Professional Documents
Culture Documents
Published by:
http://www.sagepublications.com
On behalf of:
Additional services and information for Journal of Hospitality & Tourism Research can be found at:
Subscriptions: http://jht.sagepub.com/subscriptions
Reprints: http://www.sagepub.com/journalsReprints.nav
Permissions: http://www.sagepub.com/journalsPermissions.nav
Citations: http://jht.sagepub.com/content/16/1/51.refs.html
Jonathan D. Barsky
University of Sari Francisco
ABSTRACT
Focusing on the hotel industry, this study builds a functional definition of
customer satisfaction and a practical approach to facilitate its measurement. Field
research conducted in the United States and Japan support these new approaches.
Contributions from the literature provide a theoretical basis for the development of
a customer satisfaction model. The variables of this model were converted into a
survey instrument (customer comment card) and tested at a large urban hotel. Key
Words: customer satisfaction, expectations, discomfirmafion, importance, guest
comment card.
INTRODUCTION
The guest surveys found in most hotels are used in a variety of ways. As a
consequence of these comment cards, managers may be given bonuses, employ-
ees may be rewarded or retrained, resources may be reallocated, mailing lists may
be developed, and guests are given an outlet for their opinions.
The guest surveys used in many hotels present imcomplete or inaccurate
information that cannot help hotel managers with decision making (see for example,
Lewis & Morris, 1987; Lewis & Pizam, 1981).The field research conducted for this
study has revealed the two most critical problems with using hotel customer
comment cards for decision making: (1)poorconstructvalidity(1hesurvey questions
are not based on any established concepts or theories, and do not measure what
they are intended to); and (2) poor statisticalvalidity (caused by the nonrandom self-
selection process used to identify survey participants).This article will address the
issue of construct validity.
No previous study has focused exclusivelyon the constructvalidity issues of the
guest comment card. Industry studies, however, have addressed statistical validity
issues, such as improving response rates by analyzing distribution methods and
changing the length of the survey form (see for example, Trice & Layman, 1984).
The purpose of the guest comment card is primarily to assess guest satisfaction
(Lewis & Piram, 1981)and to find out if the guest will return. (These two concepts
have never been shown to be empirically related.) The typical questions and
answers provided in hotel guest comment cards do not meet these primary
objectives. Typical objective ratings (e.g., Quality of food: Excellent..l 2 3 4..Poor)
only represent part of a customer's satisfaction or dissatisfactionand do not provide
sufficient data for informed decision-making by the hotel manager at the property
level or for strategic determination at the corporate level.
Developments in the study of customer satisfaction offer insight regarding the
construct validity errors commonly affecting hotel guest surveys. This article will
develop a practical definition of customer satisfaction and a simple method to
facilitate the measurement of customer satisfaction.A major thrust of this study is to
compare these new approaches with traditional industry practices of evaluating
customer satisfaction.
LITERATURE REVIEW
52 V O L U M E 16 NUMBER 1 1992
Disconfirmation Theory
Direct measures of satisfaction (for example, 'How satisfied are you with
product X?") wrongly assume that respondents know what gives them satisfaction
and that these can be linkeddirectlyto purchase behavior(Alpert,1971).Also, global
satisfaction measures, by themselves, do not offer any explanation for a single
observation.
The disconfirmation paradigm is widely accepted as a view of the process by
which consumers develop feelings of satisfaction or dissatisfaction(Cadotte, et al.,
1987). Customerscompare a new service experiencewith some standard that they
have developed. How well it stacks up against this measuring stick will determine
their belief about the service.
The first part of this process presumes that the consumer makes a purchase
decision based on expectations, attitudes, and intentions (Oliver, 1980). Later,
during or after consumption, a perception of the brand's performanceoccurs as the
consumer evaluates his or her experience (service performance).The process is
completed when a consumer compares the service performance with a
preexperiencestandard (Bearden & Teel, 1983;Cardozo, 1965; Day, 1977; Oliver,
1980) or expectation. Confirmation, satisfaction, or dissatisfaction will occur as a
result of this comparison, either matching, exceeding,or not meeting the standard.
The disconfirmation paradigm includes four components: expectations, per-
ceived performance, disconfirmation, and satisfaction. Expectations reflect a
preconsumption perception associated with the product or service. Performance is
the consumer's perception of aservice experience.Disconfirmation will result if there
is a discrepancy (positive. negative) between performance and expectations. If.
however, performance is perceived as meeting expectations, confirmation will
result. Therefore, according to the disconfirmationparadigm, overall satisfaction is
determined by combining the satisfaction outcomes for the various attributesof the
service.
The first three components are generallyaccepted as affecting satisfaction, but
whether these lead directly to satisfaction or dissatisfaction has not been estab-
lished. There is also no consensus regarding how these variables interrelate. For
example, a high quality service may result in aconsumets dissatisfactionif his or her
expectationswere too high (for example, overstated advertising claims). Investiga-
tions focusing on the relative importance of expectations and product performance
as determinants of satisfaction offer conflicting results (Churchill & Surprenant,
1982).
Another shortcoming of the disconfirmation paradigm is that it assumes that
services can be categorized into schematic criteria. Although certain elements of
product classification frameworksdo apply to services (Lovelock, 1991), the contri-
bution of the diconfirmation paradigm to this particular study will be its (modified)
three-point framework:
1. Expectations or other prepurchase standards (if any) brought to the con-
sumption experience;
2. Factors affecting perceived consumption; and
3.Confirmation (or disconfirmation)resulting (duringlafterconsumption)from
(1) and (2).
Accommodations are also made for the individualized and experiential nature
of the hotel productlservice.
Equitable Performance
Borrowing from equity theory (Adams, 1963)the equitable performance stan-
dard is formed by the consumer based on the individual's costs or investments and
anticipated rewards. This standard represents the performance level the customer
should receive, or deselve, given a perceived set of costs (Liechy & Churchill, 1979;
Miller, 1979;Woodruff, et al., 1983).
Experience-based Norms
Experience-basednorms account for a desired level of performanceas well as
possible performance levels based on experience with competitors (Cadotte,et al.,
1987).Related constructs,such as "productbased norms" and "best brand norms,"
also attempt to explain the standards of comparison used by consumers. This
process, however, has been primarily associated with tangible products and may
therefore differ when applied to a productlservice mix.
Relationship Quality
Research regardingthe relationshipbetweenthe salespersonand the customer
(Crosby,et al., 1990)suggest that relationship quality leads to satisfaction and will
significantly affect sales. This may be extended to include not only any employee
representing an organization but also to the impact that media and word of mouth
have on the relationship they create with customers.
Price
54 V O L U M E 16 NUhlBER 1 1992
Price is often an indicator of quality and can affect people’s perceptions and
behavior. Research has shown that consumers often perceive higher priced items
as having higher quality. Price will play a larger role in determining quality when a
customer is less familiar with a product or service. Retailers have been known to
increase sales frequency and revenue by increasing the price of items (Cialdini,
1990). New or complex services may also benefitfromthis product-based phenom-
ena.
Development of Expectations
It is apparent, therefore, that expectations may be derived from a variety of
sources. It is also likely that consumers use differentsources and process informa-
tion in a variety of ways.
Although the disconfirmation paradigm is generally accepted as the construct
that, thus far, best explains customer satisfaction. the studies done on the
disconfirmation theory raise more questions than they answer. For example, how
does the disconfirmation process actually occur? Do people use other sources or
types of information in developing expectations? What are the most common
sources? Which are the most important? Do sources vary by product or service, by
situation, by consumer? How are these sources selected, and how can business
operations affect the selection process? Can the frustration from a delayed flight
carry over and affect guest expectationsof their hotel?
Other questions suggested but not addressed by the disconfirmationtheory
relate to perceived performance. Because it is not possible to distinguish confirming
and discontiming perception from perceptions, prior experience or lack of expen-
ence may vary among customers.The expectationframeworksdescribed above are
more appropriate for servicesthat are close substitutes. Expectationsabout new or
different services are likely to be revised concurrent to consumption.
What other factors specific to the consumption situation affect a customer’s
perception of product or service performance? One researcher has suggested that
marketing efforts may affect customer perception of service performance (Bitner,
1990). 0ther“personal”variables may also substantiallyaffect a customer‘sproduct
or service experience.
Extensive research could address these complicated issues. However,for the
purpose of this study and its practical application, the task of interpreting the
disconfirmation paradigm will be left up to the consumer. This will provide only one
component of the customer satisfaction model and does not, by itself, imply
satisfaction. Customers will be asked how well their expectations were met. Their
responsewill represent the guests’confirmingordisconfirmingthe product orservice
performance with any preexperience standards they may have. The response will,
if necessary, require guests to combine preexperience standards (expectations)
with their perception of product or service performance (quality, etc.). If for any
reason guests have not formulated any preexperience standards or expectations,
their responses will simply reflect their perception of the product or service experi-
ence.
Figure 1
The “EM” Component of the Customer Satisfaction Model
Factors affecting perceived performance
(situation,marketing mix, attributions,purchase level of productlservice.
and other personalhndividualdifferences)
Perceived performance
Disconfirmation (EM)
Expectations
56 V O L U h l E 16 NUhIBER 1 6 1992
Expectancy-Value Theory
According to expectancy theory (Tolman, 1932).customers often make some
estimationabout aproduct,its benefits, and the likely outcomesof using the product.
According to Tolman, people will learn to perform (or increase their probability of
performing) behaviorthat they expect to lead to positive outcomes. Social psycholo-
gists have stressed looking at two dimensions of attitudestoward objects (Fishbein,
1967; Riteer, 1966),and their work has influenced studies in marketing research
(Bass & Talarzyk, 1969;Hansen, 1969).Common to all these is that overall attitude
(satisfactionldissatisfaction)is a function of beliefsabout an object's attributes (that
is, the belief that a product or service possesses a particularattribute;see Figure l ) ,
and the strength of these beliefs (that is, the relative importance of each attribute to
the customer's overall satisfactionwith the product or service).
The impact of attribute importance on consumer decision making is widely
recognized (Heeler, et al., 1979; MacKenzie, 1986; Quelch, 1978).The notion that
a consumer will weigh various attributes associated with a product or service is
based on expectancy theory.
Expectancytheory assumes that people act on the basis ofwhat they value and
what they anticipatewill result from their actions. Rational behavior is also presumed
as a result of acquiring and assimilatingmost product information prior to a decision.
Because of the relatively unknown and experientialnature of services, this may be
more viable for tangible products, such a s toothpaste, than for the combination of
products and services associated with a hotel room purchase.
The first systematicversion of expectancy theory (Vroom, 1964) included the
followingvariables: action, outcome, expectancy instrumentality,and valence. The
model predicts a person's motivation to perform an action by assigning subjective
probabilitiesthat various outcomes will occur.
Probably the most well known and widely accepted expectancy model is
Fishbein's expectancy-valueformulation (Fishbein and Ajzen, 1975):
A = (b)(e) (11
where A is the attitude toward an object, action, or event; b is the belief about the
object's attributes or about the act's consequences; and e is the evaluation of the
attributesorconsequences.Thus,accordingto this model (Equation I ) , a customers'
attitudestoward a product or service can be estimated by multiplying their evaluation
of each attributeassociated with the product or serviceby their subjectiveprobability
that the product or service has that attribute, and then summing the products for the
total set of beliefs. Stated differently, a person's attitude is related to the strength of
his or her beliefs that link the object to various attributes, multiplied by his or her
evaluation of the attributes. The notion of importance is captured within the belief
variable. Attitudes are therefore predictable from the s u m of the resulting products
(Fishbein & Ajzen, 1975).Following Fishbein and Ajzen, marketing studies focusing
on the variety of attributesassociated with consumer purchasing are all based on the
expectancy attitude model.
Hogarth (1980) presented a similardecision making model and suggested that
people implicitlyvalue each alternative being considered by summing the contribu-
tions of each dimension, weighted by their importance:
Value of alternative = Sum of (Relativeweight x Value) (2)
of all dimensions
This model assumes that people will select the alternative with the highest
value. Hogarth's model has demonstrated strong predictive abilities in a variety of
business situations,including production-scheduling,college admission decisions,
and judgments by auditors (Dawes & Corrigan, 1974; Meehl, 1955).
Originally, weighted expectancy formulaswere meant to predict effort and may
be more applicable (in general) to decisions regarding choices as opposed to the
formation of postconsumption attitudes. The contribution of the expectancy theory
for the development of a customer satisfaction model will, therefore, be limited
primarilyto its focuson attribute utility (value).Theassessmentof attributeutility may
occur by a guest at any time (pre-through postconsumption)and may be revised
concurrentto consumption.Forthe current model, a postconsumption measurement
will be obtained and will identify a net-value utility (representingthe combined effect
of any variations occurring throughout the consumption experience).
The equation that most closely resembles the model presented in this study
was recently introduced by Carmen (1990).
Q = li (Pi - Ei) (3)
Here, Q is overall quality; I is the importanceof serviceattribute i; the sum is over
the number of serviceattributes;Pis perception;and E is expectation.This definition
ofquality includesall the significantcomponentssupported in the customersatisfac-
tionliteratureandresemblestheFishbeinandHogarthattitudemodels.Althoughthis
formula is an expression of quality, Carmen asserts that this labeling (of the
dependent variable) was imposed on the equation and suggests that other names
may be more appropriate. He further maintains that the difficulty of measuring
58 V O L U M E 16 H N U M B E R I W 1992
expectations (E) and the errors associated with subtracting these values from
perceptions (P) are a major problem.
The current study will label Carmen's "Quality"term (Q) "CustomerSatisfaction"
(CS). The EM construct (Figure 1) addresses Carmen's measurement problem by
combining expectations (E) with perceptions (P) in a single customer observation,
"Expectations Mer (EM). Thus, Carmen's model of quality serves as a stepping
stone to the customer satisfactionmodel by using the s a m e independent variables
and with the same arithmeticrelationshipsas previous research and common sense
suggest. The contribution of expectancy theory will, therefore, be limited to its
emphasis on attribute utility (importance) and its mathematical association of
importance with perception and expectation.
applications. This study will demonstrate the theoretical and practical value of
including importance as a separate measure in customer surveys.
Traditional methods of customer feedbackdonot consider attributeimportance.
Researchers have assumed that the weights attached to product or service at-
tributes are based on deep-seated cultural norms and values. Thus, the issue of
importance has been treated as a stable, predetermined, exogenous variable in
most consumer research. Most survey instruments, therefore, consider the impor-
tance levels among all attributes to be equal. Only open-endedquestions, or space
for "comments,"are able to capture any variation in importance among responses.
These, however, are subjective responses and introduce additional questions of
measurement error.
Borrowing from the expectancy-valueequation (Equation 3),a separate term,
'I" representing "importance" will serve as the intervening variable between
disconfirmation and satisfactionin the customer satisfactionmodel to be used in this
study.
This term-related to Fishbein's "evaluation"and Hogarth's "value"terms and
the same as Carmen's expressionof importance-will express the importancelevel
of individual product or serviceattributes. This variable will also be used to compare
importance levels among all salient attributes. The following are the operational
assumptions of the 'I" variable:
* Under conditions of very high or very low evaluation by consumers, higher
importance ratings will result; and under less extreme conditions, lower
importance ratings will be obtained.
*Thevariation in importance ratings will reflect the relative value of the various
service attributes to consumers. Lower importance ratings are likely to play
less of a role in affecting overall satisfaction; whereas higher importance
ratings are likely to play a more critical role in determining customer
satisfaction.
Figure 2
The Customer Satisfaction Model
Factors affecting perceived performance
(situation, marketing mix, attributions,purchase level of productlservice
and other personallindividualdifferences)
Perceived Performance
Customer
Disconfirmation x Importance, e = Satisfaction
(EM) (1) (CS)
Expectations
Other pre-experience standards
(ideal levels, equity, prior experience with same or similar products/
services,and other personal individual differences)
e = error term
Survey
The survey form developed for this study reflects previous academic research
and the practices and preferences of the US. and Japanese hotel industries. It is
based on 20 interviewswith general managers ofleading US. and Japanese hotels
Table 1
Customer Satisfaction Scores (CS)
Standard
62 V O L U M E 16 D N U h f B E R 1 W 1992
Table 2
lnterscale Reliability: Matrix of Spearman Correlation Coefficients
EM :EM
E X P E C T A T I O N S M E T
LOC PARK RECEP ROOM SERVS F&B E.ATT FACIL PRICE
E LOC 1.000
X
P PARK .095 1.000
E
C RECEP .270 .250 1.000
T
A ROOM .283 .067 .118 1.000
T
I SERVS .356 .286 .536 .386 1.000
0
N F&B .321 .115 .433 .217 .390 1.000
S
E.AT .263 -133 .455 .126 .448 .381 1 .OOO
M
E FACIL .396 .217 .245 .448 .416 .362 .279 1 .OOO
T
PRICE .386 .239 226 .328 .247 .289 .290 .324 1 .OD0
Note: N = 100
64 V O L U M E 16 NUMBER 1 1992
Table 3
lnterscale Reliability: Matrix of Spearman Correlation Coefficients
I:I
I M P O R T A N C E
LOC PARK RECEP ROOM SERVS F&B E.AlT FAClL PRICE
I
LOC 1.000
M
PARK -.014 1.000
P
RECEP .195 -.011 1.000
0
ROOM -157 .024 317 1.000
R
SERVS .200 .123 A83 .360 1.000
T
F&B .115 .060 .282 .074 .376 1.000
A
E.ATT .121 .098 325 .308 .465 .214 1 .OOO
N
FACIL .136 .062 .340 319 538 .271 .523 1 .OOO
C
PRICE .238 ,258 .140 .197 .352 .119 .220 .348 1 .OOO
E
Note: N = 100
Table 4
lnterscale Reliability: Matrix of Spearman Correlation Coefficients
EM :I
E X P E C T A T I O N S M E T
LOC PARK RECEP ROOM SERVS F&B E.AlT FAClL PRICE
I
LOC -463 -.lo0 .070 -108 .275 .129 .lo7 .193 .070
M
PARK -.lo7 .224 -.033 .024 .113 .On .033 -.126 .037
P
RECEP .120 .050 .210 .241 355 .191 .130 321 .017
0
ROOM -118 -.133 -.049 .296 .159 -079 .092 -207 -.005
R
SERVS .148 -.008 .024 .183 .294 .232 .298 .262 .130
T
F&B -162 -080 .221 -236 .289 -291 .263 .208 .114
A
E.ATT .005 -.194 .081 .076 .198 .083 .440 .188 -.070
N
FACIL .145 -.075 .059 .265 .235 .123 .204 .385 -.002
C
PRICE .lo8 .OOO .092 .021 .I 70 .061 .083 .096 .238
E
Note: N = 100
of correlation occurring between these measures. For example, did the EM scores
for "Parking" and "Location" measure the same thing? Were these numbers so
similar on each survey that, perhaps, these two attributes were taken to mean the
s a m e thing? Their correlationcoefficientidentifiesthe strength of this relationshipfor
each of the 100 surveys in the "Neat" sample. This particular comparison demon-
strated the lowest correlation (.095;see Table 2).
The highest correlation among the EM scores was between 'Services" and
'Reception" (.536),indicating similar responses for these attributes and quite likely
some overlap in their perception. "Services"was also part of the highest correlation
comparison among the I scores. "Facilities" and 'Services" showed the most
correlation (538)and were also closely ranked in numerical mean scores. While not
statistically significant, this correlation suggests that these two areas performed
similarly and, most likely, that their meanings were also confused. The words
describing these functions should be changed to distinguish and clarify their
meanings.
Table 4 shows the correlationsbetween the EM and I scores and is particularly
importantbecause it reveals how differentthese measures are; thus, it displays low
correlations.
Showing the most similarity were the "Location" scores for EM and 1. This
correlation is smaller than the largest coefficients in Tables 2 and 3. and further
represents the consistently high marks given to this hotel for its location.
The EM scores for "Parking" and the I scores for "Employee Attitude" showed
the least correlation (-.194; Table 4) between these two measures. This demon-
strates the contrast between the considerable importance placed on employee
attitudes and the low evaluations of the available parking.
The low numbers found throughoutthese three tables (especiallynumbers less
than .2) indicate low correlationsbetween the attribute measurements and therefore
support the reliability of the survey data as obtained in this study. Each of the 18
attributes in this survey instrument display independence and t h u s support the
operationalizationof the customer satisfaction model and the conclusions (below).
The other reliability test performed evaluates the overall consistency of the
observations in terms of the construct under consideration. In other words, how
reliablewasthissurveyinmeasuring customersatisfaction?Table5summarizesthe
results of this internal consistency test.
The test for interitem reliability was conducted by separating the odd and even
observationsfrom all the data. These data include the 18 attribute evaluationsand
the answers to the other survey questions (see Figure 2). The high correlation
coefficients (between .8 and 1.0) indicate a strong relationship between the survey
data obtained and the overall measurement of the customer satisfaction construct.
Each of the 18attributes in this survey instrumentdisplay strong independence
and thus support the operationalizationof the customer satisfactionmodel. A strong
relationshipwas also found to exist between thesurvey data obtained and the overall
measurement of the customer satisfaction construct.
66 VOLUhiE 16 I N U M B E R 1 m 1992
Table 5
lnteritem Reliability: Split Half Correlation
68 V O L U M E 16 H N U M B E R 1 1992
REFERENCES
Adams, S. J. (1963, October). Towards an understandingof inequity. Journal of
Abnormal and Social Psychology, 67,422-36.
Band, W.A. (1978, October).The art of listening to customers. Sales and Marketing
Management in Canada, p. 6.
Barsky, J.D. (1990). THEORY S: Total customer service. The Cornell Hotel and
Restaurant Administration Quarterly, May, 89.
Barsky, J.D. (1992a). The quality sample. The Comell Hotel and Restaurant
Administration Quarterly, February.
Barsky, J.D. (1992b). Customer satisfaction applications. The Comell Hotel and
Restaurant A dministration Quarterw, Apri I.
Bass, F. M., & Talarzyk, W. W. (1969). A study of attitude theory and brand
preference. Journal of the American MarketingAssociation, Fall, 272-9.
Cohen, J. 8.. Fishbein, M., & Ahtola, 0. T. (1972). The nature and uses of
expectancy-value models in consumer attitude research. Journal of
Marketing Research, 9, 456-60.
Crosby, L. A., Evans, K. R., & Cowles, D. (1990). Relationship quality in services
selling: An interpersonal influence perspective. Journal of Marketing, 54,
68-81.
Dawes, R., & Corrigan, 6. (1974). Linear models in decision making. Psychologi-
cal Bulletin, 81.95-106.
70 V O L U M E 16 H N U M B E R 1 H 1 9 9 2
Heeler, R. M., Okechuku, C., & Reid, S. (1979). Attribute importance: contrasting
measurements. Journal of Marketing Research, 76,60-63.
Hogarth, R.M. (1980). Judgement and choice: The psychology of decision. New
York: John Wiley & Sons.
Katz, D. (1960). The functional approach to the study of attitudes. Pubh Opinion
Quarterly, 24,163-204.
Lewis, R. C., & Morris, S. V. (1987, February). The positive side of guest
complaints. The Cornell Hotel and Restaurant Administration Quarterly,
pp. 13-15.
Liechy, M., & Churchill, Jr., G. A. (1979). Conceptual insights into consumer
satisfaction with services. in Niel Beckwith (Ed.), Educators Conference
Proceedings (pp. 509-515). (Chicago: American Marketing Association.
Ryan, M. J.. & Holbrook, M. B. (1982). Importance, elicitation order, and expect-
ancy x value. Journal of Business Research, 70.310.
Tolman, E.C. (1932). Purposive behaviorin animals and men. New York:
Appleton-Century.
Trice, A. D., & Layman, H. (1984, November). Improving guest surveys. The
Comell University Hotel and Restaurant Administration Quarterly, pp.
10-13.
Vroom, V.H. (1964). Workand motivation. New York: John Wiley and Sons.