Documente Academic
Documente Profesional
Documente Cultură
Finding Report
prepared by :
AVNEET KAUR AHUJA
April,
Background
Study background
Delhi Metro Rail Service started its operations in 2002 which has affected the lives of Delhi/NCR people on a
very large scale. It was designed to ease the traffic congestion on roads, reduce air pollution and provide an
easy, pocket-friendly and convenient- efficient mode of public transportation. For Delhi Metro to achieve its
objective of delivering quality service to its customers, it is imperative to study to know the level at which the
customers are satisfied with its services
Study Objectives
To understand the various driving factors impacting customers satisfaction with Delhi Metro
Study Design
Quantitative cross-section study: Metro commuters/riders were interviewed using a structured
questionnaire
Target Group
Male & Female
Introduction
service quality (SQ) is defined as customers perception of how well a service meets or exceeds
Study focused on service quality and customer satisfaction only to Performance Model
(SERVPERF)
Reliability of Delhi
Metro in comparison to
Other Public
Transportaiom
Overall Customer
Satisfaction with
Delhi Metro
Cleanliness
Temperature
Audio
announcements
quality
Seating
availability
Metro
Connection with
other
Transportation
Facilities for
ATMS, caf and
Shops
Frequency
Punctuality
Operational
hours
Staff
availability&
behaviour
Complaint
handling systems
Time required
for booking
tickets/tokens
Accessibility
Travel time
Fare price
General safety
Safety women
Safety checks
Crowd mgt
Facilities for
disabled and sr.
citizen
Abnormal
conditions
information
Encroachment
and tackling
beggars
Way-finding
signage & general
information
Variable Name
Description
X1
FREQUENTLY_X1
X2
PUNCTUALITY_X2
X3
OPT_HOURS_X3
X4
ACCESSIBILITY_X4
X5
TRAVEL_TIME_X5
X6
FARE_PRICE_X6
X7
GENERAL_SAFETY_X7
X8
SAFETY_WOMEN_X8
X9
SAFETY_CHECKS_X9
X10
X11
CROWD_MGT_X10
FACILITIES_PHYS_SR_CITIZENS
_X11
X12
CLEANLINESS_X12
X13
TEMPERATURE_X13
X14
AUDIO_QUALITY_X14
X15
X16
X17
X18
X19
SEATING _X15
CONNECTIONS_OTHER_TRANS_
X16
ATMS_SHOPS_X17
ABNORMAL_INFO _X18
ENCROACHMENT_X19
X20
SIGNAGE_INFO_X20
X21
X22
SATFFBEHAVIOUR_X21
COMPLAINT_HANDLE_X22
X23
DV
Y1
QUEUE_BOOKING_X23
Variable Name
SATISFACTION_Y1
Y2
RELIABLE_Y2
Metro connections with other modes of public transports like feeders, DTC buses, etc
Availability of ATMs, shops, caf, shopping avenues at metro stations
Information during abnormal conditions like about delay travel time status, forecast of restoration of metro services, etc
Encroachment and tackling beggars outside Metro stations
Reliability of Delhi Metro for you in comparison to Other Public Transports (DTC Buses, Taxi Cab),
Research Methodology
the questionnaire comprised of two sections:
First Section:
Gender, Age group, Education level, Occupation, Frequently and Primary Purpose of using Delhi
Metro
Second section:
various factors influencing customer satisfaction with Delhi metro, overall Satisfaction and
reliability of Delhi metro in comparison to other public transportation on 5-point likert scale
Where,
Y - Dependent variable
- Constant (intercept)
- Regression coefficients with respect to independent variables
Error Term
aggregation combination of two or more attributes into a singe variable using sum, variance
or distance in order to change the scale & to get more stable data
this
study considered distance (D) between overall customer satisfaction and reliability of Delhi
Metro
Formula - given the coordinates of two points, the distance between the points is defined by:
Where
dx is the difference between the x-coordinates of the points
dy is the difference between the y-coordinates of the points
Y
Ov erall Cu stom er Sat isfac tion
Relability
(0,0)
Hypothesis Formulation :
Hypothesis No.
Null Hypothesis (H0)
Service Quality Dimension : Reliability
Hypothesis 1
Hypothesis 2
Hypothesis 4
Hypothesis 5
Hypothesis 6
Hypothesis 7
Hypothesis 8
Hypothesis 9
Hypothesis 10
Hypothesis 11
Hypothesis Formulation :
Service Quality Dimension : Tangibility
Hypothesis 12
Hypothesis 13
Hypothesis 14
Hypothesis 15
Hypothesis 16
No significant relationship between Cleanliness and Joint Significant relationship between Cleanliness and Joint effect of customer
effect of customer satisfaction and reliability.
satisfaction and reliability.
No significant relationship between Temperature & Lighting
Service and Joint effect of customer satisfaction and reliability. Significant relationship between Temperature & Lighting Service and
Joint effect of customer satisfaction and reliability.
No significant relationship between Quality Audio
Announcements and Joint effect of customer satisfaction and Significant relationship between Quality Audio Announcements and
reliability.
Joint effect of customer satisfaction and reliability.
No significant relationship between Seat availability and Significant relationship between Seat availability and Joint effect of
Joint effect of customer satisfaction and reliability.
customer satisfaction and reliability.
No significant relationship between Service to Other Modes
Of Public Transports and Joint effect of customer Significant relationship between Service to Other Modes Of Public
satisfaction and reliability.
Transports and Joint effect of customer satisfaction and reliability.
Hypothesis 18
Hypothesis 19
Hypothesis 23
Sample size
the formula adapted to determine the sample size is as follows:
A total number of 121 questionnaires were therefore collected for this study
Google Forms
the data collection period of the study is of one week
invitation online sampling technique is used where potential respondents are alerted that
To cover 166 Delhi metro stations and apply sampling technique in order to collect data via face to
face structured questionnaires requires time and manpower for this study
Focus was on selected 23 independent variables which may not completely represent all the
variables that might influence customer satisfaction
Survey answers could be partial and numerous checks such as mood/feelings of respondents,
surveys fraud, limited sampling, and inability to reach challenging population for instance elderly
people are out of research results
Consumer is a focal point in this market research study and his/her satisfaction drivers are difficult
to judge precisely and accurately
The results from this study should be taken with caution
at the pattern of responses for each respondent thereby cleaning the invalid data points from the dataset
Reliability Test
a Cronbach's alpha coefficient reliability test was carried out obtain
i.e., how closely related a set of items are as a group for the 103 sample size.
If value of Cronbach's alpha > 0.600 the questionnaire items dictated reliability
if value of Cronbach's alpha < 0.600 the dictated questionnaire items unreliability.
Cronbach's Alpha
.907
N of Items
23
51
Male
Female
10
43
21
62
18-24
25-30
31-40
41-50
51-60
42
College but
not Graduate
Up to 1 2 th
class
38
Literate but
no form al
schooling
10
8
43
17
22
Daily
Weekly
Occassionally
Alternativ e
day s in a week
Monthly
Occupation
16
Post graduate
41
Primary Purpose
Frequency Usage
Education
Graduate
533
10
50
30
84
40
24
24
Student
Pvt Service
Business
Govt Service
Household
Others
Studies
Job
Shopping/han
gouts
Others
Business
Martial Status
49
Age (yrs)
Gender
dependent variable.
Sign F: tests whether R- square is significantly different from zero i.e., whole model is
significant or not
rejection of null hypothesis if p-value from F-test is <0.10
T ratio: checks the reliability of the estimate of the individual beta. looks at p- values at 10%
Status of residuals
D=2
No autocorrelation
D=0
Positive autocorrelation
D=4
Negative autocorrelation
R^2
.734
.539
Std. Error
of
Adj R^2 Estimate
.405
.685
N
103
Change
.539
F Change
4.022
df1
23
df2
79
Sig. F
Change
.000
Durbin
Watson
1.821
Test of Autocorrelation:
- using d-lower limit ( 1.2) and d-upper limit ( 2.1), the decision fall under indecisive zone
-Durbin-Watson statistic is 1.821 and (rho) using formula of d 2(1- ) 1-d/2 is 0.09. Since DW is less
than 2 and p not equal 0 (0.09>0), a test for positive autocorrelation should be conducted
(Ho): No statistically significant autocorrelation in residuals, (rho )=0
(Ha): statistically significant positive autocorrelation in residuals, > 0
-Thus, used Modified d test. Rejected at 5% level of significance (d< d-upper limit) where d-upper limit
(2.1), concluded that there is statistically significant first order positive autocorrelation in residuals
Regression
Residual
Sum of Squares
43.435
37.091
df
23
79
80.527
102
Mean Square
1.888
.470
F
4.022
sign
.000
-sign F=0.000 (F statistic= 4.022 with d.f. =23, 79) is less than p=0.10. Thus, the model is statistically
significant and can be used for prediction purposes
-only variables i.e. FREQUENTLY_X1 and CLEANLINESS_X12 were statistically significance since
their p-values <0.10. Ho is rejected.
-rest, other variables were statistically insignificant since their p values was (p>0.10). Ho is accepted
Variables
Intercept
Coefficients
2.160
Std. Error
t- Statistic
p-value
.000
.199
.117
1.696
.094
Ho Rejected
.061
.123
.497
.621
Ho Accepted
.074
.091
.808
.421
Ho Accepted
.078
.079
.994
.323
Ho Accepted
.014
.104
.135
.893
Ho Accepted
.051
.128
.402
.689
Ho Accepted
.017
.112
.156
.877
Ho Accepted
.098
.096
1.029
.307
Ho Accepted
-.147
.092
-1.602
.113
Ho Accepted
.041
.082
.501
.618
Ho Accepted
-.021
.090
-.234
.816
Ho Accepted
.227
.106
2.152
.034
Ho Rejected
-.153
.118
-1.300
.197
Ho Accepted
.079
.103
.760
.449
Ho Accepted
.021
.084
.254
.800
Ho Accepted
-.150
.099
-1.512
.134
Ho Accepted
-.121
.130
-.932
.354
Ho Accepted
.120
.113
1.064
.291
Ho Accepted
.080
.089
.897
.372
Ho Accepted
-.007
.104
-.067
.947
Ho Accepted
.190
.129
1.475
.144
Ho Accepted
.196
.142
1.378
.172
Ho Accepted
FREQUENTLY_X1
PUNCTUALITY_X2
OPT_HOURS_X3
ACCESSIBILITY_X4
TRAVEL_TIME_X5
FARE_PRICE_X6
GENERAL_SAFETY_X7
SAFETY_WOMEN_X8
SAFETY_CHECKS_X9
CROWD_MGT_X10
FACILITIES_PHYS_SR_CITIZENS_X11
CLEANLINESS_X12
TEMPERATURE_X13
AUDIO_QUALITY_X14
SEATING _X15
CONNECTIONS_OTHER_TRANS_X16
ATMS_SHOPS_X17
ABNORMAL_INFO _X18
ENCROACHMENT_X19
SIGNAGE_INFO_X20
SATFFBEHAVIOUR_X21
COMPLAINT_HANDLE_X22
and assume that the error terms follows the AR (1) scheme, namely ,
Knowing the value of (rho), the problem can be easily solved. In this study rho 1-d/2 is 0.09.
If
time holds true at time t, it also holds true at time (t-1). Hence,
Where
run regression to this new equation involving Y on X in difference form, not in original form
because et is no longer autocorrelated
cause problems in estimating the regression coefficient. In this situation, the overall
p-value may significant but the p-value for each predictor may not be significant
o
Examine tolerance (defined as 1/1-R^2 and the Variance Inflation Factor (VIF) , Collinearity
diagnostic and Pearson correlation matrix
VIF
Status of predictors
VIF = 1
VIF > 5 to 10
X1
.209
X2
X3
.318 .242
X4
X5 X6 X7 X8 X9 X10
.263 .355 .301 .319 .366 .264 .286
X2
X11 X12 X13 X14 X15 X16 X17 X18 X19 0 X21
.211 .334 .444 .346 .388 .604
.709 .378 .402 .510
.195
.233 .424
.391
.118
.013 .022
.119
.330
.230
.352 .213
.159
.136
.165 .232
.246
.169
.165 .231
.263
.172
.160 .319
-.00 .272
4
.430
X22
.412
X23
.376
.431
.442
.271 .221
.324
.426
.487
.533
.285
.744
.397
ATMS_SHOPS_X17
ABNORMAL_INFO _X18
ENCROACHMENT_X19
.271
SIGNAGE_INFO_X20
SATFFBEHAVIOUR_X21
.596
COMPLAINT_HANDLE_X22
QUEUE_BOOKING_X23
.172
.596
This is not sufficient, since multicollinearity can exist with no pairwise correlations being high
Remarks
Toleran
ce
VIF
FREQUENTLY_X1
0.433
2.308
PUNCTUALITY_X2
0.397
2.518
OPT_HOURS_X3
0.471
2.125
ACCESSIBILITY_X4
0.604
1.654
TRAVEL_TIME_X5
0.418
2.394
FARE_PRICE_X6
0.47
2.129
GENERAL_SAFETY_X7
0.369
2.708
SAFETY_WOMEN_X8
0.431
2.318
SAFETY_CHECKS_X9
0.421
2.373
CROWD_MGT_X10
FACILITIES_PHYS_SR_CITIZE
NS_X11
0.471
2.121
0.499
2.003
CLEANLINESS_X12
0.446
2.24
TEMPERATURE_X13
0.352
2.839
AUDIO_QUALITY_X14
0.46
2.175
SEATING _X15
CONNECTIONS_OTHER_TRAN
S_X16
0.431
2.321
0.397
2.52
ATMS_SHOPS_X17
0.288
3.468
ABNORMAL_INFO _X18
0.313
3.199
ENCROACHMENT_X19
0.445
2.248
SIGNAGE_INFO_X20
0.479
2.086
SATFFBEHAVIOUR_X21
0.315
3.176
COMPLAINT_HANDLE_X22
0.239
4.179
QUEUE_BOOKING_X23
0.375
2.665
multicollinearity exist
multicollinearity exist
multicollinearity exist
multicollinearity exist
multicollinearity exist
multicollinearity exist
multicollinearity exist
multicollinearity exist
multicollinearity exist
multicollinearity exist
multicollinearity exist
multicollinearity exist
multicollinearity exist
multicollinearity exist
multicollinearity exist
multicollinearity exist
multicollinearity exist
multicollinearity exist
multicollinearity exist
multicollinearity exist
multicollinearity exist
multicollinearity exist
multicollinearity exist
Dimension
1
Eigenvalue
22.667
Condition Index
1.000
.212
10.351
.168
11.606
.123
13.603
.113
14.176
.102
14.933
.082
16.617
.078
17.010
.064
18.788
10
.063
18.970
11
.049
21.494
12
.039
24.258
13
.034
25.720
14
.030
27.553
15
.028
28.300
16
.026
29.518
17
.022
31.821
18
.021
32.541
19
.019
34.283
20
.016
37.110
21
.014
40.438
22
.011
44.565
23
.010
48.495
24
.008
52.799
Strong multicollinearity
exists
Moderate
multicollinearity exists
Removal of Multicollinearity
REMOVAL OR REMEDY OF
MULTICOLLINEARITY DEPENDS ON THE
KNOWLEDGE ABOUT NATURE OF
INTERDEPENDENCE AMONG INDEPENDENT
VARIABLES
Factor Analysis:
Dropping Variable:
Best Delhi Metro Customer Satisfaction Model Equation using Stepwise, backward and forward regressions :
Model Summaryb
Model
.690
Change Statistics
Std.
Error of
R
F
R
Adjusted
the
Square Chang
Sig. F
Square R Square Estimate Change
e
df1 df2 Change
.476
.443
.663
.476
14.523
96
.000
Remarks
DurbinWatson
1.995
The adjusted R^2 has improved from 40.5% to 44.3% and the Standard Error of the Estimate has
decreased from 68.5% to 66.3% towards customer satisfaction
F=0.000 (F statistic= 14.523 with d.f. =6, 96) is less than both p=0.05 and therefore the model is
statistically significant.
ANOVAb
Sum of
Squares
df
Mean Square
Sig.
Regression
38.316
6.386
14.523
.000
Residual
42.211
96
.440
Total
80.527
102
Model
1
Best Delhi Metro Customer Satisfaction Model Equation using Stepwise, backward and forward regressions :
Coefficients
Unstandardized
Remarks
Collinearity
Coefficients
Testing at p
Statistics
Remarks
<0.10 and
p=<0.05
Model
1
Std. Error
Sig.
2.328
.398
5.855
.000
FREQUENTLY_X1
.274
.080
3.426
.001
.875
1.143
OPT_HOURS_X3
.117
.064
1.822
.072
.893
1.120
CLEANLINESS_X12
.232
.079
2.922
.004
.742
1.347
CONNECTIONS_OTHER_TR
-.145
.069
-2.091
.039
.764
1.309
-.129
.063
2.050
.043
.826
1.211
.306
.081
3.782
.000
.694
1.440
(Constant)
(Remark)
Reject Ho
Tolerance
VIF
Reject Ho,
1<VIF<5
Moderately
correlated, thus
multicollinearity
exist
ANS_X16
ENCROACHMENT_X19
COMPLAINT_HANDLE_X22
six variables show a significant relationship with customer satisfaction at 10% sign level
in
in
Removal of Heteroscedasticity
REMOVAL OR REMEDY OF
HETEROSCEDASTICITY DEPENDS ON THE
KNOWLEDGE ABOUT NATURE OF AMONG THE
UNEQUAL VARIANCE BETWEEN
INDEPENDENT VARIABLES
18-24 Years
Conclusion
Regression model identifies the two key factors that influence customers satisfaction
frequency
cleanliness
Best regression model predict that only following factors are significantly influence
customers satisfaction
frequent of metros
operational hours of metro
cleanliness of metros/stations
metro connections with other transportations
encroachment and tackling beggars outside metro stations
complaint handle system
Moving Forward
Continuously improve the survey
Delete, add and change questions
add more sample size to improve the model
Responses especially from elderly metro riders
Compare with other data points (secondary & Primary researches) for the broader
picture