1): 217–227. doi:10.1016/J.Tranpol.2010.08.005.
Oliver, R. L. 2010. Satisfaction: A Behavioral Perspective on the Consumer. New York: M.E. Sharpe.
de Oña, Juan, Rocío de Oña, Laura Eboli, and Gabriella Mazzulla. 2013. “Perceived Service Quality
in Bus Transit Service: A Structural Equation Approach.” Transport Policy 29: 219–226. doi:10.
1016/j.tranpol.2013.07.001.
Quattro. 1998. “Quality Approach in Tendering Urban Public Transport Operations.” Final Report
by QUATTRO Consortium on behalf of European Commission DGVII, Luxembourg: Office for
Of
ficial Publications of the European Communities.
Rojo, Marta, Luigi dell’Olio, Hernán Gonzalo-Orden, and Ángel Ibeas. 2013. “Interurban Bus
Service Quality from the Users’ Viewpoint.” Transportation Planning and Technology 36 (7):
599–616. doi:10.1080/03081060.2013.845432.
Stradling, Stephen G., Jillian Anable, and Michael Carreno. 2007. “Performance, Importance and
User Disgruntlement: A Six-Step Method for Measuring Satisfaction with Travel Modes.”
Transportation Research Part A: Policy and Practice 41 (1): 98–106.
Tabachnick, B. G., and L. S. Fidell. 2007. Using Multivariate Statist ics. New York: Pearson.
Transportation Research Board. 1999. A Handbook for Measuring Customer Satisfaction and
Service Quality. Washington, DC: National Academy Press.
Tyrinopoulos, Yannis, and Constantinos Antoniou. 2008. “Public Transit User Satisfaction:
Variability and Policy Implications.” Transport Policy 15 (4): 260–272. doi:10.1016/j.tranpol.
2008.06.002.
16 M. MAHMOUD AND J. HINE
Downloaded by [McMaster University] at 07:29 19 February 2016
