Measuring Customer Satisfaction and Service Quality 81
CHAPTER 10. CONVERTING SERVICE QUALITY RESEARCH
FINDINGS INTO TRANSIT AGENCY PERFORMANCE
MEASURES
10A. Introduction
The assessment of the determinants of transit service quality has so far focused on the analysis of the
survey that measures transit users' attitudes towards service and derives the implied priorities for transit
service improvements. This analysis has provided useful insights into the factors that make up transit
rider satisfaction which influence mode choice behavior and consequently the observed transit ridership.
The interpretation of the survey results by managerial staff in each of the three transit agencies further
underscores the importance and usefulness of developing and maintaining a market research program
that focuses on customer satisfaction. The robustness and resonance of the survey findings with
management's opinions about the service offered bring to focus the steps that are required to take action
to improve service.
In this chapter we build upon the existing analysis framework by structuring the discussion of
performance measurement from a transit agency's management perspective. Instead of focusing on the
quality of service perceived and expected by the customer, we shift to ways of measuring the quality of
service actually offered by the transit agency. The ability to accurately measure performance allows the
agency both to evaluate its service and to define realistic and measurable goals for service
improvements.
We first discuss the importance of linking transit riders' perspectives to objective disaggregate measures
of transit performance. The different types of analyses that can be conducted are discussed along with
the desired elements of an ongoing data collection plan that focuses on the greatest possible level of
detail.
The performance measures are then identified in a manner that is consistent with customers' experience
by breaking down a transit trip to its individual components and by defining customer expectations of
service. Each of the 46 transit service attributes that were evaluated in the survey is related to the
different components of the transit trip to identify service attributes that share common characteristics.
The 10 most important aspects of service that have been identified through the survey analysis for each
transit agency are then tabulated to identify service attributes that are common to rail and bus transit
systems in each of the three cities. For each of those service attributes we define customers'
expectations and discuss a range of mostly simple performance measures that can be used to measure
the ability of the transit agency to offer service that meets these expectations.
10B. A Transit Agency's Perspective to Transit Performance Measurement
The consumer-oriented approach to transportation service planning is rooted in the assumption that the
observed transit ridership and transit market share are the result of the mode choices made by each
individual commuter. The analysis framework presented in Figure G.1 of Appendix G highlights the
importance of transit level of service, individual traveler characteristics, and communication and
marketing channels on the formation of travelers' perceptions and consequently on their likelihood of
riding transit.
Measuring Customer Satisfaction and Service Quality 82
The analysis of the transit rider survey has provided a way of evaluating the link between riders'
perceptions and their overall satisfaction with transit service. A better understanding of transit
customers' needs and wants would allow the transit agency to identify the strengths and weaknesses of
transit service against competing modes and the differences in service for individual routes within the
transit system.
Examples of successful customer-driven approaches to the design and marketing of transit service
quality are documented in a recent study of four European transit systems.
14
The common theme among
these case studies is the intent to demonstrate the transit agency's commitment to service quality and its
sensitivity to customer input by promising a standard of service. This allows customers to evaluate the
ability of the transit agency to provide the level of service to which it was committed.
Among the service improvements that were considered and implemented in the transit systems under
study were the provision of more frequent service, the improvement of reliability, purchase of new
equipment, improved customer relations, electronic payment facilities, and more convenient
connections. A similar review of 40 transit systems in the United States
15
identified increases in transit
ridership that system managers attributed to level of service adjustments, pricing changes, marketing
and information initiatives, enhancement of service coordination, and market segmentation.
Therefore, the next important step in the process from a transit agency perspective is to develop a
strategy of service improvements that is responsive to its customers' expressed needs and wants. In
particular, a transit agency needs to define the type and level of service improvements that need to be
implemented to address weaknesses in service for those service attributes considered most important by
its customers.
The collection of data reflecting riders' perceptions of transit service along with an ongoing program of
transit performance data collection at the transit line and route level by different times of day and days
of the week can be used by a transit agency to:
•
identify existing weaknesses of transit service as reflected in the responses provided by transit
riders and in the performance measures being monitored;
•
set priorities for service improvements by focusing on the aspects of transit service that need to
be addressed first and by identifying the service routes and segments of the market that will be
affected the most;
•
design and implement the identified improvements in transit service; and
•
design an information dissemination program that will properly communicate the improvements
to the riding public.
A recent Transit Cooperative Research Program study
16
approaches the subject of quality of transit
service by adopting a total quality management (TQM) framework for public transportation. To meet
the objectives of increased productivity, reduced costs, and higher ridership through improved rider
satisfaction the study focuses on controllable factors that influence public transit performance.
Recognizing the human service character of public transit, the study focuses on "putting customers first"
by responding to customer expectations and by translating market research into actionable procedures.
Measuring Customer Satisfaction and Service Quality 83
An important consideration in the outlined approach is the ability to "manage by fact" and establish a
range of measures that can be used to monitor and evaluate performance. Among the criteria for
developing these performance measures that are included in the report are the:
•
validity of data that are sampled by credible unbiased methods;
•
completeness of data that cover a broad spectrum of aspects of service;
•
policy sensitivity of data that can be used to support managerial decisions;
•
timeliness of data that can be processed, analyzed and interpreted on time;
•
transparency of the data collection process;
•
inexpensive data that may already be collected for another purpose; and
•
ability to interpret data by developing measures that are easy to understand, compare,
and communicate to management and the public.
The ability to make the linkage between riders' statements and measures of transit performance is
therefore instrumental in providing transit management with the means of evaluating alternative service
improvements aimed at enhancing rider satisfaction and transit ridership. Such an evaluation can be
supported by an ongoing data collection effort that captures differences by transit route, time of day,
and day of the week and focuses on a comprehensive list of transit performance indicators. As a result,
the ongoing analysis of the transit performance measures can be used to:
•
provide transit management with a systemwide overview of transit operations for
different transit modes;
•
evaluate transit performance on a route-specific level of detail by focusing on individual
segments of the transit network;
•
monitor changes in transit service over time to identify deteriorating conditions or to
highlight improvements in service in response to service intervention;
•
identify the variation in transit level of transit service by collecting data specific to a
service area, time of day, or day of the week for the service attributes of interest; and
•
guide the development of marketing and communication strategies to inform transit
customers and potential customers of the desirable service features.
10C. Overview of Transit Performance Measures
The collection of transit performance data to support the monitoring, evaluation, and the implementation
of improvements in service presents a challenge to transit agencies. Although transit agencies might be
interested in collecting a wide array of information, the cost of collecting and analyzing a large amount
of transit performance and service quality data presents a constraint to transit agencies.
Measuring Customer Satisfaction and Service Quality 84
As a result, the data collection and analysis activities should be concentrated on those aspects of transit
service that are both crucial to their operations and that more accurately reflect the needs and wants of
customers and potential customers. The objective is to match the most important perceptions to specific
aspects of transit service and to identify one or more corresponding service performance indicators.
These measures will differ by transit agency given the different priorities expressed by riders, the
differences in the nature of services offered, and the feasibility and cost of collecting the relevant data.
Travelers' need to travel reflects their need to participate in an activity that is located elsewhere. In this
context, travelers' choices of residential location, workplace, time-of-day of travel, and transportation
mode reflect their desire to minimize the disutility of travel. In the case of transit riders, the disutility of
travel encompasses the whole travel experience from the planning of a transit trip at their point of origin
through the walk egress portion of the trip to get to their final destination. To better understand and
measure the service that a transit rider receives, the total travel experience has been broken into the trip
components and service dimensions shown in Table 10.1.
Table 10.1
Correspondence Between Trip Components and Dimensions of Service
Prior to their trip, transit riders may need to seek information about the most convenient route, departure
time, transfers, and fare to get to his or her destination. Sources for such information include printed
transit route maps and schedules, information provided over the phone by individuals at a passenger
information center, and electronic versions of schedule and fare information. Although such information
is seldom needed for routine trips, it can be of great value to infrequent transit users and non-users who
are unfamiliar with the system.
The level of transit fares is another aspect of transit service that contributes to the disutility of travel and
affects riders' perceptions of transit's attractiveness. Although transit fares are often lower than the
corresponding operating, maintenance, and parking costs of private modes, fare levels can have an
adverse impact on the price-sensitive frequent traveler segment of the travel market. The availability of
different types of fares, such as monthly passes, ten-ride discount tickets, and electronic fare media with
value storage capabilities, and fare restrictions increase travelers' flexibility to choose an optimal
payment strategy that fits their own travel patterns.
Measuring Customer Satisfaction and Service Quality 85
The travel components of a transit trip include:
•
the access to the transit station/bus stop,
•
the time spent waiting for transit service,
•
the in-vehicle experience of riding transit,
•
potential transfer(s) to different transit services, and
•
the egress to the final destination.
The access and egress walk components of the trip are only in part linked to the everyday operations of
a transit system. Although the number, location, and spacing of stations and stops and the adjacent land-
use development may affect transit service considerably, they are primarily linked to the original design
of the service. On the other hand, riders' perceptions of the accessibility of rail stations and bus stops
can be positively influenced by interventions such as kiss-and-ride facilities, availability of long-term
station parking, sidewalk maintenance, availability of well-lit access paths, and maintenance programs
for stairs, escalators, and elevators leading to platforms.
The time waiting at the station or stop, the in-vehicle component of the trip, and the transfer to another
transit route are all characterized by:
•
traditional measures of transit service such as wait time, travel time, and service
reliability;
•
the station/stop and vehicle environments that the transit riders experience; and
•
the availability and quality of information available to riders at rail stations, bus stops,
and en route.
Table 10.2 provides a link between the components of a transit trip, the dimensions of transit service,
and the 46 attributes of service that were used in the transit rider survey. These linkages illustrate both
the depth of the rider survey and the potential range of corresponding measures of performance. The list
of candidate performance measures can be extended even further considering that a variety of measures
can be defined for attributes like service reliability depending on the nature of service. A range of
surrogate measures may be needed to properly reflect riders' feelings of security at stations, stops, and
on-board transit vehicles.
Measuring Customer Satisfaction and Service Quality 86
Table 10.2
Ratings of Service By Trip Component and Service Dimension
Measuring Customer Satisfaction and Service Quality 87
Table 10.2
Ratings of Service By Trip Component and Service Dimension
(continued)
Measuring Customer Satisfaction and Service Quality 88
In the remainder of this chapter, we focus on the 10 most important determinants of service for each of
the transit systems under study. Table 10.3 summarizes the findings and highlights the similarities and
differences across the three systems and the two CTA lines that were examined.
The two service attributes that emerged as the most important across all routes sampled were the
frequency and reliability of transit service, both of which reflect important policy-sensitive aspects of
transit service design. The third service attribute that was mentioned by riders in all three transit
systems but only in one of the CTA lines was the freedom from the nuisance behaviors of others, an
important but subtle and difficult to quantify service dimension. The remaining "top ten" service
attributes were split between those that were perceived as important by riders in Chicago and
Lynchburg and those that were mentioned by riders of the Sun Tran service who mostly focused on
frequency-related issues.
In sections 10D to 10M, we focus the discussion on the individual service dimensions and the
corresponding measures.
10D. Frequency of Transit Service
Based on the customer satisfaction surveys, frequency of transit service is among the most important
elements of transit service. Frequency was at the top of riders' lists for each of the three agencies where
transit riders were surveyed.
Frequency has two interpretations for transit riders. First, it refers to the hours of operation of transit
services. Many routes and services are available only during weekday peak periods, and sometimes
riders need to make trips served by the routes and services on weekends and on off-peak times of
weekdays. Limitations in transit service hours obviously affect travelers who need to travel during the
hours or days when there is no service. In addition, some potential transit riders choose not to use transit
services because the particular services are unavailable for their anticipated return trips or because they
cannot be certain about the time of their return trips and need to be certain that they do not get stranded.
Limitations in transit services and routes are almost always necessary for reasons of cost-effectiveness.
The low ridership levels that would be generated on many routes simply cannot justify the cost of
providing services at these times. However, from the customers' point of view, having service at all
hours and on all days is desirable. A straightforward customer-oriented measure of this aspect of service
frequency is the hours per day and days per week that transit service is available for each route.
The second interpretation that customers have of service frequency is how often buses and trains come
when the route is being operated. This can be measured most directly by the wait time that customers
experience. When service headways (the time between successive trains or buses) are relatively short,
wait time can be assumed to be one-half the headway. As headways get longer and people begin to
arrive for specific scheduled trains or buses, wait times level out. However, the general inconvenience
of having only a few buses or trains from which to choose continues to increase as headways are
increased. Since headways and wait times usually vary by time of day and between weekdays and
weekends, measuring them for customers' actual time of travel is likely to greatly improve the
relationship between customer ratings and the service measures. Therefore, bus and train headways can
be used as straightforward measures of service convenience reflecting the frequency of service by route,
time of day, and day of the week.
Measuring Customer Satisfaction and Service Quality 89
Table 10.3
Similarities and Differences Across Transit Systems
In addition, customers making trips that require one or more transfers are likely to view the frequency
of the second and subsequent routes or services as especially important because those frequencies will
dictate the amount of time that the customers can expect to spend making transfers. Transfer time is
usually considered to be particularly onerous by transit riders. For this reason, it is recommended that
measures of the time spent transferring are developed at least for the most important origin-destination
pairs in the area served by transit.
The frequency of service is the primary determinant of actual customer wait times and one of the most
important determinants of their level of satisfaction with transit service delivery. Closely related to
service frequency (in customers' minds) is service reliability — the ability to stay on the expected
schedules. The next section discusses this aspect of service.
Measuring Customer Satisfaction and Service Quality 90
10E. Reliability of Transit Service
The large number of transit agencies reporting measures of service reliability reflects the importance of
providing reliable and predictable service both from a transit operations and a transit rider's perspective.
Furthermore, the variety of definitions of on-time reliability reflects the different perspectives of transit
agencies in measuring this service attribute (Appendix G).
It is highly advantageous both to operators and customers to maintain consistent and predictable service
on transit routes and lines. For operators, a lack of regularity and uniformity leads to the inefficient use
of resources (with some vehicles overloaded while others are underutilized), increased costs, and lower
systemwide productivity. Two-thirds of transit operators view maintaining reliability as very important
element of transit service delivery.
17
For customers, non-uniform and inconsistent service increases the
level of uncertainty and uneasiness they feel at stops and stations, exacerbates crowding on vehicles and
at station and stop waiting areas, and makes transfers more difficult and time-consuming.
The reliability of transit service is most often measured by on-time performance, which reflects how
closely the delivery of transit service matches the published schedule. Specific measures of on-time
performance include:
•
percent of trains or buses reaching pre-specified points on time in different time periods,
where on time is defined as arriving in a pre-specified time window;
•
variance in travel times between two points;
•
average minutes of bus or train delay measured at specific locations; and
•
regularity of service (schedule adherence) at specific locations.
There are certain dimensions to on-time performance that make its measurement complicated. The
objective of a transit rider is to arrive at his/her destination on-time, regardless of any en-route schedule
variations. It is possible for trains or buses to be badly off schedule, and still get a passenger to the
destination at the desired time. At the same time, transit riders are interested in minimizing the time
spent waiting for vehicles since it is a component of travel time that is perceived as more onerous than
invehicle travel time. It is also possible for the on-time performance measures to poorly conform to
riders' experiences in this regard.
In analyzing on-time performance measures, it is often difficult to compare different types of services
and different types of routes. Most on-time performance measures will have disparate ranges for
different transit modes because the modes are affected by different exogenous factors. For instance, it is
quite difficult to meaningfully compare the on-time performance of a commuter rail line with that of an
urban bus because the bus is more vulnerable to weather problems and highway incidents. Riders
recognize the inherent reliability differences, and usually customer satisfaction levels will be based on
different levels of expectation.
Even within mode comparisons are difficult. To facilitate the assessment of on-time performance a
distinction needs to be made between frequent transit service that is offered in small regular intervals
and infrequent service that is provided according to a published schedule. In addition, the length of the
route is likely to skew on-time performance results.
Measuring Customer Satisfaction and Service Quality 91
Because of these difficulties in comparing on-time performance for different services, it is also difficult
to develop meaningful systemwide on-time performance measures. The most effective measures are
obtained for specific services or small groups of services. They are best analyzed through comparisons
over time as opposed to comparisons with each other.
There are also a number of operations measures that can be used as a surrogate measure for transit
reliability. These measures are supply-driven and reflect the ability of the transit agency to provide the
required amount of service rather than the quality of service. These measures could be used as surrogate
indicators in cases where there is no option for additional data collection and analysis and include:
•
the frequency of service breakdowns which is usually expressed as the average number
of miles between breakdowns including a vehicle failure, road call, or service
interruption, and
•
vehicle availability which measures the number of vehicles that are available for service
suggesting that the likelihood that service will be delivered as scheduled decreases with
fewer available vehicles.
10F. Explanations and Announcement of Delays
For transit riders, one of the most difficult aspects of delays in service is the associated uncertainty
about what has happened and how long they will need to wait for a train or bus. Riders are much more
accommodating of delays when they are provided with information regarding the reasons for the delay
and the likely length of the delay. The information allows riders to better plan ahead, and at a broader
level, it helps to make riders feel like the transit system recognizes that the delays are a problem and
that the transit workers are actively working on the problems.
A number of transit systems try to provide delay information to riders through on-board and station
public address systems. In addition, some agencies have experimented with providing electronic map
information on-board vehicles, at stations, and at bus stops. Automated Vehicle Location (AVL)
systems allow operators to post real-time or close-to-real-time information for passengers.
In Europe, many transit agencies pride themselves on passenger amenities, especially the provision of
customer information.
18
In London, where uncertainty about delays is among the most common sources
of rider dissatisfaction, arrival time and destination information is beaconed to transit stops. In Stuttgart,
the transit agency makes use of their AVL-based transit fleet management system to provide traveler
information at terminal kiosks and through an in vehicle route guidance system.
19
In addition to the more high-tech communications devices, transit agencies also provide likely-delay
information to passengers through newsletters, flyers, and telephone customer service representatives.
A number of measures can be used to gauge how well delay information is being disseminated to riders,
including:
•
availability of on-board and station public address systems;
•
availability of other electronic real-time displays;
•
frequency and clarity of announcements and messages;
Measuring Customer Satisfaction and Service Quality 92
•
percentage of significant delays for which correct information was provided to
passengers on-board affected vehicles;
•
percentage of significant delays for which correct information was provided to
passengers waiting at affected stations or bus stops; and
•
percentage of foreseeable delays (construction, maintenance, weather-related) of which
customers are made aware.
Transit agencies also commonly measure the quality of their customer communications that are not
directly related to delays. Some agencies reported measures that are aimed at quantifying each of the
different communication efforts that transit agencies carry out. Examples of such measures include the
percentage of calls by the public answered within 90 seconds; the number of service requests received
by the public; and the number of calls received asking for transit-related information.
The number of complaints expressed by transit passengers is used by some agencies as a surrogate of
service performance and is often reported on a monthly basis. This measure presents an effort by the
transit agencies to be responsive to their clients' needs and wants. Agencies collect and analyze
complaints by type (e.g. facilities, operators) and by mode and normalize the frequency of complaints
by dividing by the number of transit riders or the number of transit service miles provided.
10G. Crowding On-board Trains and Buses
A common complaint about public transit systems in large cities is that trains and buses are often too
crowded. Generally, the most common reasons that vehicles get overcrowded is that there is a service
frequency or reliability problem, so the fact that crowding is of importance to survey respondents
reinforces the importance of measuring frequency and reliability.
The crowding on-board trains and buses is an easily quantifiable measure through the calculation of
various load factors. The load factors reflect the discrepancy between the available transit capacity and
the corresponding transit ridership. Load factors can be expressed as the number of passengers on a
vehicle divided by the vehicle's design capacity, the number of passengers divided by the crush capacity
of the vehicle, or the number of passengers on a vehicle divided by the number of available seats.
Passenger loading estimates are best obtained through direct observation of vehicles passing
prespecified points (usually the maximum loading points).
10H. Behavior of Other Riders
Security concerns are an important element of customer satisfaction. In the surveys, these concerns
manifested themselves as concerns about the behavior of other riders. If transit customers perceive that
the nuisance behavior of other riders is tolerated, then their level of concern about their personal
security will increase. Where there is a high level of so-called "quality-of-life" crimes and rules
violations, there is more of a feeling that there is no one in charge of the system.
One way to measure the level of nuisance behavior is to track police arrest and citation records. The
weakness of this approach is that it is confounded by changes in the level of effort by police to enforce
system rules and by the general presence of police within the system. The presence of police officers
within the system will tend to shift crimes and incidents to different places in the system, so measured
improvements may not accurately reflect riders' experiences.
Measuring Customer Satisfaction and Service Quality 93
Some transit agencies have tried to obtain measurements on the amount of nuisance behavior by
discretely sending observers into the system to collect information on fare evasion and other minor
crimes and rules violations. OC Transpo in Ottawa has developed Transecure, a neighborhood watch
program within its system to allow police to locate and respond to bad behavior or suspicious activities.
Information from such a program is likely to be better than arrest or citation data because those
observing the bad behavior will not be recognized as police. If a system is able to spend enough
resources to obtain a statistically significant sample of time periods and locations, then changes over
time can be monitored and compared to survey results.
10I. Smoothness of the Ride
The smoothness of the ride and the stops is an indicator of rider comfort that is not easily quantified.
Smoothness can be measured on a subjective basis by having transit staff ride transit vehicles that are in
operation and to rate the ride quality. Alternatively, scientific instruments could be used to measure the
forces being experienced by riders as the vehicles traverse their routes.
These measures are more difficult to use and interpret than other measures discussed in this chapter. A
number of factors contribute to the relative smoothness of the transit ride, including:
•
the condition of the railroad track or the roadway;
•
the operating condition of the rail or bus vehicles;
•
the speed of the bus and the composition of the roadway traffic; and
•
the experience of the rail and bus operator.
Riders' dissatisfaction about the smoothness of the trip can be caused by problems related to any or all
of these factors. Therefore, developing direct measures to quantify smoothness will not necessarily help
a transit operator determine whether or how to make improvements to the system to improve customer
satisfaction. Given this problem, it is probably unlikely that smoothness measures would be helpful to
transit operators unless they were specifically designed to isolate the different factors that go into ride
smoothness.
10J. Cost Effectiveness, Affordability, and Value
The cost of travel by transit is almost always subsidized by local, state and/or national governments in
an effort to promote transit use, alleviate roadway congestion, and improve the mobility of the transit-
dependent segments of the population. However, in almost all cases the users are required to pay fares
to use transit systems. Fare levels affect customer satisfaction and ridership.
For any given customer, the measure that is directly related to the questions of cost effectiveness,
affordability, and value is the cost per transit ride. Because most systems offer some type of discounted
multi-ticket fare as an option to a one-way ticket, the cost per ride may be different depending on the
ticket type that individuals use. If monthly passes or another type of unlimited ride ticket types are
available, the cost per ride will also vary based on the amount of system usage.
Measuring Customer Satisfaction and Service Quality 94
In most cases, the average cost per ride that individuals pay will vary by traveler market segment
because ticket type choice will vary by market segment. Developing separate measures for different
traveler market segments may be the best way to relate customer satisfaction with transit fare levels.
10K. Availability of Seats
Availability of seats is a special case of crowding on transit vehicles that is discussed above under
section 10G. One can measure the ratio of the number of people on a vehicle to the number of seats on a
vehicle to quantify the availability of seats.
10L. Frequency of Delays due to Repairs/Emergencies
The paramount importance of delays and reliability to transit passengers was discussed above under
section 10E. However, the analysis of the survey results suggests that riders do not consider all delays
equally. Delays that are due to breakdowns or accidents are particularly irksome to transit riders
because they are to some extent preventable. Weather-related delays, while certainly viewed negatively,
have a lesser impact than delays due to bus or train mechanical problems.
Transit agencies commonly quantify the reliability of transit vehicles with the measures mean distance
between failures (MDBF) or average failures per vehicle. Operations staff use these measures to detect
problems with vehicles of one type or another, so separate values are calculated for each vehicle type in
the system. The primary advantage of these measures is that most agencies collect this information on a
continuing basis anyway, so no additional data collection is necessary.
The primary disadvantage of these measures is that they are not collected for the purpose of measuring
the quality of customer service delivery. To relate the measure to particular riders' customer satisfaction,
it is sometimes necessary to obtain detailed information about the vehicle types being used on specific
routes and to calculate route-specific or service type-specific weighted averages of the mean distance
between failures. In addition, the type and circumstances of failures will have a large impact on
customers' perceptions and this information is not necessarily captured by the maintenance measures. It
would probably be quite useful to categorize the specific problems causing the breakdowns, whether or
not passengers were able to be rerouted once a vehicle broke down, and the response time to address the
incident.
The frequency of transit-related accidents was another category of measures cited by many agencies.
Some of the agencies normalize the number of accidents per miles of service while other agencies break
out accidents by type including passenger accidents, employee accidents, preventable accidents, vehicle
accidents, etc. Measures of accident incidence are usually reported on a monthly and a mode-specific
basis.
10M. Passenger Environment On-board Vehicles and at Stations/Stops
The general environment through which passengers travel on transit has a great deal to do with their
level of satisfaction. However, it is difficult to develop a consistent and objective approach to
measuring the quality of the passenger environment.
Some agencies employ professionals whose responsibilities include monitoring the system from the
customer's point-of-view. These individuals are trained to consistently rate stations and vehicles
according to specific objective measures or on qualitative pre-set scales. This information is then
Measuring Customer Satisfaction and Service Quality 95
aggregated and tracked over time to measure how the passenger environment changes. The information
is shared with the operations managers who are responsible for the specific elements being evaluated, so
that they are able to evaluate the quality of their departments' service delivery.
New York City Transit uses its passenger environment survey to obtain data on a wide range of subway
categories
20
, including:
Station
•
lighting at different locations within stations;
•
public address system clarity;
•
condition of escalators and elevators;
•
presence and readability of system maps in the stations;
•
amount of litter on the platforms and track bed;
•
amount of stains and spills on the platforms;
•
amount of graffiti in the station;
•
quality of the station signage;
•
condition of public phones;
•
condition of turnstiles, gates, token vending machines;
•
courtesy and appearance of token booth personnel;
•
availability of maps and system information in the station.
Subway Cars
•
exterior graffiti;
•
condition of doors;
•
lighting;
•
air conditioning, fans, car temperature;
•
clarity of station stop and safety announcements;
•
amount of litter, spills, and stains in the car and;
•
presence of graffitied, scratched, and cracked windows;
•
appearance of guards.
Bay Area Rapid Transit (BART) performs a similar quarterly review of its facilities.
21
The BART
survey includes 31 specific measures that are organized around organizational areas of responsibility.
The BART measures include:
Facilities Management
•
Station Patio Cleanliness
•
Parking Lot Cleanliness
•
Landscape Appearance
Station Operations
•
Station Cleanliness
•
Station Graffiti
•
Restroom Cleanliness
•
Advertisements in Stations
•
Brochures in Kiosks
Measuring Customer Satisfaction and Service Quality 96
Station Agents
•
Agent Available or Sign in Place
•
Agent in Uniform
•
Agent wearing Name Badge
BART Police
•
BART Police Personnel in Stations
•
BART Police Personnel in Parking Lots/Garages
•
BART Police Personnel on Trains
Public Address Announcements
•
P.A. Arrival Announcements
•
P.A. Transfer Announcements
•
P.A. Destination Announcements
Rolling Stock
•
Train Exterior Graffiti
•
Train Doors Operative
•
Train Interior Graffiti
•
Train Interior Cleanliness
•
Train Window Etching
•
Temperature on Trains
•
Advertisements on Trains
Elevator/Escalator Availability
•
Station Elevator Availability
•
Escalator Availability - Street
•
Escalator Availability - Platform
Automatic Fare Collection Availability
•
Fare Gate Availability
•
Ticket vending Machine Availability
On-Time Performance
•
Train on Time
•
Customer on Time
A number of the passenger environment measures are subjective and qualitative. The careful training of
observers and tests to ensure that ratings are being made consistently are essential for the data collection
effort to be effective. However, despite the difficulty in establishing and monitoring the data collection
effort, passenger environment surveys are probably the best way for transit agencies to understand their
systems from customers' perspectives.
Measuring Customer Satisfaction and Service Quality 97
ENDNOTES
14
European Conference of Ministers of Transport, Round Table 92:
Marketing and
Service Quality in Public Transport,
Organization for Economic Cooperation,
Paris, France, 1993.
15
Transit Cooperative Research Program, Research Results Digest, Number 4,
Transit Ridership Initiative
, Transportation Research Board, National Research
Council, Washington D.C., February 1995.
16
Transit Cooperative Research Program, Research Results Digest, Number 3,
Total
Quality Management in Public Transportation
, Transportation Research Board,
National Research Council, Washington D.C., October 1994.
17
National Cooperative Transit Research & Development Program, Synthesis 15,
Supervision Strategies for Improved Reliability of Bus Routes
, Transportation
Research Board, National Research Council, Washington D.C., September 1991.
18
Transit Cooperative Research Program, Research Results Digest, Number 22,
International Transit Studies Program - Report on 1996 Missions
, Transportation
Research Board, National Research Council, Washington D.C., October 1997.
19
Transit Cooperative Research Program, Research Results Digest, Number 20,
International Transit Studies Program - Report on the First Three Missions
,
Transportation Research Board, National Research Council, Washington D.C.,
May 1997.
20
Charles River Associates Incorporated,
Metropolitan Transportation Authority
Comprehensive Line Improvement Study,
March 1994.
21
Aaron Weinstein and Rhonda Albom,
Securing Objective and Reliable Data on
the Quality of the Passenger Environment — The Redesign of BART's Passenger
Environment Measurement System,
presented at the 77
th
Annual Meeting of the
Transportation Research Board (January 1998).
This page left intentionally blank.
Measuring Customer Satisfaction and Service Quality 99
CHAPTER 11. AN OVERVIEW OF DATA COLLECTION AND
ANALYSIS METHODS
In this chapter we outline the broadly defined desirable features of a data collection and analysis plan.
The differences in the level of service offered and the nature of the markets served by each transit
system do not allow the development of a unique set of specific data collection and analysis procedures.
Furthermore, the identification of a different set of priorities for service improvements by riders of
different transit systems further stresses the need for a customized approach to data collection and
analysis.
The broadly defined principles guiding the data collection and analysis approach are presented in two
sections. We first outline the elements of a data collection plan that minimizes biases and aggregation
errors, provides data that are internally consistent and relevant from a passenger perspective, and
accounts for the statistical significance of the collected data at a reasonable cost. We conclude our
discussion by outlining different ways of analyzing the collected transit performance data and
summarizing the results.
11A. Principles of Data Collection
In order to gauge the quality of customer service by measuring specific service attributes, it is essential
that the transit agency consider the quality of the data that are being collected and the appropriateness of
the chosen data collection method(s). As noted in the previous chapter, data on different service
measures can be obtained by a variety of manual and automatic methods.
The manual methods include observation of service attributes by field inspectors, by field worker data
collection staff, and by "mystery riders," transit agency staff or contractors who ride the system as
customers would without letting transit workers know who they are or where they will be. In many
cases, inspectors assemble the data that would be used in evaluating service attributes for their own
purposes, thus the added cost of using this information for customer service evaluation is low. Special
data collection procedures by transit staff and mystery riders can be used to obtain the other service
measures.
Some transit service measures can be recorded automatically. For instance, systems that use buses
equipped with AVL systems can automatically collect data on vehicle headway, on-time performance,
and ridership allowing us to calculate a multitude of performance measures discussed in this report.
Furthermore, the implementation of an AVL system allows the development of passenger information
systems that can be used to provide estimated time of arrival to waiting passengers, display vehicles on
an electronic map at a bus stop or rail station, and provide en route information to transit passengers.
A review of the current status of AVL bus transit systems in the U.S. along with a detailed technical
review of different AVL architectures and technologies is presented in a recent TCRP report.
22
The
advantage of such a data collection mechanism is that a variety of accurate performance data can be
automatically collected at the route level by time of day and day of the week. At the same time, the
challenge with these data is the ability to properly sample, organize, and analyze the information that is
gathered in order to obtain the meaningful measures that are being sought.
Measuring Customer Satisfaction and Service Quality 100
Planners need to be aware that there are several potential problems with any given measure that can
reduce its usefulness in analyzing service delivery. Among the potential problems are:
•
bias;
•
aggregation error;
•
inconsistency;
•
irrelevancy form the passenger perspective;
•
insignificance; and
•
cost to assemble and analyze data.
These issues are discussed below.
Bias.
In this context, bias refers to a set of systematic errors that tend to overstate or understate the
performance of the system for a specific measure. Performance measures should be as free from bias as
possible. Examples of biased measures include data from a non-representative sample of routes or
services and data assembled with methods that cause the observed situation to be different than that
experienced by riders. If an agency were to assess the reliability of its bus system by measuring on-time
performance only on routes of one type, say routes that serve major downtown stops, erroneous
conclusions about the system as a whole are likely. Similarly, if an agency were to evaluate aspects of
customer service by having uniformed inspectors observe transit employees' interactions with
customers, then it is likely the results of such an evaluation would not reflect conditions when
inspectors were not present.
Aggregation Error.
If service measures are collected at too gross a level, important nuances of
customer service delivery will be lost. For instance, if on-time performance was calculated on a
systemwide basis and was used to gauge customer satisfaction with on-time reliability, it is possible that
the measure is masking significant differences between different routes and lines. If a small number of
routes have significantly poorer performance than the system as a whole, their effect on the objective
service measures will understate the negative effect that they have on customer satisfaction.
Inconsistency.
Because the most effective way to analyze service measures is to analyze changes over
time and differences between different routes and services, the measures of service delivery and the
scales used to record them should be consistent over time, from location to location, and from one
evaluator to another. This is particularly important for the more subjective measures such as cleanliness.
If inspectors or members of the field staff are employed to rate stations or vehicles on cleanliness, each
one of them should have consistent ratings. In addition, the ratings should not vary with time. This is
sometimes difficult because changes in the level-of-acceptability of certain conditions are likely to occur
over time, particularly if a system invests in improvements in the specific aspect of service under study.
When agencies employ staff to make subjective measurements of service measures, the following steps
should be taken whenever possible:
•
develop objective measures whenever possible (e.g., use a thermometer to measure the
temperature on vehicles, rather than a field worker rating of temperature);
•
train the field workers extensively, employing actual field evaluations, to ensure that
different fieldworkers rate things consistently;
•
test inter-rater variations in ratings to ensure that raters remain consistent (sometimes the
best way to test this is to have raters have some overlapping responsibilities).
Measuring Customer Satisfaction and Service Quality 101
Irrelevancy to Customers.
Often, it is possible to use already-collected measures of performance to
evaluate service delivery to customers. Of course, whenever this is possible it is desirable from an
efficiency point-of-view. However, because these data are collected for purposes other than the
evaluation of customer service delivery, planners need to assess the relevancy of the measure to
customers. For example, information on on-time performance is commonly collected at train and bus
terminals. In many cases where ridership is highly directional or is skewed to be on only part of the
route or line, on-time performance at a particular terminal point may be largely irrelevant to customers.
If a morning peak train runs close to schedule going into the CBD but then is delayed after it has made
it past the CBD, the delay is irrelevant to the large majority of riders. In this case, a better on-time
performance measure would be one that was collected at a CBD station.
Insignificance.
In order to draw valid conclusions from the assessment of service measures, an agency
needs to ensure that enough data are sampled and assembled to make the conclusions statistically
significant. An agency should first define finite elements of its system, such as stations, buses in a
particular time period, or buses on a particular route. As a second step, statistical sampling methods
should be applied to determine how many of the elements need to be studied or observed in order to
make statistically valid conclusions. If information is assembled in an ad hoc way, it is possible that
variations in service quality will never be accurately observed.
Cost to Assemble Data.
Finally, as for any primary data collection effort, the costs of getting particular
types of data need to be considered and traded-off with the benefits of the data that would be collected.
In general, the errors introduced by the potential problems described above can be reduced somewhat
through more and better data collection efforts that almost always increase the cost of data collection.
Although it is difficult to determine the cost-effectiveness of data collection efforts, the agency should
set as a priority maintaining data on the measures associated with the three or four of the most important
aspects of service from the customer's point-of-view.
For those aspects of service that are perceived as less important, an agency should probably obtain data
through less rigorous methods, perhaps using less accurate measures that are already collected or are
easily collected. In developing cost estimates for service data collection, an agency should seriously
consider the added benefits of maintaining the data assembly over time, rather than on a one-time basis.
In addition, an agency should consider collecting detailed high-quality data for specific elements of the
system, rather than the system as a whole.
11B. Approaches to the Analysis of Performance Data
The ultimate objective of the analysis of the transit performance measures is to facilitate a focused and
accurate assessment of any existing weaknesses in service and the measures that need to be taken in
response to these performance problems. To provide transit management with a means of identifying
the strengths and weaknesses of transit service and supporting its evaluation the analysis should, as
stated earlier:
•
provide transit management with a systemwide overview of transit operations for different
transit modes;
•
evaluate transit performance on a route-specific level of detail by focusing on individual
segments of the transit network;
Measuring Customer Satisfaction and Service Quality 102
•
monitor changes in transit service over time to identify deteriorating conditions or to
highlight improvements in service in response to service intervention;
•
identify the variation in transit level of transit service by collecting data specific to a
service area, time of day, or day of the week for the service attributes of interest; and
•
guide the development of marketing and communication strategies to inform transit
customers and potential customers of the desirable service features.
To provide transit management with these insights, we demonstrate four different broadly defined ways
in which the collected transit performance data can be analyzed. We use as a hypothetical example a
measure of bus on-time reliability as reflected in the percentage of buses arriving late at the central
business district bus terminal. We have also assumed that comparable data on on-time performance are
available for four different points in time between 1979 and 1997. The figures that are presented and
discussed allow us to:
•
measure bus performance at a systemwide level and compare it with differences in
performance at the bus route level;
•
identify trends in systemwide and route-specific levels of bus performance over time;
•
assess differences in the perceptions of different market segments including bus riders and
nonusers, frequent and infrequent transit riders, riders using different routes, and riders
with different socioeconomic characteristics; and
•
compare riders' perceptions to measures of transit service to identify whether the strengths
and weaknesses perceived by riders actually reflect the level of transit service that is
currently provided.
These layers of analysis correspond to an ever-increasing level of complexity. It is therefore not
expected that all layers of analysis will be employed by each agency to study each of the important
aspects of service. Furthermore, the more complex analyses presented below also require a wealth of
data that may be maintained only for a few important measures of service.
I. Cross-Sectional Analysis of Transit Performance
The analysis of on-time transit reliability at a single point in time can provide a snapshot of transit
performance both at a systemwide and at a transit route level. Although the systemwide measure can be
a useful indicator of overall performance especially when monitored over time, it is also important to
focus on the performance over sections of the transit system to identify potential differences by line.
Figure 11.1 presents a hypothetical example where the aggregation at the bus system level without any
attention to the disaggregate route level of detail would mask important differences in performance by
bus route. As shown in Figure 11.1, the overall on-time performance for the transit bus system is
reflected on a satisfactory systemwide average of 87% of buses arriving within a specified time interval.
However, a more detailed analysis of on-time performance at the route level suggests that there are
considerable differences in route performance that would ordinarily be masked by focusing solely on
the systemwide average measure.
Measuring Customer Satisfaction and Service Quality 103
Figure 11.1
Comparative Route Analysis
Therefore, on the basis of such a cross-sectional analysis of the bus system, the analysis would conclude
that:
•
the overall level of bus on-time performance is satisfactory, but
•
there are important differences by route which suggest that:
•
route E experiences a significant amount of buses that are late and should be
identified as a priority for service improvements;
•
route B operates at an acceptable better-than-average level but should be monitored to
prevent any deterioration in service; and
•
route A should be used as a benchmark of on-time performance for the whole
system.
II. Historical Analysis of Transit Performance
An additional layer of analysis can be provided by the study of systemwide and route specific on-time
performance over time. Such an analysis can be used to identify trends of progress and deterioration in
transit service that are not provided by the snapshot provided by the cross-sectional analysis.
A review of the hypothetical historical patterns of on-time performance for the same system shown in
Figure 11.2 uncovers some important trends that could help explain the differences in on-time reliability
across the system. In particular, it appears that the systemwide trend of deteriorating on-time
% of Trains Late
Measuring Customer Satisfaction and Service Quality 104
performance has been reversed in the past three years. However, there are some important underlying
differences among the three routes suggesting that:
•
the current poor on-time performance for route E is the result of an ongoing deterioration in
transit level of service and reflects problems that date back more than a decade and that
have gradually affected transit service;
•
route B has enjoyed improved on-time reliability over the past three years reflecting the
systemwide trend; and
•
route A has maintained an excellent level of service over time.
Thus, despite the improvement in systemwide service performance the identified route-specific patterns
of stability, progress, and deterioration in service performance over time can be used to support route-
specific interventions.
Figure 11.2
Performance Monitoring Over Time
III. Riders' Attitudes and Transit Performance
The third layer of analysis that supplements the cross-sectional and historical analysis of transit
performance data focuses on the joint analysis of transit riders' attitudes and transit performance. Two
general types of analysis can be accommodated within this context. First, an analysis of the differences
in attitudes across segments of the transit market can help identify opportunities for marketing to
different groups of riders. Second, a comparison of attitudes and transit performance can help identify
riders' misperceptions and identify opportunities for communicating service improvements to transit
riders.
Measuring Customer Satisfaction and Service Quality 105
Figure 11.3 illustrates the differences in perceptions among users and nonusers as reflected on their
ratings of five different aspects of transit service. A rating scale of 0 to 10 was used with higher values
corresponding to more positive perceptions of transit service. As shown in Figure 11.3, current transit
riders rate all aspects of transit service, with the exception of safety while using the system, higher than
nonusers do. The pattern of differences in the respondents' ratings suggests that:
•
the transit agency needs to allocate resources to enhance riders' perception of feeling safe and
secure while riding the transit system;
•
the perception of safety and security among nonusers does not appear to be the primary
reason for not using the transit system;
•
the gap between users' and nonusers' perceptions is greater for "frequency of transit service"
and "transit on-time performance" which are perceived rather positively by current riders of
the transit system; and
•
there are considerable opportunities to improve nonusers' perceptions of transit service along
most of the dimensions of transit service as part of an effort to increase transit ridership.
Although the example of such an analysis is presented at the systemwide level for transit riders and
nonusers it can be further expanded along two additional dimensions. First, route-specific analyses can
be conducted for routes and groups of routes that are of greatest interest to the transit authority. Second,
comparisons of attitudes among market segments can be expanded to account for differences among
frequent and infrequent riders, male and female riders, and riders with different degrees of captivity to
transit. These analyses can provide insight into the appeal of different transit routes to distinct segments
of the market.
Finally, it is possible that the availability of transit performance and survey data at similar points in time
allow comparisons between riders' perceptions and transit performance measures. Such comparisons are
again most meaningful if they can be repeated over time and across different routes of the system. The
availability of such data supports a fourth layer of analysis that can be used to relate patterns of change
in transit performance to changes in riders' perceptions.
The comparisons that can be made allow us to identify cases where service improvements have a
positive impact on riders' perceptions and cases where despite improvements in transit service transit
riders' perceptions continue to remain rather low.
Measuring Customer Satisfaction and Service Quality 106
Figure 11.3
Perceptions of Users and Nonusers for Route A
Figure 11.4 offers an example of comparisons that can be made using historical attitudinal data and
corresponding performance data at the route level to identify the extent to which there is a correlation
between traveler perceptions and transit performance.
The bar chart and the left hand axis illustrates the average ratings given by riders of routes A and E on a
scale of 0 to 10 with higher values corresponding to more positive perceptions of service. The line
graph and the right hand axis correspond to the on-time performance reflecting the percentage of buses
arriving late for the A and B routes at the three study years.
The comparisons that can be made suggest that:
•
riders' ratings for route E are consistently lower than those by riders of route A properly
reflecting the historically better on-time performance of route A;
•
route E riders' ratings of the transit service have dropped over time in a manner that is
consistent with the deteriorating performance of route E;
•
the gap between the ratings for route A and E has widened over time again properly
corresponding to the widening gap in the level of transit on-time performance offered by
each route; and
•
the drop over time in riders' ratings of route A is not consistent with the high level of on-
time performance for route A.
Measuring Customer Satisfaction and Service Quality 107
These observations suggest that riders' evaluations are generally consistent with the level of service that
is provided. The need to improve the on-time performance along route E is supported both by the
existing low level of on-time reliability on that route as well as the low ratings provided by riders. It is
expected that the implementation of such service improvements will enhance route E riders' perceptions
and bring them closer to the ratings provided by riders on route A.
Finally, the apparent inconsistency between the historically high level of on-time reliability for route A
and the steady or decreasing ratings by route A riders suggests that other aspects of the performance for
this route need to be examined more closely. It is possible that due to deterioration in other service
characteristics for route A, riders provide ratings for on-time reliability that are lower than expected.
However, if there are no apparent weaknesses in other aspects of route A service, the implementation of
a marketing campaign aimed at riders of route A may be considered to stress the existing high level of
service.
Figure 11.4
Performance Measure versus Riders' Perceptions
Measuring Customer Satisfaction and Service Quality 108
ENDNOTES
22
Transit Cooperative Research Program, Synthesis 24,
AVL Systems for Bus Transit
,
Transportation Research Board, National Research Council, Washington D.C., 1997.
APPENDIX A
Measuring Customer Satisfaction and Service Quality A-1
Appendix A
CUSTOMER SATISFACTION/DISSATISFACTION RESEARCH
— AN HISTORICAL PERSPECTIVE
Consumer behavior as a distinct discipline dates only from the mid 1960s. Interest in understanding and
tracking specific consumer problems grew dramatically in the late 1970s under the broad label of
consumer satisfaction/dissatisfaction (CS/D) research. Its growth coincided with (and was abetted by) a
growing interest on the part of both government regulators and leaders with the consumer movement in
making the policy formulation process more rational and systematic. Critics of past consumer policy
formulation had argued that it was too often influenced by chance events, letter-writing campaigns,
media publicity, and partisan political agendas. The earliest comprehensive CS/D studies were, in fact,
motivated by the policy planning needs of a public regulatory agency, the Federal Trade Commission
(Technical Advisory Research Program (TARP) 1979), and a private non-profit sector organization,
Ralph Nader's Center for Study of Responsive Law.
Pioneering studies by Handy and Pfaff in the mid 1970s developed raw and weighted indexes of
consumer satisfaction with food products across seven broad food categories. After that point, research
on the topic grew rapidly.
Since 1985, two different patterns have emerged. First, there has been a considerable drop in CS/D
research from a public policy perspective. At the same time, however, there has been substantial growth
in interest in the topic of consumer satisfaction research within the private sector. This has been driven
primarily by the growth of the service sector of the economy where managers have realized that
tracking satisfaction is crucial to success when intangibles such as personal attention and atmospheres
are the "product." A number of private satisfaction tracking services have emerged. Many of these
services have made extensive use of earlier methodological developments in social policy research.
Initial studies on CS/D sought to calibrate the amount and types of dissatisfaction in the marketplace as
a basis for policy planning. This body of research was largely descriptive (TARP 1979). Wide variation
was found across purchase categories. These studies differ widely in the basic measure of
dissatisfaction they used. Some focused on more or less objective measures of "problems," others on
subjective feelings of "dissatisfaction." Some counted any negative experience whatsoever, some only
"serious" dissatisfactions, and some only the most recent problem. Also, there was the issue of
opportunity for problems. Measures did not always control for frequency of purchase. Definitional
problems persist today.
Most of the early studies were based on survey data. An alternate approach was complaints data, data
on the extent to which consumers voluntarily speak up about their dissatisfactions. Such data have the
advantage of not requiring field surveys; however, they are typically biased in two important ways.
First, some types of problems in some types of industries are more likely to be voiced than others, and
some problems are less serious than others, and or less costly than others. Monopolies are often
relatively "immune" to complaining except from a small elite. Still other industries are more
encouraging of complaints. Finally, not all consumers complain. These problems have led researchers in
recent years to fall back on the more costly, but more objective, survey research methods.
Finally, most CS/D research from 1975 to 1985 was conducted within product and goods producing
industries. Only after 1980 were initial concepts and models developed to measure consumer
satisfaction/dissatisfaction within service industries.
Measuring Customer Satisfaction and Service Quality A-2
Appendix A
LITERATURE SEARCH SUMMARY FOR SERVICE QUALITY AND
CUSTOMER SATISFACTION MEASUREMENT —
OUTSIDE TRANSIT INDUSTRY
Conceptual Model of Service Quality and Its Implications for Future Research
, A. Parasuraman,
Valerie A. Zeithaml, and Leonard L. Berry, Journal of Marketing, Fall 1985, Vol. 49, Number 4,
pp. 41-50.
Overview
The attainment of quality in products and services was a pivotal concern of the 1980s. While quality in
tangible goods has been described and measured by marketers, quality in services is largely undefined
and unresearched. The authors attempt to rectify this situation by reporting the insights obtained in an
extensive exploratory investigation of quality in four service businesses and by developing a model of
service quality. Propositions and recommendations to stimulate future research about service quality are
offered.
Quality and measurement are not easily articulated by consumers (Takeuchi and Quelch 1983).
Interpretation and measurement of quality also present problems for researchers. While the substance
and determinants of quality may be undefined, its contribution to increasing market share and return on
investment is unequivocal.
Existing Knowledge About Service Quality
Knowledge about goods quality is insufficient to understand service quality. Three well-documented
characteristics of services — intangibility, heterogeneity, and inseparability — must be acknowledged.
Because they are performances rather than objects, precise manufacturing specifications concerning
uniform quality can rarely be set. Because of intangibility, the firm may find it difficult to understand
how consumers perceive their services and evaluate service quality (Zeithaml 1981).
Second, services, especially those with high labor content, are heterogeneous: their performance often
varies from producer to producer, from customer to customer, and from day to day. Consistency of
behavior from service personnel (e.g., uniform quality) is difficult to ensure (Booms and Bitner 1981)
because what the firm intends to deliver may be entirely different from what the customer receives.
Third, production and consumption of many services are inseparable (Carmen and Langeard 1980,
Upah 1980). In labor intensive services, quality occurs during service delivery, usually in an interaction
between the client and front-line personnel.
Service quality literature traditionally agrees that service quality is a measure of how well the service
level delivered matches customer expectations. Delivering quality service means conforming to
customer expectations on a consistent basis. (Lewis and Booms 1983)
Measuring Customer Satisfaction and Service Quality A-3
Appendix A
Insights from Exploratory Qualitative Investigation
A set of discrepancies or gaps exists regarding executive perceptions of service quality and the tasks
associated with service delivery to consumers. These gaps can be major hurdles to attempting to deliver
a service which consumers would perceive as being high quality. Figure A.1 on the following page
shows the five gap areas identified.
These are:
GAP 1:
Consumer expectation — management perception gap
Discrepancies between executive perceptions and consumer expectations. Service
firm executives may not always understand what features denote high quality to
consumers in advance, what features a service must have in order to meet
consumer needs, and what levels of performance on those features are needed to
deliver high quality service.
GAP 2:
Management perception — service quality specifications
Constraints (resources, or market conditions) which prevent management from
delivering what the consumer expects, or the absence of total management
commitment to service quality.
GAP 3:
Service quality specifications — service delivery gap
Difficulty in standardizing employee performance even when guidelines exist for
performing services well and treating consumers correctly.
GAP 4:
Service delivery — external communications gap
Media advertising and other communications by a firm can affect consumer
expectations. Promising more than can be delivered will raise initial expectations
but lower perceptions of quality when the promises are not fulfilled. Also firms
can neglect to inform consumers of special efforts to ensure quality that are not
visible to consumers thereby affecting consumer perceptions of the delivered
service.
GAP 5:
Expected service — perceived service gap
How consumers perceive the actual service performance in the context of what
they expected. The quality that a consumer perceives in a service is a function of
the magnitude and direction of the gap between expected service and perceived
service.
Measuring Customer Satisfaction and Service Quality A-4
Appendix A
Figure A.1
Service Quality Model
A Quality Service Model
The foundation of the model is the set of gaps shown in Figure A.1. Service quality as perceived by a
consumer depends on the size and direction of GAP 5 that, in turn, depends on the nature of the gaps
associated with the design, marketing, and delivery of services. The gaps on the marketer side of the
equation can be favorable or unfavorable from a service quality perspective. That is, the magnitude and
direction of each gap will have an impact on service quality.
Measuring Customer Satisfaction and Service Quality A-5
Appendix A
The Perceived Service Quality Component
This exploratory investigation suggests that, regardless of the type of service, consumers used basically
similar criteria in evaluating service quality. These criteria seem to fall into 10 key categories that are
labeled "service quality determinants." These determinants are listed in Table A.2 below. Overlap
among the 10 determinants may exist.
Table A.2
Determinants of Service Quality
1 RELIABILITY involves consistency of performance and dependability.
2 RESPONSIVENESS concerns the willingness or readiness of employees to provide
service. It also involves timeliness of service.
3 COMPETENCE means possession of the required skills and knowledge to perform
the service.
4 ACCESS involves approachability and ease of contact.
5 COURTESY involves politeness, respect, consideration, and friendliness of contact
personnel.
6 COMMUNICATION means keeping customers informed in language they can
understand and listening to them. It may mean that the company has to adjust its
language for different consumers — increasing the level of sophistication with a
welleducated customer and speaking simply and plainly with a novice.
7 CREDIBILITY involves trustworthiness, believability, and honesty. It involves
having the customer's best interests at heart.
8 SECURITY is the freedom from danger, risk, or doubt.
9 UNDERSTANDING/KNOWING THE CUSTOMER involves making the effort to
understand the customer's needs.
10 TANGIBLES includes the physical environment and representations of the service.
It is quite possible that the relative importance of the 10 determinants in molding consumer expectations
(prior to service delivery) may differ from their relative importance vis-à-vis consumer perceptions of
the delivered service. Figure A.3 on the following page indicates that perceived service quality is the
results of the consumer's comparison of expected service with perceived service.
Measuring Customer Satisfaction and Service Quality A-6
Appendix A
Figure A.3
Determinants of Perceived Service Quality
Two of the determinants which consumers appear to have difficulty evaluating are
competence
(the
possession of the required skills and knowledge) and
security
(freedom from danger, risk, or doubt).
Consumers are probably never really certain of these attributes, even after experiencing the service.
Perceived service quality is posited to exist along a continuum ranging from ideal quality to totally
unacceptable quality, with some point along the continuum representing satisfactory quality. The
position of a consumer's perception of service quality on the continuum depends on the nature of the
discrepancy between the expected service (ES) and perceived service (PS). When ES > PS perceived
quality is less than satisfactory; when ES = PS perceived quality is satisfactory; and when ES < PS,
perceived quality is more than satisfactory and will tend toward ideal quality.
Although the preliminary research showed that consumers used similar criteria in judging service
quality, the group participants differed on the relative importance of those criteria to them, and their
expectations along the various quality dimensions. Research needs to determine whether identifiable
service quality segments exist and whether and in what ways consumer expectations differ.
Measuring Customer Satisfaction and Service Quality A-7
Appendix A
Takeuchi, Hirotaka and John A Quelch (1983), "Quality Is More Than Making a Good Product,"
Harvard Business Review
, 61 (July-august), 139-145.
Zeithhaml, Valerie A. (1981) "How Consumer Evaluation Processes Differ Between Goods and Services,"
in
Marketing of Services
, J. Donnelly and W. George, eds., Chicago: American Marketing, pp. 186-190.
Carmen, James M. and Eric Langeard (1980), "Growth Strategies of Service Firms,"
Strategic
Management Journal
, 1 (January-March), pp. 7-22.
Upah, Gregory D. (1980), "Mass Marketing in Service Retailing: A Review and Synthesis of Major
Methods,"
Journal of Retailing
, 56 (Fall), pp. 56-76.
Lewis, Robert C. and Bernard H. Booms (1983), "The Marketing Aspects of Service Quality," in
Emerging Perspectives on Services Marketing,
L. Berry, G. Shostack and G. Upah, eds., Chicago:
American Marketing, pp. 99-107.
Churchill, G.A. Jr. and C. Suprenaut (1982), "An Investigation into the Determinants of Customer
Satisfaction,"
Journal of Marketing Research
, 19 (November), pp. 491-504.
A National Customer Satisfaction Barometer: The Swedish Experience
, Claes Fornell, Journal of
Marketing, January 1992, Volume 56, Number 1, pp. 6-21.
Overview
Many individual companies and some industries monitor customer satisfaction on a continual basis, but
Sweden is the first country to do so on a national level. The annual Customer Satisfaction Barometer
(CSB) measures customer satisfaction in more than 30 industries and for more than 100 corporations.
The new index is intended to be complementary to productivity measures. Whereas productivity
basically reflects quantity of output, CSB measures quality of output (as experienced by the buyer). The
author reports the results of a large-scale Swedish effort to measure quality of the total consumption
process as customer satisfaction. Efforts to measure customer satisfaction on a nationwide basis are now
underway in several other countries including the U.S., Japan, and Norway.
The U.S index is the result of a joint venture between the American Quality Foundation and the
University of Michigan Business School. The significance of customer satisfaction and its place within
the overall strategy of the firm are discussed.
Inherent Differences Among Industry and Firm Customer Satisfaction Levels
Substantial literature suggests that market share leads to profitability (see Buzzell and Gale 1987).
Customer satisfaction also is believed to lead to profitability (Business International 1990).
Traditionally, much more effort is devoted to the offense for customer acquisition then to the defense to
protect the present customer base (Fornell and Wernerfelt 1987, 1988). However, in the face of slow
growth, a good defense is critical. Defensive strategy involves reducing customer exit and switching.
One way of accomplishing this objective is to have highly satisfied customers. While improving market
share and improving customer satisfaction individually result in higher profitability, it is far from
certain that market share and customer satisfaction themselves are positively correlated. If an industry
or company enjoys high levels of customer satisfaction, decreases in market share (perhaps because of a rise
in cost) are less likely to affect profitability. Decision making in this situation is a combination of price
Measuring Customer Satisfaction and Service Quality A-8
Appendix A
and quality. However, it is more difficult for a firm with a large market share to also have a high
average level of customer satisfaction, especially if customer needs or wants are heterogeneous.
The ideal point conceptualization as one aspect of customer satisfaction suggests a new hypothesis
about market structure and customer satisfaction. The contention is that the monopoly will have a lower
score on customer satisfaction indexes than other non-monopoly industries, if it faces a heterogeneous
demand. Lower customer satisfaction in this case is partially a reflection of the difficulty in serving a
heterogeneous market with a limited variety of service or product offerings. On the other hand, we
would expect that industries characterized by a good fit between the levels of demand and supply
heterogeneity (homogeneity) to have higher customer satisfaction ratings than those with a poor fit.
Industries, including monopolies, that supply a high quality homogeneous product to a homogeneous
market should have high satisfaction.
Also explored is the impact of customer satisfaction on repeat business and customer loyalty in different
industries. Loyal customers are not necessarily satisfied customers, but satisfied customers tend to be
loyal customers. Customer switching barriers comprise a host of factors that also bring about retention.
Switching barriers make it more costly for the customer to switch to another supplier or mode.
Transaction costs, learning costs, loyal customer discounts, customer habit, emotional cost, and
cognitive effort, coupled with financial, social, and psychological risks on the part of the buyer, all add
up to switching barriers. However, previously insulated organizations become vulnerable, for they are
seldom well prepared and have not made the investments in quality and customer satisfaction necessary
to prevent customer exit. Low barriers and weak customer satisfaction force the company to compete on
price. With high satisfaction there is less price sensitivity.
Uses of the Sweden Customer Satisfaction Barometer (CSB)
To combine premises, the proposition that evolves from the ideal-point model and the switching-barrier
effect suggests that customer satisfaction should be lower in industries where repeat buyers face high
switching costs and where the industry offers a homogeneous product to a heterogeneous market. With
this presumption in mind, the CSB in Sweden offers the following information:
•
industry comparisons
•
comparisons of individual firms with the industry average
•
comparison over time
•
predictions of long-term performance
•
Though empirical evidence is limited, increases in customer satisfaction are generally
believed to: (1) shift the demand curve upward and/or make the slope of the curve steeper
(i.e. lower price elasticity, higher margins), (2) reduce marketing costs (customer
acquisition requires more effort, (3) reduce customer turnover, (4) lower employee
turnovers (satisfied customers affect the satisfaction of front-line personnel), (5) enhance
reputation (positive customer word of mouth), (6) reduce failure costs (handling customer
complaints).
•
answers to specific management questions
(Such as the effects of overall quality and price, the impact of customer expectations, the
quality increase necessary to retain dissatisfied customers, price sensitivity, switching
patterns, customer complaints, and effects of word of mouth.)
Measuring Customer Satisfaction and Service Quality A-9
Appendix A
Highlights of CSB Measurement
The literature on customer satisfaction/dissatisfaction suggests that satisfaction is an overall postpurchase
evaluation. There is no consensus on how to measure it. Hausknecht (1990) identifies more than 30
different measures that have been used in previous research. There are three different dimensions: (1)
general satisfaction (as in the studies by Moore and Shuptrine 1984; Oliver and Bearden 1983; Oliver
and Westbrook 1982; and Westbrook 1980), (2) confirmation of expectations (as in studies by Oliver
1977; Swan, Trawick, and Carroll 1981), and (3) the distance from the customer's hypothetical ideal
product (Tse and Wilton 1988, and Sirgy 1984). Customer satisfaction for the CSB is defined as a
function of these three indicators, thus the fallibility of measures is acknowledge and taken into account.
The traditional view of satisfaction/dissatisfaction as the discrepancy between perceived performance
and expectation (P-E) is not dismissed
a priori
in CSB. However, CSB measurement allows for the
possibility of dissatisfaction even when expectations are confirmed (a negative correlation). For
example, if low quality is expected but the product is purchased nevertheless (because of supply
restrictions or price), the expectations are confirmed. Clearly, the fact that expectations are confirmed is
not sufficient for satisfaction.
Presumably, customers take both price and quality into account. To avoid compounding the two, for the
CSB, each was measured in the light of the other — by price (given quality) and quality (given price).
For most industries surveyed, sample frames were drawn via random digit dialing with screening for
customer status. In no cases were company customer lists used as sample frames. Hence data were
costly but presumably more objective.
Almost all customer satisfaction research is hampered by highly skewed distributions for satisfaction.
For example, in studies ranging from shoes to medical care, more than 80% of the customers were
satisfied. Only in captive markets might repeat buyers be dissatisfied in general. Skewness is a problem,
but it is a statistical one. Highly skewed variable distributions do not lend themselves to conventional
tests of significance and, what is equally serious, lead to downward biases in correlation analysis, low
reliability, and sometimes misleading arithmetic means. In CSB, the problem of skewness was handled
by (1) extending the number of scale points (usually 5 or 7) to 10 to allow respondents to make finer
discriminations, (2) using a multiple-indicator approach for greater accuracy, and (3) estimating via a
version of partial least squares (PLS).
CSB Results
The results of the CSB fit the reasoning presented. Overall, CSB scores are significantly higher in
industries where heterogeneity/homogeneity in demand is matched by the supply. Staple foods and
automobiles score at the top of the CSB; the police force and television broadcasting are at the bottom.
(Transportation services were not measured as a part of the Sweden CSB.) Overall, it is noteworthy that
services score lower than products, both among monopolies and among competing firms.
The Effect on Customer Loyalty
Just as price elasticity varies among firms and industries, so does "customer satisfaction elasticity." It is
very important to determine how sensitive the present customer base is to satisfaction. In view of the
current business emphasis on quality, one may well get the impression that quality and customer
satisfaction are equally important to all firms or industries. Customer satisfaction is more important (for
Measuring Customer Satisfaction and Service Quality A-10
Appendix A
loyalty) in some industries than in others. Industries with low elasticities are those in which switching
costs are high (police, postal services, etc.)
The most meaningful measurement of quality is how it affects customer satisfaction. Changes in
satisfaction are predictors of future performance.
Buzzell, Robert D. and Bradley T. Gale (1987),
The PIMS Principles
, New York: The Free Press.
Business International (1990),
Maximizing Customer Satisfaction: Meeting the Demands of the New
Global Marketplace,
Research Report. New York: Business International Corporation.
Fornell, Claes and Birger Wernerfelt (1987), "Defensive Marketing Strategy by Customer Complaint
Management: A Theoretical Analysis,"
Journal of Marketing Research
, 24 (November) pp. 337-346.
Hausknecht, Douglas R. (1990), "Measurement Scales in Consumer Satisfaction/Dissatisfaction,"
Journal of Consumer Satisfaction, Dissatisfaction, and Complaining Behavior
, 3, pp. 1-11.
Moore, Ellen M. and F. Kelly Shuptrine (1984), "Disconfirmation Effects on Consumer Decision
Making Processes,"
Advances in Consumer Research
, Vol. 11, Thomas C. Kinnear, ed. Ann Arbor, MI:
Association for Consumer Research, pp. 299-304.
Oliver, Richard L. and William O. Bearden (1983), "The Role of Involvement in Satisfaction
Processes"
Advances in Consumer Research
, Vol. 10, Richard P. Bagozzi and Alice M. Tybout, eds.
Ann Arbor, MI: Association for Consumer Research, pp. 250-255.
Oliver, Richard L. and Robert A. Westbrook (1982), "The Factor Structure of Satisfaction and Related
Postpurchase Behavior," in
New Findings in Consumer Satisfaction and Complaining
, Ralph L. Day
and H. Keith Hunt, eds. Bloomington: Indiana University, pp. 11-14.
Westbrook, Robert A. (1980), "A Rating Scale for Measuring Product/Service Satisfaction,"
Journal of
Marketing
, 44 (Fall), pp. 68-72.
Oliver, Richard L. (1977), "Effect of Expectation and Disconfirmation on Post-Purchase Product
Evaluations: An Alternative Interpretation,"
Journal of Applied Psychology,
62 (4) pp. 480-486.
Swan, John E., Frederick Trawick, and Maxwell G. Carroll (1981), "Effect of Participation in
Marketing Research on consumer Attitudes Toward Research and Satisfaction With Service,"
Journal
of Marketing Research
, 18 (August), pp. 365-363.
Tse, David K. And Peter C. Wilton (1988), "Models of Consumer Satisfaction Formation: An
Extension,"
Journal of Marketing Research
, 25 (May), pp. 204-214.
Sirgy, Joseph M. (1984), "A Social Cognition Model of Consumer Satisfaction/Dissatisfaction,"
Psychology and Marketing
, I (2), pp. 27-43.
Measuring Customer Satisfaction and Service Quality A-11
Appendix A
Expectations, Performance Evaluation, and Consumers' Perception of Quality
, R. Kenneth Teas,
Journal of Marketing, October 1993, Volume 57, Number 4, pp. 18-34.
Overview
The author examines conceptual and operational issues associated with the measurement framework
defined as customer "perceptions-minus-expectations" (P-E) identified by Parasuraman, Zeithaml, and
Berry (1985). The examination indicates that the P-E service gap premise is of questionable validity
because of a number of conceptual problems involving the (1) conceptual definition of expectations, (2)
theoretical justification of the expectations component of the P-E framework, and (3) measurement
validity of the expectation (E) and revised expectation (E*) measures specified in the published service
quality literature.
The P-E model and two alternative perceived quality modes that are designed to address the problems
associated with the P-E model are empirically tested and the implications of the conceptual issues
examined in the study and of the empirical findings are explored.
Definition Problems
Alternative definitions of expected or ideal service exist. Conceptualizing service expectation as ideal
standards is a problem under each of the interpretations examined.
Classic attitudinal model point interpretation (Ginter 1974; Green and Srinivasan 1978).
In these
models, the ideal point is the perfect or utility maximizing level of the attribute. For example, if the
attribute has a non-maximum ideal point, once the ideal point is reached "there are negative utility
returns for further increases in the attribute" (Lillien, Kotler, and Moorthy 1992, p.9). Favorableness of
an evaluation of an attitude object is positively related to the closeness of the object to the ideal object.
Feasible ideal point interpretation.
A second interpretation of the service quality ideal standard is that it
represents a feasible level of performance under ideal circumstances. However, the "feasible ideal
point" conception of E is not compatible with the service quality P-E measurement specification, when
finite classic ideal point attributes are involved.
Operational Definition Problems
Empirical research has identified important problems concerning the operationalization of the service
expectation (E) concept. Respondents may assign unrealistically high ratings to the E response scales.
Carmen (1990) questions the validity of expectation measures when consumers do not have "well-
formed expectations." Research by Teas (1993) indicates that a considerable portion of the variance in
responses to the E scale is because of variance in respondents' interpretations of the question being
asked, rather than to variance in respondents' attitudes.
To correct respondents high ratings on E scales, Parasuraman, Berry, and Zeithaml (1990) proposed a
revised expectation (E*) measure, based on ratings of the attribute's "essentialness" for excellent
service. However, using the revised definition of expectation (E*), in conjunction with the P-E
measurement specification, suggests that high performance on essential attributes (high E* scores)
reflects lower quality than high performances on attributes that are less essential (low E* scores). This
measurement result is illogical.
Measuring Customer Satisfaction and Service Quality A-12
Appendix A
Results of Testing Alternative Perceived Quality Frameworks
The results suggest a considerable portion of the variance of service quality expectation measures may
be because of respondents' misinterpretations of the question or the scales. The empirical testing also
indicates that the inclusion of attribute weights in the P-E or other alternative frameworks does not
improve the validity of the models. This result is similar to the findings of other research that indicates
importance weights often do not increase, and may decrease, the predictive validity of multiattribute
models (Bass and Wilkie 1973).
The conceptual and operational definition problems with the P-E "gap" framework and alternative
tested models create ambiguity concerning the interpretation and theoretical justification of these
perceived quality concepts.
Parasuraman, A., Leonard L. Berry, and Valerie A. Zeithaml (1990),
An Empirical Examination of
Relationships in an Extended Service Quality Model
, Cambridge, MA: Marketing Science Institute.
Parasuraman, A., Valerie A. Zeithaml, and Leonard L. Berry (1985), "A Conceptual Model of Service
Quality and Its Implications for Future Research,"
Journal of Marketing
, 49 (Fall) pp. 41-50.
Ginter, James L. (1974), "An Experimental Investigation of Attitude Change and Choice of a New
Brand,"
Journal of Marketing Research
, 11 (February), pp. 30-40.
Green, Paul E. And V. Srinivasan (1978), "Conjoint Analysis in Consumer Research: Issues and
Outlook,"
Journal of Consumer Research
, 5 (September), pp. 103-23.
Lillien, Gary L., Philip Kotler, and K. Sridhar Moorthy (1992),
Market Models
, Englewood Cliffs, NJ:
Prentice Hall, Inc.
Carmen, James M. (1990), "Consumer Perceptions of Service Quality: An Assessment of the
SERVQUAL Dimensions,"
Journal of Retailing
, 66 (Spring) pp. 33-55.
Teas, R. Kenneth (1993), "Consumer Expectations and the Measurement of Perceived Service Quality,"
Journal of Professional Services Marketing
, 8 (2), pp. 33-54.
Bass, Frank and William L. Wilkie (1973), "A Comparative Analysis of Attitudinal Predictions of
Brand Preference,"
Journal of Marketing Research
, 10 (August) pp. 262-269.
Competing Based on the Customer's Hierarchy of Needs
, Doug Schaffer, National Productivity
Review (Summer 1995) pp. 9-15.
Even when companies improve their performance, they often have difficulty achieving real competitive
advantage in the face of often astounding operational improvements, since most customers just do not
seem very excited. This is largely because customers have been excluded from improvement efforts to
date. For companies to better perform in ways that matter to their customers, they must know why
customers buy from them in the first place. This represents a shifting hierarchy of needs that requires
companies to improve their performance in ways that will make their customers sit up and take notice.
It is typical for companies to launch improvement programs in response to competitive pressures, then
several years down the road report improvements that primarily affect internal operations. Published
reports often list fewer engineering problems or defects, streamlined purchasing processes, lower
Measuring Customer Satisfaction and Service Quality A-13
Appendix A
receivables, improved employee safety, etc. All are worthy goals and certainly contribute to a healthy
balance sheet, but may be only of marginal interest to customers. Many programs to improve corporate
performance are more effective in reducing costs and improving profitability than spurring growth and
increasing market share.
Most companies have a rudimentary understanding of why customers buy their product or select their
service. However, most would be hard-pressed to explain how much of a customer's decision is based
on service characteristics, value, or reputation.
In his 1954 work,
Motivation and Personality,
Abraham Maslow proposed a theory of human
motivation characterized by a hierarchy of needs. Inserting the needs of the customer into Maslow's
model yields a model of customer motivation (Exhibit A.4).
Exhibit A.4
Hierarchy of Customer Needs
First on the list is how closely a product or service matches what the customer needs. The product must
be available when the customer needs it. Customers expect a good value — the relationship of the cost
to perceived benefit. Customers also expect quality and reliability. They never want to be stranded,
inconvenienced, or endangered by products or services. (Customers employ a standard of zero
tolerance.) Customers want to be treated well, never put down or demeaned. Customers also have come
to expect an occasional value-added extra that makes it easier to do business with a company and
improves the cost/benefit ratio. Finally, customers faced with a problem expect the supplier to recover,
to fix the problem without harassing the customer.
Measuring Customer Satisfaction and Service Quality A-14
Appendix A
Once customers have decided to purchase a product or service from a particular supplier, their overall
satisfaction and willingness to do business with that supplier in the future rest with the supplier's ability
to satisfy needs at the top of the hierarchy. Those who fail to manage the customer relationship at the
top of the hierarchy loose customers despite the value, quality and availability of their products.
Eventually, they create a reputation for themselves that waves off potential customers and erodes their
sales base.
Any performance improvement effort should begin with an analysis of the company's performance
against its customers' hierarchy of needs. Strengths and weaknesses should be identified and priorities
set based on this analysis.
Best Practice for Customer Satisfaction in Manufacturing Firms
, Griffin, Abbie, Greg Gleason,
Rick Preiss, and Dave Shevenaugh, Sloan Management Review (Winter 1995)
The most frequently measured Customer Satisfaction (CS) variables were expressed as numbers. Most
companies use simple scales that assume satisfaction ranges linearly between 0 and 10 or 0 and 100.
More elaborate measures of customer satisfaction that look at performance relative to expectations, or
disconfirmation measures of satisfaction, are not frequently used. CS measures are often upwardly
biased, not linear. Customers are the subset of the total population who are already somewhat satisfied
with products and services, so the response population does not form a normal distribution about the
midpoint, which is what most analytical procedures for linear scales assume. However, if you cut the
scale off at 5 and consider only the responses above 5, the response distribution of the "average" firm
might be much closer to a normal distribution about the new midpoint, 7.5 of 75 percent. This truncated
scale would more closely conform to the standard statistical assumptions for linear interval scales.
The process of linking goals to performance through measuring CS is exploratory and preliminary at
even the most forward-thinking companies. First, companies must formalize and quantify the
relationship between CS and firm performance. By determining how CS improves performance or what
specific CS components correlate with different improvements, corporations can focus on only the most
effective endeavors, allowing them to become more efficient in implementation.
Delivering CS is at an early evolutionary state in most U.S. firms. Most firms are not focused on
satisfying customers, even though research now correlates CS with improved performance. A firm's CS
implementation process must reflect the needs of individual customer segments, and the overall
program must be flexible enough to allow each business unit to develop measures and processes that fit
its management needs.
Avoid Top Box Problem by Using Another Box
, Dan Prince, President, Prince Marketing Research,
Marketing News, June 1995, p. H-32.
This article suggests an alternative to the "top box problem" when measuring customer satisfaction.
This alternative uses a three-point scale. Respondents are asked to rate overall satisfaction, and
satisfaction on individual attributes, as (1) much better than expected, (2) about as expected, and (3)
worse than expected. If a customer chooses (1), it means they are expressing delight with the product or
service, not just satisfaction. The research showed that if a customer is delighted, there is a 90% chance
they will purchase the product or service again. If (2) is chosen, the customer is expressing satisfaction
with a low product or brand loyalty. And finally, if (3) is chosen, the customer is dissatisfied with the
product or service.
Measuring Customer Satisfaction and Service Quality A-15
Appendix A
This alternative approach provides two benefits:
•
it measures a customer's view against his or her expectation, and
•
it gets rid of the top box problem of skewness — bias to the top of the scale.
Finally, using this alternative approach enables management to understand how well their product or
service actually measures against their customers' expectations.
DEFINITIONS:
Top box problem:
Most customers — if they are still your customer — will tend to give overall satisfaction scores
that fall into one of the top boxes on your answer sheet, usually, "excellent" or "good" (7 to 10
on a 10-point scale).
A second variation of the top box problem is that when respondents are asked, "How satisfied
are you with X," followed with a request to rate X on a scale of importance, most customers will
say each variable is either "very important" or "important."
Rational and Adaptive Performance Expectation in A Customer Satisfaction Framework
, Johnson,
Michael D., Eugene W. Anderson, and Claes Fornell, Journal of Consumer Research, Inc., Vol.
21, March 1995, pp. 595-707.
There is an extensive and growing body of research on customer satisfaction that focuses primarily on
disaggregate or individual-level satisfaction with particular goods or services. Relatively little attention
has been paid to the determinants of market-level satisfaction, which is defined here as the aggregate
satisfaction of those who purchase and consume a particular product offering (e.g., Ford Escort owners
or Federal Express users). Studying customers in the aggregate is one way to establish empirical
generalizations in the domain of satisfaction research.
The modeling of customer satisfaction depends critically on how satisfaction is conceptualized. Two
general conceptualizations of satisfaction exist: transaction-specific satisfaction and cumulative
satisfaction. Consumer researchers are often concerned with satisfaction as an individual, transaction-
specific measure or evaluation of a particular product or service experience. Alternately, satisfaction is
viewed as a cumulative, abstract construct that describes customers' total consumption experience with
a product or service. This conceptualization of satisfaction is more consistent with existing views.
Satisfaction is a customer's overall evaluation of his or her purchase and consumption experience to
date. Measures of this satisfaction component can serve as a common denominator for describing
differences across firms and industries, while transaction-specific evaluations provide information only
about shortrun product or service encounters. Cumulative satisfaction is a fundamental indicator of a
firm's (or market's) current and long-run performance.
To construct indices of customers' satisfaction at the market level for individuals who purchase and
consume a particular product or service offerings, three measures are proposed: aggregate expectations,
perceived performance, and satisfaction. Expectations are measured first by asking customers how well
they expected the product or service to perform. Two measures are then collected to operationalize
performance (perceived quality relative to price paid and a rating of how much the customer has paid
Measuring Customer Satisfaction and Service Quality A-16
Appendix A
relative to how well the product or service has performed). Finally, three measures are used to
operationalize satisfaction: overall satisfaction, confirmation of expectations, and distance from the
customer's hypothetical ideal product or service in the industry. Three-stage (extrapolative, adaptive,
and rational) least square estimates are used to determine market expectations and satisfaction. In every
case, satisfaction is positively affected by both performance and expectations.
The results show that there is a significant carryover effect for customer satisfaction from period to
period. That is, market satisfaction is a relatively stable, cumulative phenomenon that changes gradually
over time.
Green, Paul E. and Tull, Donald S. Research for Marketing Decisions; 3rd edition; Prentice-Hall,
Inc. 1975 (Englewood Cliffs, New Jersey), pp. 478-484.
In a typical customer satisfaction study, respondents evaluate overall satisfaction, followed by ratings
on many individual attributes. A key question for researchers is which attributes are most important in
determining overall satisfaction. Not all attributes have equal impact. A method of prioritizing is needed
to allocate limited resources more efficiently.
Researchers have suggested many procedures for dealing with this problem. Several are considered by
Green and Tull (1975), Hauser (1991), and
The Maritz Marketing Research Report
(1993). Work
continues in this area; no true "answer" for all applications has emerged. However, derived importance
measures are usually preferred over stated importance measures.
Stated importance measures ask respondents to explicitly state their perception of the importance of
each attribute, usually using a 10-point scale. The results of this method can be straightforwardly
interpreted; however, the results can be few, if any, statistical differences among attributes, so the aim
of the method — to prioritize attributes — is thwarted. (How does a mean 7.8 rating differ specifically
from a mean 7.5 rating?)
Derived importance
methods rely on the statistical association between ratings (predictors) and an
overall rating. The importance of an attribute is statistically determined from this relationship. Green
and Tull consider four derived importance measures. If, in the very unlikely case that all attributes are
uncorrelated with each other, all four yield identical measures of relative importance. Measures
discussed by Green and Tull are:
•
Bivariate (Pearson) correlation:
This measure has the advantages of familiarity and
simplicity. Unlike the other three, it's not affected by adding or deleting other attributes in
a regression equation; however, joint effects with other attributes go undiscovered.
•
Standardized regression coefficient or beta weight:
Model misspecifications and the
influence of other attributes in the regression model are particularly troublesome in this
approach. This measure can be very unstable.
•
The product of the beta weight and the corresponding Pearson correlation:
This measure
is a compromise between the two former measures.
•
The coefficient of part determination:
This model represents an incremental gain in
predictive power but is adversely influenced by the inclusion or exclusion of particular
attributes in the model.
Measuring Customer Satisfaction and Service Quality A-17
Appendix A
All four measures exhibit limitations. However, an important consideration is that it is common in
customer satisfaction research for attributes to be correlated — sometimes highly — with each other.
This makes it difficult to measure the separate effects of the individual attributes on overall satisfaction.
The latter three measures are all subject to instability when attributes are highly correlated. When
interrelations exceed .5 — a fairly frequent occurrence for customer satisfaction data — the beta
weights can shift dramatically.
Moreover, the latter three measures can also be affected by the addition or deletion of particular
attributes to the regression model. The multiple regression model used for the latter three measures must
have the correct functional form.
In the face of these problems, use of the first measure, simple bivariate correlation is recommended.
However, considering each attribute in isolation is also unrealistic.
Green and Tull offer an alternative to combat multicolinearity; namely, to transform the original
attributes into an uncorrelated set of new variables using the technique of principal component analysis.
The principal components reveal the colinearity in the data while allowing analysis such as stepwise
multiple regression to be performed without multicolinearity — and without deleting one of more of the
highly correlated attributes.
This approach has the added advantage of using multivariate techniques that can be explained and
described.
APPENDIX B
Measuring Customer Satisfaction and Service Quality A-19
Appendix B
MODERATOR'S GUIDE
A. Introduction
This is a nationally based study to explore customer requirements for transit service. We want to know
how riders view quality of service. What features of service are important? What are the most
troublesome aspects of riding transit? How can a transit agency best improve its service? These are the
kinds of questions we want to ask. We also want to know how you define quality service and get your
reactions to various ideas about how a transit agency can monitor their quality of service. Let's start by
having each of you introduce yourself.
1. Current transit usage, frequency of usage, trip purposes, how long have they been using transit, cars
in the household, primary reasons for using transit over other modes of transportation.
B. Description of Ideal Transit Service
1. How would you define the 'ideal' transit service?
2. What would you change about your current transit service to make it closer to the 'ideal'?
3. How do you define low quality transit service?
C. Discussion of Basic Transit Requirements
1. What are the basic requirements for transit service?
2. How would you define the dimensions of service quality?
— safety
—comfort
— ease of using the system
— convenience
— performance/reliability
— facilities
—value
Measuring Customer Satisfaction and Service Quality A-20
Appendix B
D. Review of Specific Transportation Attributes
SAFETY
1. What does "safety" mean when using rail/bus?
2. Here are some features related to "safety" mentioned by others. How important is each in your
decision to use transit?
— Safety from crime while riding
— Safety at stations/bus stops
— Safety related to the behavior of other persons
— Safety related to the rail/bus operation
3. Are there other aspects of "safety" we failed to discuss?
COMFORT
1. How do you define "comfort" when riding rail/bus?
2. Here are some features related to "comfort" mentioned by others. How important is each in your
decision to use transit?
— Availability of seating at the station/bus stop
— Availability of seats on the train/bus
— Smoothness of the train/bus ride
— Comfort of the seats
— Degree of crowding on the train/bus
— Comfortable temperatures on the train/bus
— Availability of handrails/grab bars
3. Are there other aspects of "comfort" we failed to discuss?
Measuring Customer Satisfaction and Service Quality A-21
Appendix B
EASE OF USING THE SERVICE
1. How would you define an "easy" system to ride?
2. Here are some features related to "ease of using a service" mentioned by others. How important is
each in your decision to use transit?
— Knowing when trains/buses arrive and depart
— Availability of information at a station (RAIL ONLY)
— Availability of printed schedules
— Ease of getting information by telephone
— Courtesy/helpfulness of ticket agents (RAIL ONLY)
— Ease of purchasing tickets/passes/tokens
— Visibility of station names from on the train (RAIL ONLY)
— Visibility of train/bus names/route numbers/colors from the outside
— Ease of getting on/off train/bus
— Ease of paying fare
— Ease of making connections/transfers
— Knowledgeable and courteous conductors/drivers on-board
— Availability of information about delays from conductors/drivers
— Clear/timely stop announcements
3. Are there other aspects of "ease of use" we failed to discuss?
CONVENIENCE
1. What does "convenience" mean when riding rail/bus?
2. Here are some features related to "convenience" mentioned by others. How important is each in
your decision to use transit?
— Availability of stations/bus stops close to home
— Availability of stations/bus stops close to work
— Availability of stations/bus stops close to shopping
— Availability of parking at stations/bus stops
3. Are there other aspects of "convenience" we failed to discuss?
Measuring Customer Satisfaction and Service Quality A-22
Appendix B
PERFORMANCE/RELIABILITY
1. What does "performance and reliability" have when riding rail/bus?
2. Here are some features related to "performance and reliability" mentioned by others. How
important is each in your decision to use transit?
— Frequency of service
— Travel time by train/bus
— On-time performance
— Wait time when transferring
3. Are there other aspects of "performance and reliability" we failed to discuss?
CONDITION OF VEHICLES AND FACILITIES
1. How do you define vehicles and facilities in good condition?
2. Here are some features related to the condition of vehicles and facilities mentioned by others. How
important is each in your decision to use transit?
— Cleanliness of the train/bus interior
— Trains/buses clean of graffiti
— Stations/bus shelters clean of graffiti
— Cleanliness of train stations/bus stops
3. Are there other aspects of the condition of vehicles and facilities we failed to discuss?
VALUE
1. How would you define "value" with respect to riding rail/bus?
2. Here are some features related to "value" mentioned by others. How important is each in your
decision to use transit?
— Cost of a one-way ride
— Cost of a transfer
— Availability of discounted fares, e.g., senior citizens, students
— Availability of volume discounts, e.g., monthly passes
— Cost of parking at stations/bus stops
3. Are there other aspects of "value" we failed to discuss?
Measuring Customer Satisfaction and Service Quality A-23
Appendix B
E. Defining Service Quality
1. How should a transit agency measure/monitor its own quality?
2. What information should a transit agency collect and use to monitor its quality?
3. Reactions to collecting the following quality measures.
— percent of trips on-time
— headway consistency
— breakdowns
— communication measures
— number of accidents
— vehicle availability
If I told you that the reports that 92% of all trips on the
line arrive within four
minutes of their scheduled arrival time, what does that mean to you?
What does it mean if I say that on average buses break down every 3,500 miles?
4. Do these measures present an honest picture of the quality of service provided by ?
5. How should a transit agency demonstrate that its customers come first?
F. Closing
1. What does quality of transit service mean to you as a rider?
Measuring Customer Satisfaction and Service Quality A-24
Appendix B
BASIC DIMENSIONS
!
safety
!
comfort
!
ease of using the system
!
convenience
!
performance/reliability
!
facilities
!
value
SAFETY
!
Safety from crime while riding
!
Safety at stations/bus stops
!
Safety related to the behavior of other persons
!
Safety related to the rail/bus operation
COMFORT
!
Availability of seating at the station/bus stop
!
Availability of seats on the train/bus
!
Smoothness of the train/bus ride
!
Comfort of the seats
!
Degree of crowding on the train/bus
!
Comfortable temperatures on the train/bus
!
Availability of handrails/grab bars
CONVENIENCE
!
Availability of stations/bus stops close to home
!
Availability of stations/bus stops close to work/shopping
!
Availability of parking at stations/bus stops
PERFORMANCE/RELIABILITY
!
Frequency of service
!
Travel time by train/bus
!
On-time performance
!
Wait time when transferring
Measuring Customer Satisfaction and Service Quality A-25
Appendix B
EASE OF USING THE SERVICE
!
Knowing when trains/buses arrive and depart
!
Availability of information at a station
!
Availability of printed schedules
!
Ease of getting information by telephone
!
Courtesy/helpfulness of ticket agents
!
Ease of purchasing tickets/passes/tokens
!
Visibility of station names from on the train
!
Visibility of train/bus names/route numbers/colors from the outside
!
Ease of getting on/off train/bus
!
Ease of paying fare
!
Ease of making connections/transfers
!
Knowledgeable and courteous conductors/drivers on-board
!
Availability of information about delays from conductors/drivers
!
Clear/timely stop announcements
CONDITION OF VEHICLES AND FACILITIES
!
Cleanliness of the train/bus interior
!
Trains/buses clean of graffiti
!
Stations/bus shelters clean of graffiti
!
Cleanliness of train stations/bus stops
VALUE
!
Cost of a one-way ride
!
Cost of a transfer
!
Availability of discounted fares, e.g., senior citizens, students
!
Availability of volume discounts, e.g., monthly passes
!
Cost of parking at stations/bus stops
