3rdConference of Transportation Research Group of (3rd CTRG) Performance Analysis of Sub Urban Rail System in - A Case Study

Rahul Raoniara, Amudapuram Mohan Raob*, S. Velmuruganc

a Post Graduate Student, Academy of Scientific and Innovative Research (AcSIR), CSIR-Central Road Research Institute, -110025, India. b Senior Scientist, Traffic Engineering and Safety Division, CSIR-Central Road Research Institute, New Delhi-110025, India. [email protected] c Senior Principal Scientist and Head of Division Traffic and Safety Division, CSIR-Central Road Research Institute, New Delhi- 110025, India. * Corresponding Author

Abstract Today customer’s satisfaction is one of the key parameter which is considered by the transport organizations for assessment of service quality. Due to growing importance of quality in our life, customer’s desire for good quality of transport and superior quality of services has been increased. India is one of the leading nations at world level contributing better living standards, economy, and employment and also having rapid growth of population which contribute to an increase in demand of better safe transportation facilities. Delhi’s Suburban Rail transportation system is one of the cheapest modes for commuting. This paper aims to identify the major cause of lesser use of suburban rail transport system and identify the attributes, on whose improving may further leads to increase in service quality. Three techniques namely Important Performance Analysis (IPA), Customer Satisfaction Index (CSI) and Structure Equation Modeling (SEM) are used to identify the current service quality in terms of Current Service Satisfaction and Expectation.

Results indicates the current passengers of Delhi’s Suburban Rail system are not satisfied with the overall service quality provided. CSI value obtained is 0.45 which is quite low compared to the highest level of service quality measure i.e. ranges in between 0.81 to 1 and SEM analysis indicates, Delhi Suburban rail transit system isn’t having a positive impact on passenger satisfaction in Delhi City.

Keywords: Performance Evaluation Techniques, Survey and Scale Design, IPA, CSI, SEM.

1. Introduction Transportation facilities evolve day by day in all cities irrespective of their status. Delhi transportation system connectivity lies at par with major cities across the world. Public transit systems such as Metro; Bus and Rail are the most extensively used by the commuters on daily basis. The study aims to identify the service quality and performance gap in terms of user perception and to identify the major attribute that defines transit service performance.

2. Literature Review The performance of transit system can be enumerated based on two distinct dimensions i.e., Service and Service quality. Service is described as “the business transaction that takes place between a donor (Service provider) and Receiver (Customer) in order to produce an outcome that satisfies the customer” (Ramaswamy, 1996) [1]. Whereas, Service quality gives the measure of how well the service level is delivered to the commuters as per its expectation. Parasuraman (1988) defines service quality as a comparison between customer expectation and perception of service [2]. 3rdConference of Transportation Research Group of India (3rd CTRG)

The model proposed by Parasuraman (1985) known as SERVICE QUALITY (SERVQUAL) which comprises ten dimensions namely; Reliability, Responsiveness, Competence, Access, Courtesy, Communication, Credibility, Security, Understandability and Tangibles which later in 1988 reduced to five dimensions: Reliability, Assurance, Tangibles, Empathy and Responsiveness, known as RATER. This method calculates the service quality by measuring the gap in between service delivered and service perceived by the commuters [3]. Federal Administration of the U.S (1999) developed a simple and effective measurement method of customer satisfaction for transit services termed as Impact Score Technique [4].

3. Study Methodology 3.1 Study area Delhi suburban rail system is operated by northern railway for the National Capital Region (NCR) which was started in 1975 to serve the goods and passenger around Delhi city covering 25 km around the city. Delhi suburban railway services cover Delhi along with adjoining district of Faridabad, and other places of and . Most of the railway path covered by the EMU and MEMU passenger trains covering the Delhi use the same track used by long distance trains. The study area covers the suburban rail system covering south Delhi is considered, i.e. the line going from New station to Badarpur and ring rail which covers many area of Delhi is covered. User perception data is collected at three rail stations namely Tughlakabad, Okhla and .

3.2 Parameters Identification and Scale Design The selection of performance evaluation parameters and scale design solely depends upon the transit system. The parameters were formulated using all possible demographic and performance variables to capture passenger’s responses. The questions include 10 demographic questions and 19 performance measuring variables. The details of the demographic parameters are presented in Table 1 and performance variable are presented in Table 2. In this study a dual level questionnaire prepared by incorporating satisfaction level with the base line of comparison Expectation level. Hence a constant scale of 5 Likert scale [5] units has been selected which indicates minimum satisfaction and expectation level as 1 and maximum satisfaction and expectation level as 5. The questionnaire was designed to capture the existing satisfaction level for the system and also to capture the expected level of each parameter.

3.3 Data Collection The user opinion survey was conducted at entry/exit and on platforms of railway stations. The survey was conducted at three stations for one day from 6:00 am to 10:00 pm. The survey was conducted by trained enumerators. The samples are collected by random sampling technique by considering various socio economic parameters. The partly filled forms are rejected and not considered for the analysis. Around 1195 samples were collected at three selected stations, these data is used for further analysis. The user characteristics obtained from collected sample are presented in Table 1. Table 1 User Characteristics Characteristics Statistics 1. Gender Male (85.4%), Female (14.6%) 2. Age <20 (16.13%), 21-30 (43.19%), 31-40 (21.22%), 41-50 (12.14%), 51-60 (6.12%), 61-70 (1.11%), 71-80 (0.09%).

Raoniar, Mohan Rao and Velmurugan 3rdConference of Transportation Research Group of India (3rd CTRG)

3. Income <5000 (39.5%), 5001-15000 (32.9%), 15001-30000 (18.6%), 30001-45000 (5.0%), 45001-60000 (1.4%), 60001-75000 (2.1%), >75000 (0.5%). 4. Education

The demographic of present study represented in Table 1 shows majority of respondents are male (85.4%) with aged between 21 and 30, profession of the majority (49.1%) are employees. More than half (59.1%) of the user have a personal mode (Two wheeler/car) for commuting. More than half (51.7 %) passengers are regular commuters with monthly travel card.

3.4 Analysis of Data The processed data was analyzed to find what are the parameters are important and based on which parameters user are rating the system. Important performance analysis, overall satisfaction of the system is calculated. An attempt has been made to model the user satisfaction using SEM model technique. 3.4.1 Important Performance Analysis (IPA) The performance appraisal of public transport can be carried out using Important Performance Analysis (IPA) is also known as quadrant analysis in order to measure the relationship between user perception and priority, (Martilla 1977) [6]. In IPA, user satisfaction is translated into Cartesian diagram where two lines perpendicularly divide it into four quadrants shown in Figure 1. Where (Q) represents the average of average scores of level of implementation of all factors and (P) represents the average of average scores of the importance of all factors.

Raoniar, Mohan Rao and Velmurugan 3rdConference of Transportation Research Group of India (3rd CTRG)

Fig. 1 Cartesian Diagram or IPA Diagram (Supranto, 1997) The different terms of the Figure 1 are as follows:  Point (P) as the middle point of the score level expectations, obtained by dividing the total score of the average level of expectation per respondent each dimension with the existing number of dimensions.  Point (Q) as the middle point of the performance level score, obtained by dividing the total score of the average level of performance per respondent each dimension with the existing number of dimensions.  Quadrant 1: The factors that are considered important by the customer but in reality these factors have not been in line with expectations. Attributes that are included in this quadrant should get more attention or repaired so that the performance is increased.  Quadrant 2: The factors those are included in this quadrant must be maintained because of these quadrant parameters users are using the service, to retain the users these parameters has to be maintained.  Quadrant 3: The factors that are considered less important by the user but in reality are quite satisfactory. In this quadrant the performance of parameters are more than user’s expectations.  Quadrant 4: The factors that are included in this quadrant considered less important by the customer. The parameters in this quadrant are least prefer by user and their performance also low.

Performance Measure using IPA IPA has formulated to identify the rail satisfaction level of commuters' that enables identification of importance factors to identify which attributes have major impact on satisfaction. The user survey data collected in the study are converted into a Cartesian diagram to obtain the performance of rail transport. This diagram illustrates Cartesian quadrant intersection line on the average value of observations at the level of interest and axis performance assessment in order to determine the specifics of each factors lies in which quadrant shown in Figure 2.

Raoniar, Mohan Rao and Velmurugan 3rdConference of Transportation Research Group of India (3rd CTRG)

Personal Safety onboard Railway (IPA) Personal Safety at Station

4.0 Cleanliness of Trains 2.65, 3.96 2.68, 3.93 Cleanliness of Station 3.04, 3.90 2.37, 3.88 Cleanliness of Toilet 3.9 Facilities 2.21, 3.79 Availability of Seats at 2.68, 3.88 Station 2.14, 3.80 Crowding onboard 2.24, 3.79 Quadrant I Quadrant II Comfort onboard 3.8 2.05, 3.79 2.45, 3.78 Electrical Equipment 2.04, 3.74 2.27,3.78 2.17, 3.76 Functionality 1.74, 3.72 Railway Fare

3.7 2.31, 3.73 Frequency of Train Expectation (Y) Expectation 2.07, 3.70 2.20, 3.71 Regularity of Train 2.01, 3.63 Parking Facilities Quadrant IV 2.00, 3.67 Quadrant III 3.6 2.12, 3.60 Facilities for Disabled person Luggage facility onboard

Information Dessimination 3.5 at Station 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0 3.2 Timely Information Announcement and Display Current Satisfaction (X) Complaint Registration Ticket Inspection

Fig. 2 IPA Cartesian Diagram of performance Variables Table 2 shows the coordinates of different parameters with description of parameters and also explains quadrant to which it belongs. Table 2 Attribute Positions in IPA Variable Description of Satisfaction Importance Quadrant No. variable (X) (Y) I II III IV Personal Safety V1 on Board 2.14 3.80  Personal Safety at V2 Station 2.24 3.79  Cleanliness of V3 Train 2.05 3.79  Cleanliness of V4 Station 2.21 3.79  Cleanliness of V5 Toilet Facilities on Train/Station 1.75 3.72  Availability of V6 Seats at Station 2.37 3.88  Crowding on V7 Board 2.01 3.63  V8 Comfort on Board 2.00 3.67 

Raoniar, Mohan Rao and Velmurugan 3rdConference of Transportation Research Group of India (3rd CTRG)

Functionality of Electrical V9 Equipment’s (Fan/Light) 2.68 3.88  V10 Railway Fare 3.04 3.90  Frequency of V11 Train 2.20 3.71  Regularity of V12 Train 2.04 3.74  V13 Parking Facilities 2.12 3.60  Facilities for V14 Disabled Persons 2.07 3.70  Luggage V15 Facilities On Board 2.31 3.73  Information V16 Dissemination at Station 2.68 3.93  Timely Information V17 Announcement and display at Station 2.65 3.96  Complaint V18 Registration Facilities 2.17 3.76 Periodicity of V19 Ticket Inspection 2.45 3.83  Average 2.27 3.78

From the Figure 2 and Table 2 the following outcomes can be observed: The commuters strongly felt that safety on board and at station, cleanliness of train and station should be given top priority in rail services. The parameters are attracting users are fare, infrastructure facilities such as seat availability at station and train, arrival and departure announcements. These parameters are bonding the users to stick to the system. Interestingly the ticket inspection also bonding the user to the system. It is also observed that the luggage carrying facilities are performing more than user expectation. 3.4.2 Customer Satisfaction Index (CSI) Customer Satisfaction Index is a method to determine the level of satisfaction that has been achieved with respect to the service delivered. CSI was proposed by Supranto (1997) [7]. The user satisfaction measured using the average value of the level of expectation and the performance of each service item. The CSI can be calculated by the following steps:  Estimation of Mean Important Score (MIS) MIS is the mean value of each variable rate i.e. user satisfaction / expectation.

MISt = (i) 푛 ……………………………………………... ∑푡=1 푌푡 Where, n= no of respondent and Yt =Expectation value attributes 푛

Raoniar, Mohan Rao and Velmurugan 3rdConference of Transportation Research Group of India (3rd CTRG)

 Estimation of value of the Mean Satisfaction Score (MSS) It is the value of an average level of perceived per attributes.

MSSt= (ii) 푛 …………………………………….……… ∑푡=1 푋푡 Where, n= no of respondent and Xi=Real value attribute X to i 푛  Estimation of Weight Factor (WFt) The weight factor is an MIS value per attributes of the sum of all the attributes MIS.

WFt = ……………………………...... ………... (iii) 푀퐼푆푡 푝  Measurement ∑푡=1 푀퐼푆푡 of Weight Score This weight is the product of WFt with the mean level of service perceived as mean satisfaction score. WSt= WFt*MSSt ………………………………...... (iv)  Estimation of Customer Satisfaction Index (CSI)

CSI= 푝 *100% …………………………………… (v) ∑푡=1 푊푆푡 Where, p = attribute on behalf to p, HS = Highest scale used. 퐻푆

Supranto suggested rating for CSI values as; very satisfied (0.81-1.00), satisfied (0.66- 0.80), quite satisfied (0.51-0.65), less satisfied (0.35-0.50) and not satisfied (0.00-0.34).

Performance Measure using CSI CSI has formulated to identify the rail satisfaction level of commuters that enables identification of importance factors to identify which attributes have major impact on satisfaction. The values obtained from the CSI analysis is presented in Table 3.

Table 3 CSI Matrix of Railway Service in Delhi City Satisfaction Importance Weight Weight Variable No. Factor Score CSI (X) (Y) (WF) (WS) A B C D E F V1 2.14 3.80 1.00 2.28 0.46 V2 2.24 3.79 1.00 2.28 0.46 V3 2.05 3.79 1.00 2.28 0.46 V4 2.21 3.79 1.00 2.28 0.46 V5 1.75 3.72 0.99 2.24 0.45 V6 2.37 3.88 1.03 2.33 0.47 V7 2.01 3.63 0.96 2.18 0.44 V8 2.00 3.67 0.97 2.21 0.44 V9 2.68 3.88 1.03 2.33 0.47 V10 3.04 3.90 1.03 2.35 0.47 V11 2.20 3.71 0.98 2.23 0.45 V12 2.04 3.74 0.99 2.25 0.45 V13 2.12 3.60 0.95 2.16 0.43 V14 2.07 3.70 0.98 2.22 0.44 V15 2.31 3.73 0.99 2.24 0.45

Raoniar, Mohan Rao and Velmurugan 3rdConference of Transportation Research Group of India (3rd CTRG)

V16 2.68 3.93 1.04 2.36 0.47 V17 2.65 3.96 1.05 2.38 0.48 V18 2.17 3.76 1.00 2.26 0.45 V19 2.45 3.83 1.01 2.30 0.46 Total 43.16 71.80 19.00 43.15 8.63 Average 2.271 3.779 CSI overall Satisfaction 0.45

Different parameters along with individual and overall customer satisfaction Index is also presented graphically in Figure 3.

Suburban Rail CSI Values

Periodicity of Ticket Inspection 0.46 Complaint Registration Facilities 0.45 Timely Information Announcement and display… 0.48 Information Dissemination at Station 0.47 Luggage Facilities On Board 0.45 Facilities for Disabled Persons 0.44 Parking Facilities 0.43 Regularity of Train 0.45 Frequency of Train 0.45 Railway Fare 0.47 Functionality of Electrical Equipment’s (Fan/Light) 0.47 Comfort On Board 0.44

Crowding On Board 0.44 PerformanceVariables Availability of Seats at Station 0.47 Cleanliness of Toilet Facilities on Train/Station 0.45 Cleanliness of Station 0.46 Cleanliness of Train 0.46 Personal Safety at Station 0.46 Personal Safety On Board 0.46 0.4 0.41 0.42 0.43 0.44 0.45 0.46 0.47 0.48 0.49 CSI Values

Fig. 3 Customer Satisfaction Index for Different Parameters and Overall System From analysis it is observed that the satisfaction level for the parameters such as lighting (V9), fare (V10), information dissemination (V16, V17) and ticket inspection (V19) are moderately high. These results are in line with the IPA results (Quadrant 2 parameters). The overall CSI score of rail transit system is found out to be 0.45 which is below the minimum satisfactory value 0.51 indicates that Delhi’s suburban rail transit users are under satisfied with the overall performance of current services provided by the rail authority.

3.4.3 Structural Equation Modeling (SEM) Structural Equation Modeling (SEM) methodology is a powerful multivariate analysis technique in which a set of relationships between observed and unobserved variables are established. It was new method which began in the 1970s (Fornell, 1981) [8]. SEM methodology refers to a series of statistical techniques such as factor analysis, path

Raoniar, Mohan Rao and Velmurugan 3rdConference of Transportation Research Group of India (3rd CTRG)

analysis and regression models which are used to analyze data. Over the years, there has been a rapid development of different software packages such as LISREL and the AMOS which have greatly enabled the use and application of SEM techniques in diverse contexts. SEM tools consist of two parts, i) Latent variable model which describes the relation between the endogenous and exogenous latent variables and allows the direct assessment of both path strength and their underlying impact among those variable. ii) Measurement model which depicts the correlation between latent and observed variable. Both models described above consist of basic equation that describes the relationship with the independent variables. The basic equation of a latent variable model is; structural equation formulated to express causality relationship between variable which shown below in equation (vi).

Y = βX + e ………………………………………………….. (vi) Where: X = Satisfaction from the service Y = Expectation from Service e = Latent variable measurement error. β = Regression Weight (Regression coefficient of unstandardized beta)

Performance Evaluation Using SEM a) Model Configuration: In the present study of SEM analysis there are two latent factor were selected, ‘present satisfaction’ which comprises nineteen observed satisfaction variables and ‘expection’ comprises of nineteen observed expectation variables to identify the performance gap based on Confirmatory Factor Analysis (CFA) (refer Fig 4). In the present study the SEM model was developed based on confirmatory factor analysis using SPSS AMOS V22 software.

Fig. 4 Performance Measurement model A statistical test was carried out with both satisfaction and expectation data to check the level of significance and loadings (Regression weights) of each parameter. The test output of expectation is shown in Table 4. The inferences of the outputs are explained in subsequent sections.

Raoniar, Mohan Rao and Velmurugan 3rdConference of Transportation Research Group of India (3rd CTRG)

Table 4 Expectation variable estimates Standardized Standard Critical Variable Probability Regression Latent Variable Estimate Error Ratio ( No. Value (p) weights (st. (S.E.) C.R.) R.W) V1 Expectation 1.00 -- -- *** 0.775 V2 Expectation 1.01 0.035 28.7 *** 0.788 V3 Expectation 1.06 0.036 29.3 *** 0.801 V4 Expectation 1.00 0.035 28.6 *** 0.786 V5 Expectation 1.20 0.039 30.6 *** 0.828 V6 Expectation 0.90 0.032 27.8 *** 0.767 V7 Expectation 0.97 0.037 26.1 *** 0.730 V8 Expectation 1.10 0.036 30.1 *** 0.819 V9 Expectation 0.79 0.032 24.8 *** 0.699 V10 Expectation 0.57 0.033 17.3 *** 0.510 V11 Expectation 1.05 0.035 29.7 *** 0.810 V12 Expectation 1.05 0.034 30.4 *** 0.824 V13 Expectation 1.07 0.037 28.6 *** 0.786 V14 Expectation 1.13 0.038 29.9 *** 0.813 Y15 Expectation 1.00 0.035 28.8 *** 0.791 Y16 Expectation 0.78 0.033 23.9 *** 0.677 Y17 Expectation 0.83 0.033 25.3 *** 0.710 Y18 Expectation 0.96 0.034 28.4 *** 0.781 Y19 Expectation 0.85 0.034 25.0 *** 0.703

*** (asterisk) indicates full significance of attributes based on p-values. The loading factors and probability values obtained from CFA confirms that the loading values are positive defined and all manifest variables have a significant p (<0.05) value that authenticate all variables are positively and significantly form the latent factor present satisfaction (X). b) CFA Model Outputs The above observed variable loading (weights) that obtained from AMOS output forms the latent factors i.e. present satisfaction (X) and expectation (Y) and the different probability values that were obtained from analysis fully reflects the significance of each observed variable in formulation of the selected latent variables. Structural model generated for latent factor expectation (Y) can be incorporated into the equations presented in Table 5. The model output result shows in graphical form in Figure 4. For Suburban Rail system eqations for Expectations (Y) are represented in Table 5:

Table 5 Expectation estimate equations VE1= 0.775YR1 + 0.684 VE8= 0.819YR8 + 0.609 VE15=0.791YR15 + 0.617

VE2 =0..788YR2 + 0.645 VE9= 0.699YR9 + 0.671 VE16= 0.677YR16 + 0.739

VE3 = 0.801YR3 + 0.647 VE10= 0.510YR 10 + 0.959 VE17= 0.710YR17 + 0.695

VE4 = 0.786YR4 + 0.641 VE11= 0.810YR 11 + 0.591 VE18= 0.781YR18 + 0.606

VE5= 0.828YR5 + 0.679 VE12= 0.824YR12 + 0.530 VE 19= 0.703YR19 + 0.752

VE6=0.767YR6 + 0.576 VE13= 0.786YR13 + 0.731

VE7=0.730YR7 + 0.845 VE14= 0.813YR14 + 0.673

Similary Structural model generated for latent factor satisfaction (X) can be incorporated into the following equation shown in Table 6.

Raoniar, Mohan Rao and Velmurugan