SERVQUAL: A Multiple-Item Scale for Measuring Consumer ... · SERVQUAL) for assessing customer perceptions of service quality in service and retailing organizations. After a discussion

Volume 64 Number 1 Spring 1988

SERVQUAL: AMultiple-Item Scale forMeasuring ConsumerPerceptions ofService QualityA. PARASURAMAN VALARIE A. ZEITHAMLFoley's/Federated Professor of Retailing Associate Professor

and Marketing Studies of MarketingTexas A & M University Duke UniversityCollege Station, Texas Raleigh-Durham, N. Carolina

LEONARD L. BERRYFoley's/Federated Professor of Retailing

and Marketing StudiesTexas A & M University

College Station, Texas

This paper describes the development of a 22-item instrument (calledSERVQUAL) for assessing customer perceptions of service quality inservice and retailing organizations. After a discussion of the conceptu-alization and operationalization of the service quality construct, theprocedures used in constructing and refining a multiple-item scale tomeasure the construct are described. Evidence of the scale's reli-ability, factor structure, and validity on the basis of analyzing datafrom four independent samples is presented next. The paper concludeswith a discussion of potential applications of the scale.

Intensifying competition and rapid deregulation have led many serviceand retail businesses to seek profitable ways to differentiate themselves.One strategy that has been related to success in these businesses is thedelivery of high service quality (Rudie and Wansley 1985; Thompson,

The research reported in this article was made possible by a grant from the MarketingScience Institute, Cambridge, MA.

12 Journal of Retailing

SERVQUAL

DeSouza, and Gale 1985). Delivering superior service quality appears tobe a prerequisite for success, if not survival, of such businesses in the1980s and beyond.

Unlike goods quality, which can be measured objectively by such indi-cators as durability and number of defects (Crosby 1979; Garvin 1983),service quality is an abstract and elusive construct because of three fea-tures unique to services: intangibility, heterogeneity, and inseparability ofproduction and consumption (Parasuraman, Zeithaml, and Berry 1985). Inthe absence of objective measures, an appropriate approach for assessingthe quality of a firm's service is to measure consumers' perceptions ofquality. As yet, however, no quantitative yardstick is available forgauging these perceptions.

The purpose of this article is twofold: (1) to describe the development ofa multiple-item scale for measuring service quality (called SERVQUAL)and (2) to discuss the scale's properties and potential applications. Thebasic steps employed in constructing the scale closely parallel proceduresrecommended in Churchill's (1979) paradigm for developing better mea-sures of marketing constructs. Figure 1 provides an overview of the steps.

This article is divided into five sections. The first section delimits thedomain of the service-quality construct and describes the generation ofscale items (Steps 1,2, and 3 in Figure t). The second section presents thedata-collection and scale-purification procedures (Steps 4 through 9),while the third section provides an evaluation of the scale's reliability andfactor structure (Step 10). The next section deals with assessment of thescale's validity (Step 11). The final section discusses potential applica-tions of the scale.

DOMAIN OF THE SERVICE-QUALITY CONSTRUCT

In deploring the inadequacy of measurement procedures used in themarketing discipline Jacoby (1978) wrote:

Many of our measures are developed at the whim of a re-searcher with nary a thought given to whether or not it ismeaningfully related to an explicit conceptual statement of thephenomena or variable in question. In most instances, ourconcepts have no identity apart from the instrument or proce-dures used to measure them. (p. 92)

The need for scale development to be preceded by, and rooted in, a soundconceptual specification of the construct being scaled has been empha-sized by other scholars as well (e.g., Churchill 1979; Peter 1981). The

13

Journal of Retailing

HGURE 1

Summary of Steps Employed in Developing theService-Quality Scale

Step 1: DelJnition ol service quality as the discrepancy beiweer*consumers' pefceplions ol services oflered by a particular lirm anditieir expectations about lirms oflering such services.

Stap 2: Idenlificaiion ol lO dimensions making up Ihe domain of Iheservice-quality construct.

Step 3: Generation of 97 items representing the 10 dirronsions, I

IESfp 4: Collection ol expectations and perceptions data from asample of 200 respondents, each ol whom was a current or receniuser ol one ol the following services: banking, credit card, appli-ance repair or mainienance, long-distance telephone, and secur-ities brokerage.

Step 5; Scale purification ihrough the lollowing ileraiive sequence:

Computation ol coefficient alptia andilem-to-total correlations lor each di-mension.

Deletion of items whose Item-to-totalcorrelations were low and whose re-nx)val increased coellicient alpha.

Factor analysis to verily the dimen-sionality ol the overall scale.

Reassignment of items and restruc-turing of dimensions where r>eces-sary. ^^_^_

Step 6: Identification of 34 items representing 7 dimensions.

A.Step 7: Collection of expectations and perceptions data (using the34-item instrument) from four independent samples ol 200 respon-dents (each sample contained current or recent customers of a na-tionally known firm in one ol the following lour service sectors:banking, credit card, appliance repair and maintenance, and long-distance telephone).

J_Step 8: Evaluation anu I'unr^r purilication nt the 34-item scale byusing the same iterative sequence as in Step 5 on each of the fourdata sets,

Step 9: Identification of a more("SERVQtJAL") representing five dimensions.

lonious, 22-ilem scale

Step 10: Evaluation of SERVOUAL's reliability and factor structureand reanalysis of the original data (collected in Step 4) pertaining tothe 22 items, lo verily the scale's internal consistency and dimen-sionality.

Step 11: Assessment ol SERVOUAL's validity.

14

SERVQUAL

coneptual foundation for the SERVQUAL scale was derived from theworks of a handful of researchers who have examined the meaning ofservice quality (Sasser, Olsen, and Wyckoff 1978; Gronroos 1982; Leh-tinen and Lehtinen 1982) and from a comprehensive qualitative researchstudy that defined service quality and illuminated the dimensions alongwhich consumers perceive and evaluate service quality (Parasuraman,Zeithaml, and Berry 1985).

Conceptualization of Service Quality

The construct of quality as conceptualized in the services literature andas measured by SERVQUAL, the scale that is the focus of this article,involves perceived quality. Perceived quality is the consumer's judgmentabout an entity's overall excellence or superiority (Zeithaml 1987). Itdiffers from objective quality (as defined by, for example, Garvin 1983and Hjorth-Anderson 1984); it is a form of attitude, related but not equiva-lent to satisfaction, and results from a comparison of expectations withperceptions of performance.

Perceived quality versus objective quality. Researchers (Garvin 1983;Dodds and Monroe 1984; Holbrook and Corfman 1985; Jacoby and Olson1985; Zeithaml 1987) have emphasized the difference between objectiveand perceived quality. Holbrook and Corfman (1985), for example, notethat consumers do not use the term quality in the same way as researchersand marketers, who define it conceptually. The conceptual meaning dis-tinguishes between mechanistic and humanistic quality: "mechanistic(quality) involves an objective aspect or feature of a thing or event; hu-manistic (quality) involves the subjective response of people to objects andis therefore a highly relativistic phenomenon that differs between judges"(Holbrook and Corfman 1985, p. 33). Garvin (1983) discusses five ap-proaches to defining quality, including two (product-based and manufac-turing-based) that refer to objective quality and one (user-based) that par-allels perceived quality.

Quality as attitude. Olshavsky (1985) views quality as a form ofoverall evaluation of a product, similar in many ways to attitude. Hol-brook concurs, suggesting that quality acts as a relatively global valuejudgment. Exploratory research conducted by Parasuraman, Zeithaml, andBerry (1985) supports the notion that service quality is an overall evalua-tion similar to attitude. The researchers conducted a total of twelve focusgroup interviews with current or recent consumers of four different ser-vices—retail banking, credit card, securities brokerage, and product re-pair and maintenance. The discussions centered on issues such as the

15


meaning of quality in the context of the service in question, the character-istics the service and its provider should possess in order to project a high-quality image, and the criteria customers use in evaluating service quality.Comparison of the findings from the focus groups revealed that, regardlessof the type of service, customers used basically the same general criteria inarriving at an evaluative judgement about service quality.

Quality versus satisfaction. Oliver (1981) summarizes current thinkingon satisfaction in the following definition: "[satisfaction is a] summarypsychological state resulting when the emotion surrounding disconfirmedexpectations is coupled with the consumer's prior feelings about the con-sumption experience" (p. 27). This and other definitions (e.g., Howardand Sheth 1969; Hunt 1979) and most all measures of satisfaction relate toa specific transaction. Oliver (1981) summarizes the transaction-specificnature of satisfaction, and differentiates it from attitude, as follows:

Attitude is the consumer's relatively enduring affective orien-tation for a product, store, or process (e.g., customer service)while satisfaction is the emotional reaction following a dis-confirmation experience which acts on the base attitude leveland is consumption-specific. Attitude is therefore measured interms more general to product or store and is less situationallyoriented, (p. 42)

Consistent with the distinction between attitude and satisfaction, is adistinction between service quality and satisfaction: perceived servicequality is a global judgment, or attitude, relating to the superiority of theservice, whereas satisfaction is related to a specific transaction. Indeed, inthe twelve focus group interviews included in the exploratory researchconducted by Parasuraman, Zeithaml, and Berry (1985), respondents gaveseveral illustrations of instances when they were satisfied with a specificservice but did not feel the service firm was of high quality. In this way,the two constructs are related, in that incidents of satisfaction over timeresult in perceptions of service quality. In Oliver's (1981) words, "satis-faction soon decays into one's overall attitude toward purchasingproducts."

Expectations compared to perceptions. The writings of Sasser,Olsen, and Wyckoff (1978); Gronroos (1982); and Lehtinen and Lehtinen(1982), and the extensive focus group interviews conducted by Parasur-aman, Zeithaml, and Berry (1985), unambiguously support the notion thatservice quality, as perceived by consumers, stems from a comparison ofwhat they feel service firms should offer (i.e., from their expectations)with their perceptions of the performance of firms providing the services.

16

SERVQUAL

Perceived service quality is therefore viewed as the degree and direction ofdiscrepancy between consumers' perceptions and expectations.

The term "expectations" as used in the service quality literature differsfrom the way it is used in the consumer satisfaction literature. Specifi-cally, in the satisfaction literature, expectations are viewed as predictionsmade by consumers about what is likely to happen during an impendingtransaction or exchange. For instance, according to Oliver (1981), "It isgenerally agreed that expectations are consumer-defined probabilities ofthe occurrence of positive and negative events if the consumer engages insome behavior" (p. 33). In contrast, in the service quality literature, ex-pectations are viewed as desires or wants of consumers, i.e., what theyfeel a service provider should offer rather than would offer.

Dimensions of service quality. Exploratory research of Parasuraman,Zeithaml, and Berry (1985) revealed that the criteria used by consumers inassessing service quality fit 10 potentially overlapping dimensions. Thesedimensions were tangibles, reliability, responsiveness, communication,credibility, security, competence, courtesy, understanding/knowing thecustomer, and access (a description of the dimensions can be found inParasuraman, Zeithaml, and Berry 1985, p. 47). These 10 dimensions andtheir descriptions served as the basic structure of the service-quality do-main from which items were derived for the SERVQUAL scale.

Generation of Scale Items

Items representing various facets of the 10 service-quality dimensionswere generated to form the initial item pool for the SERVQUAL instru-ment. This process resulted in the generation of 97 items (approximately10 items per dimension). Each item was recast into two statements—oneto measure expectations about firms in general within the service categorybeing investigated and the other to measure perceptions about the partic-ular firm whose service quality was being assessed. Roughly half of thestatement pairs were worded positively and the rest were worded nega-tively, in accordance with recommended procedures for scale develop-ment (Churchill 1979). A seven-point scale ranging from "StronglyAgree" (7) to "Strongly Disagree" (1), with no verbal labels for scalepoints 2 through 6, accompanied each statement (scale values were re-versed for negatively worded statements prior to data analysis). The ex-pectation statements were grouped together and formed the first half of theinstrument. The corresponding perception statements formed the secondhalf. An abbreviated version of the instrument, containing a set of expec-tation statements (labeled as E's) and a corresponding set of perception

17


Statements (labeled as P's), along with directions for responding to them,is included in the appendix. Negatively worded statements are identifiedby a minus sign within parentheses in the appendix.

DATA COLLECTION AND SCALE PURIFICATION

The 97-item instrument was subjected to two stages of data collectionand refinement. The first stage focused on: (1) condensing the instrumentby retaining only those items capable of discriminating well across re-spondents having differing quality perceptions about firms in several cate-gories, and (2) examining the dimensionality of the scale and establishingthe reliabilities of its components. The second stage was primarily confir-matory in nature and involved re-evaluating the condensed scale's dimen-sionality and reliability by analyzing fresh data from four independentsamples. Some further refinements to the scale occurred in this stage.

Data Collection, First Stage

Data for initial refinement of the 97-item instrument were gathered froma quota sample of 200 adult respondents (25 years of age or older) re-cruited by a marketing research firm in a shopping mall in a large metro-politan area in the Southwest. The sample size of 200 was chosen becauseother scale developers in the marketing area had used similar sample sizesto purify initial instruments containing about the same number of items asthe 97-item instrument (e.g., Churchill, Ford, and Walker 1974; Saxe andWeitz 1982). The sample was about equally divided between males andfemales. Furthermore, the respondents were spread across five differentservice categories—appliance repair and maintenance, retail banking,long-distance telephone, securities brokerage, and credit cards. These cat-egories were chosen to represent a broad cross-section of services thatvaried along key dimensions used by Lovelock (1980, 1983) to classifyservices. For each service category, a quota of 40 recent users of the ser-vice was established. To qualify for the study, respondents had to haveused the service in question during the past three months.

Screened and qualified respondents self administered a two-part ques-tionnaire consisting of a 97-statement expectations part followed by a 97-statement perceptions part. For the first part, respondents were instructedto indicate the level of service that should be offered by firms within theservice category in question. For the second part, respondents were firstasked to name a firm (within the service category) that they had used andwith which they were most familiar. Respondents were then instructed toexpress their perceptions about the firm.

18

SERVQUAL

Scale Purification, First Stage

The 97-item instrument was refined by analyzing pooled data (i.e., datafrom all five service categories considered together). The pooling of datawas deliberate and appropriate because the basic purpose of this researchstage was to develop a concise instrument that would be reliable andmeaningful in assessing quality in a variety of service sectors. In otherwords, the purpose was to produce a scale that would have general appli-cability.

Purification of the instrument began with the computation of coefficientalpha (Cronbach 1951), in accordance with Churchill's (1979) recommen-dation. Because of the multidimensionality of the service-quality con-struct, coefficient alpha was computed separately for the 10 dimensions toascertain the extent to which items making up each dimension shared acommon core.

The raw data used in computing coefficient alpha (and in subsequentanalyses) were in the form of difference scores. Specifically, for each itema difference score Q (representing perceived quality along that item) wasdefined as Q = P - E, where P and E are the ratings on the corre-sponding perception and expectation statements, respectively. The idea ofusing difference scores in purifying a multiple-item scale is not new. Thisapproach has been used in developing scales for measuring constructs suchas role conflict (Ford, Walker, and Churchill 1975).

The values of coefficient alpha ranged from .55 to .78 across the 10dimensions and suggested that deletion of certain items from each dimen-sion would improve the alpha values. The criterion used in decidingwhether to delete an item was the item's corrected item-to-total correlation(i.e., correlation between the score on the item and the sum of scores onall other items making up the dimension to which the item was assigned).The corrected item-to-total correlations were plotted in descending orderfor each dimension. Items with very low correlations and/or those whosecorrelations produced a sharp drop in the plotted pattern were discarded.Recomputation of alpha values for the reduced sets of statements and ex-amination of the new corrected item-to-total correlations led to furtherdeletion of items whose elimination improved the corresponding alphavalues. The iterative sequence of computing alphas and item-to-total cor-relations, followed by deletion of items, was repeated several times andresulted in a set of 54 items, with alpha values ranging from .72 to .83across the 10 dimensions.

Examining the dimensionality of the 54-item scale was the next task inthis stage of scale purification and was accomplished by factor analyzing

19


the difference scores on the 54 items. The principal axis factoring proce-dure (Harman 1967) was used and the analysis was constrained a priori to10 factors. When the 10-factor solution was rotated orthogonally, no clearfactor pattern emerged. Many of the items had high loadings on severalfactors, thereby implying that the factors may not be independent of oneanother. Moreover, some degree of overlap among the 10 conceptual di-mensions was anticipated by the researchers who initially identified andlabeled the dimensions (Parasuraman, Zeithaml, and Berry 1985). There-fore the 10-factor solution was subjected to oblique rotation (using theOBLIMIN procedure in SPSS-X) to allow for intercorrelations among thedimensions and to facilitate easy interpretation.

The oblique rotation produced a factor-loading matrix that was by andlarge easy to interpret. However, several items still had high loadings onmore than one factor. When such items were removed from the factor-loading matrix, several factors themselves became meaningless becausethey had near-zero correlations with the remaining items, thereby sug-gesting a reduction in the presumed dimensionality of the service-qualitydomain. Furthermore, the highest loadings of a few of the remaining itemswere on factors to which they were not originally assigned. In otherwords, the factor loadings suggested reassignment of some items.

The deletion of certain items (and the resultant reduction in the totalnumber of factors or clusters of items) and the reassignment of certainothers necessitated the recomputation of alphas and item-to-total correla-tions and the reexamination of the factor structure of the reduced itempool. This iterative sequence of analyses (Step 5 in Figure 1) was repeateda few times and resulted in a final pool of 34 items representing sevendistinct dimensions. The alpha values and factor loadings pertaining to the34-item instrument are summarized in Table 1.

As shown in Table 1, five of the original 10 dimensions—tangibles,reliability, responsiveness, understanding/knowing customers, and access—remained distinct. The remaining five dimensions—communication,credibility, security, competence, and courtesy—collapsed into two dis-tinct dimensions (D4 and D5), each consisting of items from several of theoriginal five dimensions. The average pairwise correlation among theseven factors following oblique rotation was .27. This relatively low cor-relation, along with the relatively high factor loadings shown in Table 1,suggested that service quality might have seven fairly unique facets.

The high alpha values indicated good internal consistency among itemswithin each dimension. Moreover, the combined reliability for the 34-itemscale, computed by using the formula for the reliability of linear combina-tions (Nunnally 1978), was quite high (.94). Therefore, the 34-item in-

20

SERVQUAL

TABLE 1

Summary of Results from First Stage of Scale Purification

Dimension

Tangibles

Reliability

Responsiveness

CommunicationCredibilitySecurityCompetence

Courtesy

Understanding/KnowingCustomers

Label

Dl

D2

D3

D4

D5

D6

ReliabilityCoefficients

(Alphas)

.72

.83

,84

,79

.85

.85

Number ofItems

4

5

5

4

7

4

Factor Loadingsof Items

on Dimensionsto Which

They Belong"

63756247

7456737147

6073597666

35536656

41624750755254

80766277

21


Access D7 .78 5 5750755271

Reliability of LinearCombination .94

(Total-Scale Reliability)

• Numbers are the magnitudes of the factor loadings multiplied by 100. The loadings ofitems on dimensions to which they did not belong were all less than .3. The percentage ofvariance extracted by the seven factors was 61.7%.

strument was considered to be ready for further testing with data from newsamples.

Data Collection, Second Stage

To further evaluate the 34-item scale and its psychometric properties,data were collected pertaining to the service quality of four nationally-known firms: a bank, a credit-card company, a firm offering appliancerepair and maintenance services, and a long-distance telephone company.For each firm, an indpendent shopping-mall sample of 200 customers 25years-of-age or older were recruited by a marketing research firm in amajor metropolitan area in the East. To qualify for the study, respondentshad to have used the services of the firm in question within the past threemonths. Each sample was divided about equally between males and fe-males. As in the first stage of data collection, questionnaires were self-administered by qualified respondents.

Scale Purification, Second Stage

A major objective of this stage was to evaluate the robustness of the34-item scale when used to measure the service quality of the four firms.Therefore the data from each of the four samples were analyzed separatelyto obtain alpha values (along with corrected item-to-total correlations) anda factor-loading matrix following oblique rotation of a seven-factor solu-tion. The results from each sample facilitated cross-validation of the re-sults from the other samples.

The results of the four sets of analyses were quite consistent, but dif-fered somewhat from the first-stage findings summarized in Table 1. Spe-

22

SERVQUAL

cifically, two differences emerged. First, the corrected item-to-total corre-lations for several items (particularly among items making up the dimen-sions labeled D4 and D7 in Table 1) and the alphas for the correspondingdimensions were lower than those obtained from the first stage. Second,the factor-loading matrices obtained from all four analyses showed muchgreater overlap between dimensions D4 and D5, and between dimensionsD6 and D7. Because these differences occurred consistently across fourindependent samples and data sets, further purification of the 34-itemscale was deemed necessary.

A few items with relatively low item-to-total correlations were deleted.Furthermore, as suggested by the factor analyses, the items remaining inD4 and D5, as well as those in D6 and D7, were combined to form twoseparate dimensions. For each sample, alpha values were recomputed forthe reduced set of five dimensions and a factor analysis (involving extrac-tion of five factors followed by oblique rotation) was performed. In exam-ining the results of these analyses, an iterative sequence similar to the oneshown in Step 5 in Figure 1 was followed. This procedure resulted in arefined scale ("SERVQUAL") with 22 items spread among five dimen-sions (Dl, D2, D3, a combination of D4 and D5, and a combination of D6and D7). The expectation and perception statements in the final SERV-QUAL instrument are shown in the appendix.

An examination of the content of the final items making up each ofSERVQUAL's five dimensions (three original and two combined dimen-sions) suggested the following labels and concise definitions for the di-mensions:

Tangibles: Physical facilities, equipment, and appearance of per-sonnel

Reliability: Ability to perform the promised service dependablyand accurately

Responsiveness: Willingness to help customers and provide promptservice

Assurance: Knowledge and courtesy of employees and theirability to inspire trust and confidence

Empathy: Caring, individualized attention the firm provides itscustomers

The last two dimensions (assurance and empathy) contain items repre-senting seven original dimensions—communication, credibility, security,competence, courtesy, understanding/knowing customers, and access—that did not remain distinct after the two stages of scale purification.

23


Therefore, while SERVQUAL has only five distinct dimensions, theycapture facets of all 10 originally conceptualized dimensions.

SERVQUAL'S RELIABILITY AND FACTOR STRUCTURE

Table 2 shows the component and total reliabilities of SERVQUAL foreach of the four samples. The reliabilities are consistently high across allfour samples, with the possible exception of a couple of values pertainingto the tangible dimension. The total-scale reliability (i.e., reliability oflinear combination) is close to .9 in each of the four instances.

Results of the factor analyses of data from the four samples are summa-rized in Table 3. The overall patterns of factor loadings are remarkablysimilar across the four independent sets of results. With few exceptions,items assigned to each dimension consistently have high loadings on onlyone of the five factors extracted. The distinctiveness of SERVQUAL'sfive dimensions implied by the results in Table 3 was further supported byrelatively low intercorrelations among the five factors—the average pair-wise correlations between factors following oblique rotation were .21,.24, .26, and .23 for the bank, credit card, repair and maintenance, andlong-distance telephone samples, respectively.'

As an additional verification of the reliabilities and factor structure ofSERVQUAL, the first-stage data set that resulted in the 34-item instru-ment with seven dimensions was reanalyzed after deleting the 12 itemsthat dropped out during the second stage of scale purification. The resultsof this reanalysis are summarized in Table 4 and reconfirm the high reli-abilities and dimensional distinctiveness of the scale. The average pair-wise correlation among the five factors following oblique rotationwas .35.

It is worth noting that the interative procedure used to refine the initialinstrument was guided by empirical criteria and by the goal of obtaining aconcise scale whose items would be meaningful to a variety of servicefirms. The reliabilities and factor structures indicate that the final 22-itemscale and its five dimensions have sound and stable psychometric proper-ties. Moreover, by design, the iterative procedure retained only thoseitems that are common and relevant to all service firms included in thestudy. However, by the same token, this procedure may have deleted cer-tain "good" items relevant to some but not all firms. Therefore, whileSERVQUAL can be used in its present form to assess and compare service

' Complete matrices of the interfactor correlations can be obtained ftom the first author.

24

SERVQUAL

TABLE 2

Internal Consistencies of the Five Service-Quality DimensionsFollowing Second Stage of Scale Purification

Dimension

Tangibles

Reliability

Responsiveness

Assurance

Empathy

Label

Fl

F2

F3

F4

F5

Number ofItems

4

5

4

4

5

Reliability of Linear Combination(Total-Scale Reliability)

B

.52

.80

.72

.84

.71

.87

ReliabilityCoefficients

(Alphas)'

CC R&M

.62 .64

.78 .84

.69 .76

.80 .87

.80 .72

.89 .90

LDT

.64

.74

.70

.84

.76

.88

Items''

QlQ2

Q3Q4

Q5Q6Q7Q8Q9

QIOQllQ12Q13

Q14Q15Q16Q17

Q18Q19Q20Q21Q22

' B = Bank; CC = Credit Card Company; R&M = Repair and Maintenance Company;LDT = Long-Distance Telephone Company

'' The item numbers correspond to those of the expectation and perception statements in theappendix.

25


B

>

'S

B

Io

B5

"3cso

I2

U4

CQ

C4

PQ

O cN CNr- >n in

0 0

COO\ 00CO (N

>n

I I CO

c<i — - ^ 00r- 10 00 00

• * 00 00

tN CO -H O ONr-- vo r- 00 CO

10 VD

26

SERVQUAL

vo <N ^ —<l o <o r t

vo

O -H cs

vo O> VIoo 00 vo

in vo en ^vo r - r - vo

0 0

5IT) vo r~

555

CN ^Hen vo

ooCN

C^ C^ ON

r-* "O u^

I I00 Ov O -H cs-H -H cN CS CS

aaooa

° 2e^ _§•

c3 <u

i

8

.13. S•3 c

e §

•SIU t-i

t— 00 '—< en 00 3en Tf "^ en vo -g

llI

27


quality across a wide variety of firms or units within a firm, appropriateadaptation of the instrument may be desirable when only a single service isinvestigated. Specifically, items under each of the five dimensions can besuitably reworded and/or augmented to mEike them more germane to thecontext in which the instrument is to be used.

ASSESSMENT OF SERVQUAL'S VALIDITY

SERVQUAL's high reliabilities and consistent factor structures acrossseveral independent samples provide support for its trait validity (Camp-bell 1960; Peter 1981). However, while high reliabilities and internal con-sistencies are necessary conditions for a scale's construct validity—theextent to which a scale fully and unambiguously captures the underlying,unobservable, construct it is intended to measure—they are not sufficient(Churchill 1979). The scale must satisfy certain other conceptual and em-pirical criteria to be considered as having good construct validity.

The basic conceptual criterion pertaining to construct validity is face orcontent validity. (Does the scale appear to measure what it is supposed to?Do the scale items capture key facets of the unobservable construct beingmeasured?) Assessing a scale's content validity is necessarily qualitativerather than quantitative. It involves examining two aspects: (1) the thor-oughness with which the construct to be scaled and its domain were expli-cated and (2) the extent to which the scale items represent the construct'sdomain. As discussed in earlier sections, the procedures used in devel-oping SERVQUAL satisfied both these evaluative requirements. There-fore the scale can be considered to possess content validity.

The scale's validity was also assessed empirically by examining its con-vergent validity—i.e., the association between SERVQUAL scores andresponses to a question that asked customers to provide an overall qualityrating of the firm they were evaluating. Respondents in the second stage ofdata collection rated the service firm's overall quality (referred to hereafteras "Overall Q") by checking one of four categories—excellent, good,fair, poor. The correspondence between the Overall Q ratings and theSERVQUAL scores was examined using one-way ANOVA. The treat-ment variable in the ANOVA's was Overall Q—with three categoriesinstead of four because very few respondents checked "poor," therebynecessitating creation of a combined "fair/poor" category. The dependentvariable was the average difference score (i.e., perception-minus-expecta-tion score) on each SERVQUAL dimension as well as on the total SERV-QUAL scale (separate ANOVA's were conducted for each dimension andfor the total scale). Significant ANOVA results were investigated further

28

SERVQUAL

TABLE 4

Reanalysis of First-Stage Data for the Five-Dimensional Scale

Factor Loadingsof Items

Reliability on Dimensions

Dimension

Tangibles

Label

Fl

Number ofItems

4

Coefficients(Alphas)

.72

Items

QlQ2Q3Q4

to WhichThey Belong*

69686451

Reliability F2 .83 Q5Q6Q7Q8Q9

7563717550

Responsiveness

Assurance

F3

F4

4

4

.82

.81

QIOQllQ12Q13

Q14Q15Q16Q17

51776686

38728045

Empathy F5 .86 Q18Q19Q20Q21Q22

7881597168

Reliability of Linear Combination(Total-Scale Reliability) .92

" Numbers are magnitudes of the factor loadings multiplied by 100. The loadings of itemson dimensions to which they did not belong were all less than .3. The percentage of varianceextracted by the five factors was 63.2%.

using Duncan's multiple range test to identify significant differencesacross the Overall Q categories. The results of these analyses for eachoiF the four samples are summarized in Table 5 under the heading "Over-all Q".

The numbers reported in Table 5 are average SERVQUAL scores

29


within each Overall Q category, measured on a - 6 to +6 scale on whichthe higher (less negative) the score, the higher is the level of perceivedservice quality. In each of the four samples, the combined SERVQUALscore for those in the "excellent" category is significantly higher (lessnegative) than for those in the "good" category. Furthermore, respon-dents in the "good" category have a significantly higher combinedSERVQUAL score than those in the "fair/poor" category. A similar pat-tern of findings is evident for the scores on the individual SERVQUALdimensions as well. The strength and persistence of the linkage betweenthe Overall Q categories and the SERVQUAL scores across four indepen-dent samples offer strong support for SERVQUAL's convergent validity.

SERVQUAL'S validity was further assessed by examining whether theconstruct measured by it was empirically associated with measures ofother conceptually related variables. Respondents in each sample an-swered two general questions that provided measures of variables (labeled"Recommend" and "Problem" in Table 5) which one could expect to berelated conceptually to perceived service quality: (1) whether the respon-dents would recommend the service firm to a friend and (2) whether theyhad ever reported a problem with the services they received from the firm.Respondents answering yes to the first (Recommend) question and noto the second (Problem) question could be hypothesized to perceive higherservice quality than other respondents. As Table 5 shows, the results areconsistent with this hypothesis. These findings provide additional supportfor SERVQUAL's validity.

APPLICATIONS OF SERVQUAL

It is difficult to identify any retailer that offers no services whatsoever.Some retailers offer facilitating services, such as sales assistance and de-livery, to help sell goods. Some retailers sell services directly, in additionto offering facilitating services. Some retailers sell only services. Qualityof service is an important issue for all of these retailers. Competing goodsretailers (department stores, supermarkets) may sell many identicalproducts and quality of service is a primary means of competitive differ-entiation. Retailers that sell only services (telephone companies, airlines)have little to offer if their service is poor (Berry 1986).

SERVQUAL is a concise multiple-item scale with good reliability andvalidity that retailers can use to better understand the service expectationsand perceptions of consumers and, as a result, improve service. The in-strument has been designed to be applicable across a broad spectrum ofservices. As such, it provides a basic skeleton through its expectations/

30

SERVQUAL

perceptions format encompassing statements for each of the five service-quality dimensions. The skeleton, when necessary, can be adapted or sup-plemented to fit the characteristics or specific research needs of a partic-ular organization.

SERVQUAL is most valuable when it is used periodically to track ser-vice quality trends, and when it is used in conjunction with other forms ofservice quality measurement. A retailer, for example, would leam a greatdeal about its service quality and what needs to be done to improve it byadministering both SERVQUAL and an employee survey three or fourtimes a year, plus systematically soliciting and analyzing customer sug-gestions and complaints. The employee survey should include questionsconcerning perceived impediments to better service, e.g., what is the big-gest problem you face trying to deliver high-quality service to your cus-tomers? If you could be president for a day, what one change would youmake in the company to improve quality of service?

SERVQUAL can be used to assess a given firm's quality along each ofthe five service dimensions by averaging the difference scores on itemsmaking up the dimension. It can also provide an overall measure of ser-vice quality in the form of an average score across all five dimensions.Because meaningful responses to the perception statements require re-spondents to have some knowledge of or experience with the firm beingresearched, SERVQUAL is limited to current or past customers of thatfirm. Within this constraint, a variety of potential applications are avail-able.

One potential application of SERVQUAL is to determine the relativeimportance of the five dimensions in influencing customers' overallquality perceptions. An approach for doing this is to regress the overallquality perception scores on the SERVQUAL scores for the individualdimensions. The results of such a regression analysis for the four compa-nies in the present study are shown in Table 6 (the dependent variable wasOverall Q, coded as excellent = 4, good = 3, fair = 2, and poor = 1).

The adjusted R^ values are statistically significant in all four cases andare also quite respectable, particularly in view of the fact that the depen-dent variable had only four categories, and the first three accounted formost of the responses. A striking result in terms of the relative importanceof the five dimensions in predicting overall quality is that reliability isconsistently the most critical dimension. Assurance is the second mostimportant dimension in all four cases. Tangibles is more important in thecase of the bank than in the other three firms, while the reverse is true forresponsiveness. Empathy is the least important dimension in all fourcases. However, the relatively small magnitudes of the regression coeffi-

31


)les

et*2ea

aiJS

2one

o<j

u<

1II

ITS

O

c<

for

73>

Seal

e

. f i

nd

Pro

cs•ae

IO

20)

6

aci:

s

ecom

r

o

vera

Oo

00 §CO CA

•5 g^ Q

u .o u .c u(N 'd- so -H oOS 00 OS OS 00

O O O O OI I I I I

S> XI SI SI .Dm >o ts tN r-r-; lo tN "/ oO —> —<

I I I

00 —< -H ( S 00ON fN 00 -H 00

I I

• * 00 VOr-~ 00 r-• ^ 00

O O O OI I I I

OI

OO

o

1

?5enCN

1

u

1

OOON

1

u

1

CN *O t— en CNOS OS O O

O OI I

O -H -H

I I I

>o.C .£3OS O• ^ enen

O O O O OI I I I I

CAt/i<uB<U

U .-3 CA »

as û « t« c

I ;2

SI

I I I

I I

OSO CN

cso

C3U

CA a>

J3 B (u

= . li|lt3 « « 2 c o caH ai BS < W U 00

, 1 ^

I

B

6Soo(D

oz

CA

> ^

o0 0

o1

OSenO1

hCN

1

SI

OO

o1

OS

<N

1

SI

r—o1

enooCN

1

jo00O

1

t—

CN

1

O

1

ur-CN

CN

1

X I

ood1

r-

183

ex

oU

1

o

c/3 _O

u u u X Xo\ CN r- >o -^cs 00 (~~ —I Od CD CD -^ '-^

I I I I I

o

c3

ON

o1

?enCN

1

r~- CN

^ ^ CN

1 1

OS^^

1

uB

. • s•a;! a g ji

moc>O OSI 2

v£) CN -H OS c s OSr-; TJ| en -^ so csd —«—<—<—< —< oI I I I I I "^

O\

^ t« 5^ So oOS ^^ en en —

O O —<—'—' —< CNI I I I I \ z :

X)soOO

42"

o1

08"

o1

59"

o1

50"

o1

32"

o1

§ 13., "o q « eH ot; a! < w

u13u

•S .S u

J l32

1o

o

BBs1Ul

a

UI3

men

d

go

i

Ove

ra]

9

oZ

Yes

o

CA

> ^

1[/J

1 3

Goo

Uen

t

X

>

C^ §

idiv

idua

lD

imen

si

I.

X I

en

1

00

O1

85"

o1

36"

o1

0 0

o1

O1

o1

angi

bles

&

1

T - H

fN1

48=

en1

X)

.—,

1

hen1

hen

1

48"

o1

elia

bilit

y

05

o1

p:

1

54=

CN

1

83"

o1

qpCN

1

Soo

1

08"

o1

espo

nsiv

(

OS

u

* . H

1

s1

91=

1

16"

1

T3

OO

CN

1

o

en

1

33"

o1

ssur

ance

o1

r-

1

3N

1

85"

o1

17"

1

u

T - H

1

15"

o

mpa

thy

u

0 0

d1

1

&

CN

1

88"

o1

enCN1

ur-q1

Xivo

d1

ombi

ned

U

132

o

oo

114

ampl

e Si

;

C

1oU

oO L

1CA

Q

1

men

d

6o

Rec

S!13

Ove

i

oZ

Yes

oZ

c/1

?«

1

T3

OQo

Uen

txc

e

m

^ o

idiv

idua

lD

imen

si

CN

o1

o1

u

o1

.26"

o1

.50=

o1

..•ô1

obOO1

angi

bles

H

o1

.54'

1

ur-

1

o- H

1

•aen

CN

1

CN• ^

1

X I

1

elia

bilit

v

OS

ooO1

%

1

uenO(N

1

&o. - H

1

B1

en"^

1

.30"

1

CA

c

0c

u

r-0 0

o1

vo

1

.64=

CN

1

&o1

oCN

1

IT)

1

.39"

1

Lssu

ranc

e

ON

o1

.16

1

.34=

CN

1

.86"

o1

1

ON'—'

1

.33"

1

mpa

thy

Pu

SERVQUAL

d1

CN

\

.13=

CN

1

x>enoo

d1

•aen

1

U

'—'

1

x>Oend1

o

lom

bine

d

u

120

00

r-

ON

OO

CN

vo

ampl

e Si

>

ide,

neg

103

pect

atio

ns

u

and

tions

:ions

.le

ctal

r pe

rcep

:eed

exp

OJ

tcon

s

IBl)

.i

X

iptio

ni

Jasx:cj

§

+

1

Igin

g fr

onI

scal

e 1

CQ

g

"3âtu

. are

;'

Num

bers

>.

E

ilu

C3>

:iti

a

, an

dxp

ecta

tioi

shor

t ol

fall

VI

am

s-S.

that

dues

im

ply

>

difi

eren

t.tly

tican

1

ipts

are

:

b

1CO

s

lean

s

.2

seJl

tl

S

ifi;

not

sign

i

scri

pts

1

•S_C2

i w

itl

i

33


TABLE 6

Relative Importance of the Five Dimensions in Predicting

Ditnension

BankTangiblesReliabilityResponsivenessAssuranceEmpathy

Credit Card Co,TangiblesReliabilityResponsivenessAssuranceEmpathy

Repair & MaintenanceTangiblesReliabilityResponsivenessAssuranceEmpathy

L-D Telephone Co,TangiblesReliabilityResponsivenessAssuranceEmpathy

Overall

StandardizedSlope

Coefficient

,13,39,07,13,01

,07,33,12,17,04

Co.,04,54,11,16.01

,08,45,12.15,02

" Significance levels are for two-tailed tests

Quality

SignificanceLevel ofSlope^

,07,00.35.09.89

.26

.00,11,02,58

.48

.00,09.02.81

.17

.00

.09

.03

.78

AdustedR2

.28 (p < .00)

.27 (p < ,00)

,52 (p < ,00)

,37 (p < ,00)

cients for empathy and their lack of statistical significance should be inter-preted with caution because empathy did have a statistically significantsimple correlation with overall quality, ranging from .20 in the case of thebank to .40 in the case of the repair and maintenance company. Empathyalso had significant correlations of the same order of magnitude with reli-

34

SERVQUAL

ability and assurance (the two most important dimensions), thereby im-plying that its importance in the regression analyses may have beenmasked somewhat by possible multicollinearity. Therefore, while em-pathy is apparently the least important of the five SERVQUAL dimen-sions, it is by no means wnimportant.

Another application of the instrument is its use in categorizing a firm'scustomers into several perceived-quality segments (e.g., high, medium,and low) on the basis of their individual SERVQUAL scores. These seg-ments then can be analyzed on the basis of (1) demographic, psychogra-phic and/or other profiles; (2) the relative importance of the five dimen-sions in infiuencing service quality perceptions; and (3) the reasons behindthe perceptions reported. For example, suppose a department store foundthat a large number of SERVQUAL respondents failing in the "medium"perceived-quality group fit its prime target market based on demographicand psychographic criteria. Suppose further that reliability and assurancewere found to be the most important quality dimensions and, based onperception-expectation gap scores for items concerning these dimensions,the items relating to record-keeping accuracy and behavior of contact per-sonnel revealed the biggest gaps. With these data, the department store'smanagement would understand better what needs to be done to improve itsimage in the eyes of a very important group—customers within the firm'sprime target market who give the firm "medium" service quality scoresand who are in position to either respond to improved service from thefirm or defect to the competition.

SERVQUAL can also be used by multi-unit retail companies to trackthe level of service provided by each store in the chain. By asking respon-dents to indicate the particular store in the chain with which they are mostfamiliar, and to provide perception responses for that unit, the researchercan compare each store's average SERVQUAL score with the scores fromother stores. Service quality scores can then be a factor in store managerperformance appraisals and compensation, among other uses. Aiso,SERVQUAL scores for the individual stores can be used to group outletsinto several clusters with varying quality images, A careful examination ofthe characteristics of the stores in the different clusters may reveai keyattributes that facilitate—or hinder—the delivery of high quality service.

A retailer can also use SERVQUAL to assess its service performancerelative to its principal competitors. The two-section format of the instru-ment, with separate expectation and perception sections, makes it conve-nient to measure the quality of several firms simply by including a set ofperception statements for each firm. The expectations section does nothave to be repeated for each firm. For example, a supermarket chain could

35


include its two principal competitors in a total market survey, asking re-spondents to provide perception ratings for each of the companies withwhich they have shopping experience, A retailer that uses SERVQUAL toidentify the most salient service quality dimensions for its target markets,and to compare itself to the competition in terms of strengths and weak-nesses on these particular dimensions, will certainly have a sense of whatits priorities should be with regard to service quality.

In summary, SERVQUAL has a variety of potential applications. It canhelp a wide range of service and retailing organizations in assessing con-sumer expectations about and perceptions of service quality. It can alsoheip in pinpointing areas requiring managerial attention and action to im-prove service quality. In addition, we hope the availability of this instru-ment win stimulate much-needed empirical research focusing on servicequality and its antecedents and consequences.

REFERENCES

Berry, Leonard L, (1986), "Retail Businesses Are Service Businesses," Journal of Re-tailing, 62 (Spring), 3—6.

Campbell, Donald T, (1960), "Recommendations for APA Test Standards Regarding Con-struct, Trait, or Discriminant Validity," American Psychologist, 15 (August), 546-53,

Churchill, Gilbert A., Jr, (1979), "A Paradigm for Developing Better Measures of Mar-keting Constructs," Journal of Marketing Research, 16 (February), 64-73,

Churchill, Gilbert A,, Jr., Neil M, Ford, and Orville C, Walker, Jr. (1974), "Measuring theJob Satisfaction of Industrial Salesmen," Journal of Marketing Research, 11 (August),254-260.

Cronbach, Lee J. (1951), "Coefficient Alpha and the Internal Structure of Tests," Psycho-metrika, 16 (October), 297-334.

Crosby, Philip B. (1979), Quality is Free: The Art of Making Quality Certain, New York:New American Library,

Dodds, William B, and Kent B. Monroe (1984), "The Effect of Brand and Price Informationon Subjective Product Evaluations," Advances in Consumer Research XII.

Ford, Neil M,, Orville C, Walker, Jr,, and Gilbert A. Churchill, Jr, (1975), "Expectation-Specific Measures of the Intersender Conflict and Role Ambiguity Experienced by Indus-trial Salesmen," Journal of Business Research, 3 (April), 95-112,

Garvin, David A, (1983), "Quality on the Line," Harvard Business Review, 61 (Sep-tember-October), 65-73.

Gronroos, Christian (1982), Strategic Management and Marketing in the Service Sector,Helsingfors: Swedish School of Economics and Business Administration,

Harman, Harry H, (1967), Modern Factor Analysis, 2nd ed,, Chicago: The University ofChicago Press,

Hjorth-Anderson, Chr, (1984), "The Concept of Quality and the Efficiency of Markets forConsumer Products," Journal of Consumer Research, 11 (September), 708-718,

Holbrook, Morris B, and Kim P, Corfman (1985), "Quality and Value in the ConsumptionExperience: Phaldrus Rides Again," in Perceived Quality, J, Jacoby and J. Olson (eds,),Lexington, Massachusetts: Lexington Books, 31-57,

36

SERVQUAL

Howard, John and Jagdish Sheth (1969), The Theory of Buyer Behavior, New York: JohnWiley and Sons.

Hunt, Keith (1979), Conceptualization and Measurement of Consumer Satisfaction and Dis-satisfaction, Cambridge, Mass,: Marketing Science Institute.

Jacoby, Jacob (1978), "Consumer Research: A State of the Art Review," Journal of Mar-keting, 42 (April), 87-96.

Jacoby, J. and J. Olson, eds. (1985), Perceived Quality, Lexington, Massachusetts: Lex-ington Books,

Lehtinen, Uolevi and Jarmo R. Lehtinen (1982), "Service Quality: A Study of Quality Di-mensions," unpublished working paper, Helsinki: Service Management Institute, Fin-land OY.

Lovelock, Christopher H. (1980), "Towards a Classification of Services," in EmergingPerspectives on Services Marketing, L. L. Berry, G. L. Shostack, and G, Upah (eds,),Chicago: American Marketing Association, 72-76,

(1983), "Classifying Services to Gain Strategic Marketing Insights," Journal ofMarketing, 47 (Summer), 9-20,

Nunnally, Jum C, (1978), Psychometric Theory, 2nd ed,. New York: McGraw-Hill BookCompany.

Oliver, Richard (1981), "Measurement and Evaluation of Satisfaction Process in Retail Set-tings," Journal of Retailing, 57 (Fall), 25-48,

Olshavsky, Richard W. (1985), "Perceived Quality in Consumer Decision Making: An Inte-grated Theoretical Perspective," in Perceived Quality, J, Jacoby and J, Olson (eds,),Lexington, Massachusetts: Lexington Books.

Parasuraman, A., Valarie Zeithaml, and Leonard Berry (1985), "A Conceptual Model ofService Quality and Its Implications for Future Research," Journal of Marketing (Fall),41-50.

Peter, J, Paul (1981), "Construct Validity: A Review of Basic Issues and Marketing Prac-tices," iour/jo/o/Marifce/mg ê«arc/i, 18 (May), 133-45,

Rudie, Mary, J. and H. Brant Wansley (1985), "The Merrill Lynch Quality Program," inServices Marketing in a Changing Environment, Thomas Bloch, Gregory Upah, and Va-larie A. Zeithaml (eds.), Chicago, IL: American Marketing Association.

Sasser, W. Earl, Jr., R. Paul Olsen, and D. Daryl Wyckoff (1978), Management of ServiceOperations: Text and Cases, Boston: Allyn & Bacon.

Saxe, Robert and Barton A. Weitz (1982), "The SOCO Scale: A Measure of the CustomerOrientation of Salespeople," Journal of Marketing, 19 (August), 343-51.

Thompson, Phillip, Glenn DeSouza, and Bradley T. Gale (1985), The Strategic Managementof Service Quality, Cambridge, MA: The Strategic Planning Institute, PIMSLETTERNo. 33.

Zeithaml, Valarie (1987), Defining and Relating Price, Perceived Quality, and PerceivedValue, Report No, 87-101, Cambridge, MA: Marketing Science Institute.

37


APPENDIX

THE SERVQUAL INSTRUMENT"

DIRECTIONS: This survey deals with your opinions of services.Please show the extent to which you think firms offering servicesshould possess the features described by each statement. Do this hypicking one of the seven numbers next to each statement. If you stronglyagree that these firms should possess a feature, circle the number 7. If youstrongly disagree that these firms should possess a feature, circle 1. If yourfeelings are not strong, circle one of the numbers in the middle. There areno right or wrong answers—all we are interested in is a number that bestshows your expectations about firms offering services.

El . They should have up-to-date equipment.E2. Their physical facilities should be visually appealing.E3. Their employees should be well dressed and appear neat.E4. The appearance of the physical facilities of these firms should be in

keeping with the type of services provided.E5. When these firms promise to do something by a certain time, they

should do so.E6. When customers have problems, these firms should be sympathetic

and reassuring.E7. These firms should be dependable.E8. They should provide their services at the time they promise to

do so.E9. They should keep their records accurately.ElO. They shouldn't be expected to tell customers exactly when services

will be performed. ( —)''E l l . It is not realistic for customers to expect prompt service from em-

ployees of these firms. ( - )E12. Their employees don't always have to be willing to help cus-

tomers. ( - )El3. It is okay if they are too busy to respond to customer requests

promptly. ( —)E14. Customers should be able to trust employees of these firms.El5. Customers should be able to feel safe in their transactions with

these firms' employees.El6. Their employees should be polite.

38

SERVQUAL

E17. Their employees should get adequate support from these firms todo their jobs well.

El8. These firms should not be expected to give customers individualattention. ( - )

E19. Employees of these firms cannot be expected to give customerspersonal attention. ( —)

E20. It is unrealistic to expect employees to know what the needs oftheir customers are. ( - )

E21. It is unrealistic to expect these firms to have their customers' bestinterests at heart. ( - )

E22. They shouldn't be expected to have operating hours convenient toall their customers. ( - )

DIRECTIONS: The following set of statements relate to your feelingsabout XYZ. For each statement, please show the extent to which youbelieve XYZ has the feature described by the statement. Once again, cir-cling a 7 means that you strongly agree that XYZ has that feature, andcircling a 1 means that you strongly disagree. You may circle any of thenumbers in the middle that show how strong your feelings are. There areno right or wrong answers—all we are interested in is a number that bestshows your perceptions about XYZ.

Fl. XYZ has up-to-date equipment.P2. XYZ's physical facilities are visually appeaUng.P3. XYZ's employees are well dressed and appear neat.P4. The appearance of the physical facilities of XYZ is in keeping with

the type of services provided.P5. When XYZ promises to do something by a certain time, it does so.P6. When you have problems, XYZ is sympathetic and reassuring.P7. XYZ is dependable.P8. XYZ provides its services at the time it promises to do so.P9. XYZ keeps its records accurately.PIO. XYZ does not tell customers exactly when services will be per-

formed. ( - )PI 1. You do not receive prompt service from XYZ's employees. ( - )P12. Employees of XYZ are not always willing to help customers. ( - )PI3. Employees of XYZ are too busy to respond to customer requests

promptly. ( - )P14. You can trust employees of XYZ.P15. You feel safe in your transactions with XYZ's employees.P16. Employees of XYZ are polite.

39


P17. Employees get adequate support from XYZ to do their jobs well.P18. XYZ does not give you individual attention. ( - )P19. Employees of XYZ do not give you personal attention. ( - )P20. Employees of XYZ do not know what your needs are. ( - )P21. XYZ does not have your best interests at heart. ( - )P22. XYZ does not have operating hours convenient to all their cus-

tomers. ( —)

" A seven-point scale ranging from "Strongly Agree" (7) to "Strongly Disagree" (1), withno verbal labels for the intermediate scale points (i.e., 2 through 6), accompanied eachstatement. Also, the statements were in random order in the questionnaire. A complete listingof the 34-item instrument used in the second stage of data collection can be obtained from thefirst author.'' Ratings on these statements were reverse-scored prior to data analysis.

40

SERVQUAL: A Multiple-Item Scale for Measuring Consumer ... · SERVQUAL) for assessing customer perceptions of service quality in service and retailing organizations. After a discussion

Documents