Service quality gaps of business customers in the shipping industry

Transportation Research Part E 45 (2009) 222–237

Contents lists available at ScienceDirect

Transportation Research Part E journal homepage: www.elsevier.com/locate/tre

Service quality gaps of business customers in the shipping industry Kee-Kuo Chen a,*, Ching-Ter Chang b,1, Cheng-Sheng Lai a,2 a b

Department of Shipping and Transportation Management, National Taiwan Ocean University, 2 Pei-Ning Road, Keelung 202, Taiwan, ROC Department of Information Management, Changhua 500, National Changhua University of Education, 1 Jin-de Road, Changhua, Taiwan, ROC

a r t i c l e

i n f o

Article history: Received 1 August 2007 Received in revised form 12 December 2007 Accepted 28 February 2008

Keywords: Service quality SERVQUAL CFA MANOVA

a b s t r a c t This paper extends the gaps model of [Zeithaml, V.A., Parasuraman, A., Berry, L. 1990. Delivering Quality Service: Balancing Customer Perceptions and Expectation. The Free Press, New York] from the service provider to the business customer side by examining two service quality (SQ) gaps. One is the SQ gap between types of business customers and the other is the SQ gap among employee statuses of business customers. Besides that, the five-factor SERVQUAL measure as the initial hypothesized model is also tested. The applicability of SERVQUAL to measuring the perceived SQ of customers in the shipping industry of Taiwan is rejected empirically. The existence of the two hypothesized gaps is verified by the method of MANOVA. Ó 2008 Elsevier Ltd. All rights reserved.

1. Introduction The ‘‘gaps model” of service quality (SQ) that views services in a structured, integrated way was developed by Zeithaml et al. (1990). They asserted that closing the gap between what customers expect and what they perceive is critical to delivering quality service. As seen, their model emphasizes the provider gaps occurring within the organization that provides the services. However, the gaps model is too simple to describe the SQ perceived by business customers. We believe that there exist SQ gaps between types of business customers and among employee statuses of business customers. This article will examine these two gaps with a sample of customers from an international line shipping company of Taiwan. The inference reliability of an empirical research will not be accepted without using a scale that has been validated academically. SERVQUAL is designed to measure service quality perceived by the respondents. The basic model was that consumer perceptions of quality emerge from the gap between performance (perception) and expectations. When performance exceeds expectations, quality increases; on the other hand, when performance falls short of expectations, quality decreases. Thus, performance-to-expectation ‘‘gaps” on attributes that consumers use to evaluate the quality of a service form the theoretical foundation of SERVQUAL. Using factor analysis of several applications of the scale, Parasuraman et al. (1985, 1988, 1991, 1994) identified that the SERVQUAL scale has five perceptual dimensions, namely (1) tangibles, (2) reliability, (3) responsiveness, (4) assurance, and (5) empathy. The psychometric properties of SERVQUAL have been examined in many researches, however, the results have been mixed in the industrial setting (Brady et al., 2002; Cronin and Taylor, 1992; Durvasula et al., 1999; Teas, 1993). The need of further investigations for the validity of SERVQUAL has been called for by researchers (Durvasula et al., 1999). As a response to this, SERVQUAL will be used as the measurement scale in the current study. That is, we considered the five-factor SERVQUAL measure as the initial hypothesized model to be tested. The main

* Corresponding author. Tel.: +886 2 24622192x3437; fax: +886 2 24631903. E-mail addresses: [email protected] (K.-K. Chen), [email protected] (C.-T. Chang), [email protected] (C.-S. Lai). 1 Tel.: +886 4 7232105. 2 Tel.: +886 2 24622192x3402. 1366-5545/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.tre.2008.02.005

K.-K. Chen et al. / Transportation Research Part E 45 (2009) 222–237

223

advantage of the scale is that it can assess the SQ along each of the service items and of the dimensions (Zeithaml et al., 2006) if the validity of the application can be verified empirically. The first objective of this paper is to test the validity of applying the SERVQUAL to the shipping industry of Taiwan using confirmatory factor analysis (CFA). The second objective is to examine the hypothesized gaps using the method of factor structure invariance if the application of SERVQUAL to the shipping industry of Taiwan has been validated; otherwise the two gaps will be tested using multivariate analysis of variance (MANOVA). MANOVA will test the differences of the perceived SQ item by item. The rest of the paper is organized as follows. The SERVQUAL scale and the gaps model are described briefly in Section 2. The two gaps of business customers will be explained in Section 3. The methodology is demonstrated in Section 4. Section 5 presents the empirical results. Discussion of the findings and the implications for the management and future research are detailed in Section 6. The final section contains the conclusion. 2. SERVQUAL and gaps model The literature is very rich in terms of definition, dimensions, models and measurement issues in service quality (Asubonteng et al., 1996; Bitner et al., 1990; Dabholkar et al., 2000; Parasuraman et al., 1985, 1988, 1994), supported by a number of empirical studies from a variety of service-related application areas (Badri et al., 2005; Davis and Mentzer, 2006; Mohsin, 2005; Nadiri and Hussain, 2005; Peiro et al., 2005; Rohini and Mahadevappa, 2006; Seth et al., 2006). Some of the contemporary definitions of service quality from the literature were given in Seth et al. (2006). Among them, the SERVQUAL scale (Parasuraman et al., 1988, 1991, 1994) is designed to measure service quality perceived by the respondents from five different service categories: retail banking, long-distance telephone, securities brokerage, appliance repair and maintenance firm, and credit cards. In their original formulation, Parasuraman et al. (1985) identified 10 components of SQ from 97 service items. Subsequent empirical work involved exploratory research of these 10 components, which were collapsed into five dimensions: reliability, responsiveness, assurance, empathy and tangibles. These five dimensions along 22 service items yielded the SERVQUAL scale for measuring SQ (Parasuraman et al., 1988). While the SERVQUAL scale has been revised, refined and reformed, its primary content has remained unaltered (Parasuraman et al., 1991, 1994). Many previous works on service quality have been developed around the SERVQUAL model (Davis and Mentzer, 2006; Johnston and Clark, 2001; Nadiri and Hussain, 2005; Parasuraman et al., 1988; Peiro et al., 2005). A detailed survey of the literature on the applications of SERVQUAL can be found in Badri et al. (2005). Unfortunately, the conceptualization and measurement of service quality using SERVQUAL are not bereft of controversy. The instrument’s psychometric properties have been seriously questioned in many contexts (Asubonteng et al., 1996; Buttle, 1996; Durvasula et al., 1999; Gounaris, 2005; Badri et al., 2005). Cronin and Taylor (1992) argued for ‘‘Performance only” measurement of service quality and proposed a service quality measurement tool called SERVPERF. Teas (1993) addressed the measurement of expectations and presented the Normed Quality and Evaluated Performance model. Durvasula et al. (1999) found that the psychometric properties of the SERVPERF scores had better prediction ability than the SERVQUAL gap scores in the ocean freight services environment. Despite the criticisms, Parasuraman et al. (1991, 1994) contended that the SERVQUAL scale using the expectation/performance gaps method is a much richer approach to measuring service quality. For developing the SERVQUAL scale, Zeithaml et al. (1990) proposed the ‘‘gaps model” of SQ viewing services in a structured, integrated way at the same time. The gaps model suggests that to understand and improve the SQ delivered by the company, the managers have to close the four provider gaps (see Fig. 2). However, the gaps model did not discuss the possible distinctions of SQ perceptions among business customers. The number and composition of service quality dimensions are likely to be dependent on how complicated the service settings are. The complexity of business customers having different objectives or of employees with different statuses will cause the SQ gap between types of business customers and among employee statuses of business customers. Extant research on business-to-business SQ had not mentioned these possible issues either (Carr, 2002; Durvasula et al., 1999; Gounaris, 2005). By attempting to examine the two hypothesized gaps, this article will extend the provider gaps in the model of Zeithaml et al. to the gaps of business customers. 3. SQ gap of business customers 3.1. SQ gap between types of business customers A line company usually has two kinds of business customers – shippers and forwarders. There are at least two reasons that will make the perceived SQ of the shippers and the forwarders towards a line company different. On the one hand, although both the shippers and forwarders have business with the line companies, their motivations are different. Shippers require line company services for cargo delivery. Instead of contracting and booking the container berth directly with a line, shippers can employ a forwarder as their agents to take care of cargo delivery as shown in Fig. 1. Whether a shipper contacts directly a line or employs a forwarder as agent depends on the freight charges the shipper has to pay and the SQ the shipper perceived. Forwarders act as agents for the shippers to contract container berths with the lines for their own profit. The container handling and related services are still performed by the lines. In order to earn profit and reputation, the forwarders often ask the lines to provide extra services to their shipper customers. For example, the forwarders may sometimes ask

224

K.-K. Chen et al. / Transportation Research Part E 45 (2009) 222–237

Booking

Booking

Issue B/L

Cut off & On board

Line

Forwarder

Shipper

Cargo delivery

Issue B/L

Cargo

Bank negotiation for payment

from bank

FFWR B/L

Document

Release D/O

Released to

(cargo delivery)

Exchange D/O against B/L

D/O released

Shipping Line in Overseas country

Exchange D/O against

Importer

L/C Issuing bank

L/C Negotiating bank

Redeemed

Forwarder in Oversea Country

Document

(cargo delivery)

importer

Abbreviations. B/L: Bill of Lading; D/O: Delivery order; L/C: Letter of Credit Fig. 1. Flow chart of cargo delivery.

the lines to offer extra container spaces to satisfy their customers’ temporary needs. Otherwise, the forwarders will have little to attract shippers to employ them as agents. However, the impact of the SQ delivered by lines on the profit of shippers is not as important as that on the profit of forwarders because most of the profits of shippers result from the marginal contributions of products sold. On the other hand, forwarders are well informed of working steps in the process of cargo delivery and the extent of service quality that they can obtain from each line in the market. Thus, the forwarders’ perceived SQ would be lower than the shippers’ perceived SQ when the forwarders know that the line does not do the best to serve them. However, shippers are usually unable to grasp this situation. In view of the above, we hypothesize that there exists the SQ gap between the forwarders and shippers in the shipping industry of Taiwan and that the forwarders’ perceived SQ of the lines is lower than the shippers’ perceived SQ. 3.2. SQ gap among employee statuses Another SQ gap would arise within an organization of business customer because cargo delivery involves many steps as shown in Fig. 1. To begin with, a cargo delivery contract is usually made out by a manager of a shipper or of a forwarder with a manager of a liner. After that the subsequent steps in the flow of cargo delivery are charged by employees of low status in the organizations. These subordinates of the shippers or of the forwarders will contact the frontline line service providers. The responsibility of the managers of both shippers and forwarders is only to monitor that their subordinates execute smoothly the jobs to be done. The perceived SQ of managers and their subordinates will be different because their perceptions come from different sources. Fig. 2 describes the sources of SQ perception and the SQ gaps. Gaps 1 to 4 in Fig. 2 are the same as those in the gaps model of Zeithaml et al. (1990). The difference between Fig. 2 and the gaps model of Zeithaml et al. is seen in the upper half of the diagram. The perceived SQ of the shippers’ or the forwarders’ managers come mainly from the reports submitted by their subordinates or by word of mouth from other managers (Robertson, 1971), although they also have first-hand experience of the services delivered by the managers of the liners when the contract is negotiated and made out (Gilmore and Pine, 2002). The SQ perceived by subordinates comes mainly from the working experiences with frontline service providers of the liners. To examine the SQ of restaurants, Peiro et al. (2005) found that the employees usually overestimate the functional and relational SQ. This can also occur among the business customers. Unlike the employees, bosses are responsible for maximizing the wealth of the stockholders or the partners. Hence, they have to pursue excellence. As a matter of fact, they will tend to be stricter when rating the SQ delivered by the providers than the employees. According to these differences in personal needs, past experience and word of mouth between managers and subordinates, we hypothesize that there exists the SQ gap among employee statuses of business customers in the shipping industry of Taiwan and that the perceived SQ of managers is lower than that of their subordinates.

K.-K. Chen et al. / Transportation Research Part E 45 (2009) 222–237

Business Customer

Perceived service of management

Management Perceived Service Quality

Personal needs

Past Experience

Service Quality Gap within Enterprise

Expected service of management

Word of mouth communications

225

Expected service of Employee Employee Perceived Service Quality Perceived service of Employee

Service Provider

Service delivery (including pre- and post-contacts

GAP 4

External communications to customer

GAP 3 GAP 1

Translation of perceptions into service quality specifications GAP 2 Management Perception of Customer Expectation Fig. 2. The extended gaps model of SQ.

4. Research methodology 4.1. Sample Customer samples of both forwarder and shipper were randomly selected from a list of an international line company of Taiwan. In each selected customer, a manager and a subordinate who had been in touch with some frontline service providers of the line were surveyed. Lines often divide their organizations into a number of specialized departments and supporting activities. Perceptions of managers and subordinates concerning lines’ SQ of customers might be shaped by their experiences with the interfacing departments. Durvasula et al. (1999) showed that respondents rated high coefficient alpha reliabilities exceeding 0.9 on these interfacing departments and activities. Therefore, the respondents were not requested to identify which department or activity in the shipping line they interact with. Skilled interviewers who had experience in conducting managerial interviews administered the survey. First, a letter was sent to invite candidate managers and employees, respectively, to participate in the study. If the invitation was accepted, the interviewers delivered the questionnaire and then retrieved it upon completion. If some candidate managers and employees refused participation, we deleted their names from the study and proceeded to the next candidate until 110 valid questionnaires were collected in each category of respondents in compliance with the minimum sample size for the 22 items to be analyzed (Hair et al., 2006). Table 1 summarizes the profile of the surveyed data. A thorough review by experts (both academia and practitioners) in the shipping industry revealed little negative opinion against the SERVQUAL instrument. Therefore, the SERVQUAL instrument without modification is used in this study. The SERVQUAL scale was then administered to the respondents, and their perceptions and expectations of the SQ delivered by the line were obtained using the 22-item scale (see Appendix). The SERVQUAL gap scores were then obtained by subtracting SERVQUAL perception scores from expectation scores. Two other measures were also employed for SERVQUAL validation purposes. One is overall satisfactory evaluation of the line’s service, which is measured on a scale with 1 = extremely poor and 7 = excellent. Another is the long-term relationship proxy, which is measured on a scale with 1 = cargo volume shipped by the line is decreased, 2 = cargo volume shipped by the line is the same as before and 3 = cargo volume shipped by the line is increased.

226

K.-K. Chen et al. / Transportation Research Part E 45 (2009) 222–237

Table 1 Profile of sample data Population

Shipper

Status of respondents

Managers

Employees

Forwarders Managers

Employees

Number of visits to provide 110 responses Response rate (%)

256 43

225 49

192 57

205 54

4.2. SERVQUAL measurement model A diagram depicting the SERVQUAL measurement model is shown in Fig. 3 (Hair et al., 2006). The definitions (Zeithaml et al., 2006) of five constructs in this figure are: (1) reliability: ability to perform the promised service dependably; (2) responsiveness: willingness to help customers and provide prompt service; (3) assurance: employees’ knowledge and courtesy and their ability to inspire trust and confidence; (4) empathy: caring, individualized attention given to customers; and (5) tangibles: appearance of physical facilities, equipment, personnel, and written materials. For clear presentation and description of this diagram, the definitions of variables and symbols are explained in Table 2. In Fig. 3, n1 represents the reliability construct which causes measured variables A1 to A5; n2 represents the responsiveness construct which contains measured variables B1 to B4; n3 represents the assurance construct which causes measured variables C1 to C4; n4 represents the empathy construct which contains measured variables D1 to D5; and n5 represents the tangibles construct which contains measured variables E1 to E4. Measured variables are shown with a box by labels corresponding to those shown in the questionnaire. Each measured variable has an error term (d) associated with it. These error terms are not shown in the figure for simplicity. Each arrow from a construct to a measured variable also has a corresponding loading (k). Two-headed connectors indicate correlation coefficient between the latent constructs i and j (Ui,j). In this model, 22 measured variables are used, thus a 22  22 covariance matrix is then constructed. Consequently, the total degree of freedom available is 253 [(22  23)/2 = 253]. A total of 54 parameters will be estimated, of which 17 are factor loadings (because one loading estimate is set to be 1 for each construct), 22 are error variance terms and 15 [(5  6)/2 = 15] are construct covariance terms. Because 253 is greater than 54, the model is identified with respect to the order condition (Hair et al., 2006).

E2

E3

E1

E4

λE1,5 λE 2,5 λE 3,5 λE 4,5 Tangibles A1

A2

A3

A4

λA1,1 λA 2,1 λA3,1 λA 4,1

( ξ 5)

Φ1,5 Φ1,4

Reliability (ξ1)

Φ1,3

λA5,1

Φ 3,5 Φ 2,4

Φ 2,3

(ξ 2 )

λB1,2 B1

λB 4,2 λB 2,2 λB 3,2 B2

λD 5,4

Φ 3,4

Responsiveness

B3

C1

λC 2,3 λC 3,3λC 4,3 C2

Fig. 3. SERVQUAL measurement model (Hair et al., 2006).

D2

D3

D4

D5

Assurance (ξ 3 )

λC1,3 B4

D1

λD 2,4 λD 3,4 λD 4,4

Empathy (ξ 4 )

Φ1,2

A5

λD1,4

Φ 4,5

C3

C4

K.-K. Chen et al. / Transportation Research Part E 45 (2009) 222–237

227

Table 2 Definitions of measured variables and symbols Measured variables

Definition

A1 A2 A3 A4 A5 B1 B2 B3 B4 C1 C2 C3 C4 D1 D2 D3 D4 D5 E1 E2 E3 E4

When XYZ company promises to do something by a certain time, it does so When you have a problem, XYZ company shows a sincere interest in solving it XYZ company performs the service right the first time XYZ company provides its services at the time it promises to do so XYZ company insists on error-free records XYZ company keeps customers informed about when services will be performed Employees in XYZ company provide you prompt service Employees in XYZ company are always willing to help you Employees in XYZ company are never too busy to respond to your request The behavior of employees in XYZ company instills confidence in you You feel safe in doing transactions with in XYZ company Employees in XYZ company are always courteous to you Employees in XYZ company have the knowledge to answer your questions XYZ company gives you individual attention Employees in XYZ company give you personal attention XYZ company has your best interests at heart Employees in XYZ company understand your specific needs XYZ company has operating hours that are convenient to all its customers XYZ company has modern-looking equipment. (vessels or containers) XYZ company’s physical are visually appealing. (offices, berths and cranes) XYZ company’s employees appear neat Materials associated with the service (such as pamphlets or statements) are visually appealing at XYZ company

Symbols ni kxi;j Ui,j dxi

The The The The

latent construct i, i = 1, 2, 3, 4, 5 factor loading of the latent construct j on the measured variable xi correlation coefficient of the latent constructs i and j error term of the measured variable xi

Eq. (1) is the mathematical expression for the measurement model: xij ¼ kxi ;j nj þ dxi ;

j ¼ 1; 2; . . . ; 5; i ¼ 1; 2; . . . ; nj ;

ð1Þ

where nj, j = 1, 2, . . . , 5; represent the five latent constructs of the SERVQUAL scale, dxi is the error term, which is the extent to which the latent construct does not explain the measured variable. 4.3. Testing procedure Fig. 4 shows the three steps involved in testing the two hypothesized gaps in this paper. The three steps are discussed below. 4.3.1. Step 1: determine the method for estimating the measurement model After the identification condition has been verified, CFA begins to choose the model estimation method according to whether the data are multivariate normally distributed. If the data are multivariate normally distributed, the method of maximum likelihood estimation will be used; otherwise the asymptotic distribution-free method or weighted least squares method will be adopted (Browne, 1984). The deviation of kurtosis from normality has a severe impact on the precision of estimation (Bollen, 1989; West et al., 1995), and the Mardia’s multivariate kurtosis (Mardia and Foster, 1983) will be used as the test statistic for normality. Its critical value should absolutely exceed 25. 4.3.2. Step 2: model validated by CFA (1) Data quality assessment. The quality of data is an important factor influencing the precision of empirical results. Some problems are usually not found in the beginning of data analysis. An offending estimate suggests some serious problem with the data (Hair et al., 2006). Standardized loadings exceeding or too close to 1.0, too larger standard errors or negative error variances are diagnostic cues for examining the model quality. (2) Unidimensionality. A highly mandatory condition for construct validity and reliability checking is the unidimensionality of the measure (Anderson and Gerbing, 1991). It refers to the existence of a single construct underlying a set of measures. The accepted criterion is that the paired correlation coefficients of constructs are not equal to 1. (3) Internal consistency. Once unidimensionality of a scale is established, its statistical reliability should be assessed before it is subjected to any further validation analysis (Ahire et al., 1996). Although there has been considerable debate on which reliability estimate is the best, coefficient alpha remains a commonly applied estimate (Hair et al., 2006). An alpha value of 0.6 or above is considered to be the criteria for demonstrating internal consistency of new scales (Nunnally, 1988).

228

K.-K. Chen et al. / Transportation Research Part E 45 (2009) 222–237

Fig. 4. Testing procedure.

(4) Face validity. After internal consistency is demonstrated, the construct validity of a measurement model, which includes face, content, convergent, discriminant and nomological validity shall be tested.Face validity is the mere appearance that a measure is valid (Kaplan and Sacuzzo, 1993). This must be established prior to any theoretical test when using CFA. As the service quality constructs are identified from the literature, the selection of SERVQUAL is justified, thereby ensuring the face validity of the instrument. (5) Content validity. Content validity is the degree to which the instrument provides an adequate representation of the conceptual domain that it is designed to cover. Like face validity, content validity is a type of validity for which the evidence is subjective and logical rather than statistical (Kaplan and Sacuzzo, 1993). If the items representing the various constructs of an instrument are substantiated by expert judges, content validity can be ensured (Bohrnstedt, 1983). In this study, experts (both academia and practitioners) in the shipping industry have shown little negative opinion against the SERVQUAL instrument. Therefore, the SERVQUAL instrument with only slight modifications is used in this study. (6) Convergent validity. The items that are indicators of a specific construct should converge or share a high proportion of variance in common, known as convergent validity (Hair et al., 2006; Kaplan and Sacuzzo, 1993). When there is high correlation between a measure and other measures that are believed to measure the same construct, convergent evidence for validity is obtained. High loadings on a factor would indicate that they converge on some common point. Another criterion is the average percentage of variance extracted (VE) among a set of construct items. A VE of 0.5 or higher is a good rule of thumb suggesting adequate convergence (Hair et al., 2006). (7) Discriminant validity. Discriminant validity of a measure is the extent to which a construct is distinct from other constructs. A test of discriminant validity is to compare VE percentages for any two constructs with the square of the correlation estimate between these two constructs (Fornell and Larcker, 1981). The VE estimates should be greater than the squared correlation estimate. (8) Nomological validity. The basic idea of nomological validity is to check the performance of the measure against some criteria. Scholars (Peter and Churchill, 1986; Hogan and Nicholson, 1988) suggest that empirical evidence of relationships between measured structures of another conceptually related construct is accepted as evidence of construct validity. In the present study, nomological validity is established by correlating the scale scores with two criteria, namely total service satisfaction and repurchase propensity. (9) Goodness-of-fit assessment. Measurement model will also be verified by the goodness-of-fit (GOF) indices. GOF tests including v2 test (significant p-value with number of measured variables exceeding 12 or insignificant p-value with number of measured variables below 12; Hair et al., 2006, Table 10-2), GFI and AGFI exceeding 0.9, RMSR and RMSEA below 0.1, and NFI and CFI exceeding 0.9 will be employed to justify the validity of applying the SERVQUAL instrument to the shipping industry.

229

K.-K. Chen et al. / Transportation Research Part E 45 (2009) 222–237

4.3.3. Step 3: determine the method for testing the hypothesized gaps If the measurement model did pass all validity tests involved in Step 2, then it is well fitted and the two hypothesized gaps would be tested using factor structure invariance. Otherwise, they would be tested using multivariate analysis of variance (MANOVA). 5. Results 5.1. Multivariate normality Initial analysis suggested that 21 observations should be deleted from further analyses because of their large kurtosis values. After that, Mardia’s multivariate kurtosis remains to be 23.56, meaning that the measured variables of sample size 419 are approximately multivariate normal (Browne, 1984). Therefore, the method of maximum likelihood is employed to estimate the measurement model. 5.2. Validity of model All standardized factor loadings exceed 0.5, and each indicator t-value exceeds 10.0 (p < 0.001). Cronbach’s alpha exceeds 0.85 for each scale (see Table 3). GOF indices are shown in Table 4. The v2 fit statistic is 484.98 with 201 degrees of freedom (p < 0.001). RMSEA is 0.0581, CFI is 0.9590, and GIF is 0.904. All support the overall measurement quality given a large sample and a large number of indicators (Gerbing and Anderson, 1992). Unfortunately, not all VE in each construct exceed the respective correlation estimate between them (not shown in the paper), which provides evidence of discriminant validity. The correlations between the factor scores for each construct are shown in the upper part of Table 4. Among the five constructs, only reliability construct has significant positive correlation with repurchase propensity, and two constructs, reliability and empathy, have significant positive correlations with service satisfaction. Evidence does not support the nomological validity either. After deleting all invalid items, a reduced model is established which can meet the requirement of the discriminant validity. The factor loadings of the reduced model are shown in Table 5. There are only two constructs, assurance and empathy, remaining. The Mardia’s multivariate kurtosis of this model is 9.1896. Loadings of all remaining items exceed 0.8, and each indicator t-value exceeds 20.0 (p < 0.0001). The Cronbach’s alphas of these constructs are 0.84 and 0.93, respectively. The variances extracted from these two constructs are 77% and 82%, respectively, which is greater than the correlation estimate

Table 3 Standardized measurement coefficients and t-values obtained by CFA Abbreviation of measured variable A1 A2 A3 A4 A5 B1 B2 B3 B4 C1 C2 C3 C4 D1 D2 D3 D4 D5 E1 E2 E3 E4 Variance extracted (%) Cronbach’s alpha a b

Construct Reliability

Responsiveness

Assurance

Empathy

Tangibles

0.8080 (NAb) 0.7681(28.74a) 0.8120 (26.76) 0.8224 (23.08) 0.8802 (26.63) 0.8179 0.8532 0.8815 0.7554

(NA) (23.14) (24.61) (18.82) 0.8369 0.8886 0.7854 0.7792

(NA) (24.69) (19.86) (19.61) 0.8915 0.9220 0.8999 0.7988 0.7656

67 0.8728

69 0.8749

t-Values shown in parentheses. All are significant (p < 0.001). t-Value is not available for the fixed factor loadings.

68 0.8735

74 0.9208

(NA) (26.30) (25.09) (20.37) (19.05) 0.7735 0.8918 0.8931 0.8419 72 0.8978

(NA) (24.50) (24.06) (22.09)

230

K.-K. Chen et al. / Transportation Research Part E 45 (2009) 222–237

Table 4 GOF indices of full and reduced models Reduced model

Nomological validity Reliability Responsiveness Assurance Empathy Tangibles Absolute indices v2 d.f. p-Value GFI AGFI RMSR RMSEA

Full model

Repurchase propensity

Service satisfaction

Repurchase propensity

Service satisfaction

NA NA 0.02689 0.08378 NA

NA NA 0.08893 0.57234** NA

0.19388** 0.02639 0.08917 0.09852 0.01074

0.5908** 0.13793 0.12876 0.39267** 0.08054

23.3662 5 0.0003*** 0.9792 0.9376 0.1119*** 0.0937

484.98 201 <0.0001 0.9040 0.8793*** 0.1821*** 0.0581

Incremental indices NFI CFI

0.9845 0.9877

0.9323 0.9590

Parsimony indices PGFI PNFI

0.4896 0.4923

0.7867 0.8112

*

Significant at 5%. Significant at 1%. Criterion is not satisfied.

**

***

Table 5 Summary of statistics of reduced model Construct

Assurance Empathy a b

Abbreviation of measured variable C1

C2

0.9074 (nab)

0.8452 (20.5598)a

D1

D2

D3

0.8957 (24.2083)

0.9364 (26.2656)

0.8888 (23.8844)

VE

Cronbach’s alpha

77% 82%

0.8410 0.9286

t-Values shown in parentheses. All are significant (p < 0.0001). t-Value is not available for the fixed factor loading.

between factors, being 69%. This model is the only one that can pass the criterion of discriminant validity among all possible choices. These characteristics are not enough to validate the reduced model because the nomological validity test fails. On the other hand, some of the GOF indices in the reduced model are better than those in the full model; for instance, GFI = 0.9792, AGFI = 0.9376, NFI = 0.9845, and CFI = 0.9877, are greater than the corresponding indices in the full mode. However, the other indices do not indicate that the reduced model is superior to the full model. In particular, two parsimony indices, PGFI and PNFI, are both decreased in the reduced model. Therefore, the evidence does not support validation of the reduced model. 5.3. MANOVA assumptions Since the validity of the SERVQUAL scale was not supported empirically, the existence of the two hypothesized SQ gaps – the SQ gap between types of business customers and the SQ gap among employee statuses had to be examined by MANOVA. The most critical assumptions related to MANOVA are the independence of observations, homoscedasticity across the sub-samples, and normality. The independence of the respondents was ensured by the random arrangement of the customer list in the sampling plan. The assumption of multinormality for the 22 dependent variables was examined by Mardia’s multivariate kurtosis in Section 5.1. The homogeneity of the variance–covariance matrices is tested by two steps. First, Levene’s test is employed to test the equality of the variances for all 22 measured variables. The statistical value thus obtained is insignificant at 10%. Second, all data are separated into four sub-samples, forwarder–manager, forwarder–subordinate, shipper–manager, and shipper–subordinate. The Box’s M statistic is used for testing the equality of the entire variance–covariance matrices among four sub-samples. The result shows an insignificant value (0.246), indicating no significant differences in the variance–covariance matrices among the four sub-samples on the 22 measured variables collectively (Hair et al., 2006). This supports the results of testing the equality of variance–covariance matrices among the sub-samples.

231

K.-K. Chen et al. / Transportation Research Part E 45 (2009) 222–237

5.4. Differences in means To test the existence of the hypothesized SQ gaps, we divided all data into four sub-samples and examine the equalities of the perceived SQ mean scores item by item among four groups of respondents, respectively. Since there exists the interaction effect of customer type (forwarder and shipper) and job status (manager and subordinate), two-way with interaction MANOVA is employed to examine the effects of customer type, job status and their interaction on 22 service items. Table 6 shows that the three effects of overall test according to the statistics of Wilks’ Lambda are all significant (p-values < 0.0001). The results obtained from other test statistics, such as Pillai’s Trace, Hotelling-Lawley Trace and Roy’s Greatest Root, have the same conclusions (not presented here). In other words, there is evidence supporting the existence of SQ gaps between forwarders and shippers and between managers and subordinates of the shipping industry in Taiwan. The perceived SQ mean scores and the t-values of each service item for the four sub-samples are shown in Table 7. Fig. 5 shows the profiles of the SQ mean scores for the four samples. From Table 7 and Fig. 5, we note the following. (1) All of the perceived SQ mean scores rated by the forwarders for the 22 service items are significantly lower than those rated by the shippers as seen in the last column of Table 7. This can also be clearly observed from Fig. 5. (2) The effects of job status on the perceived SQ are not consistent among the service items. In the forwarder group, all perceived SQ mean scores relating to reliability, responsiveness and assurance (A1 to C4) for the subordinates are lower than those for managers. However, opposite results are obtained for the constructs empathy and tangibles (D1 to E4). All perceived SQ mean scores of these two constructs for the subordinate groups are higher than those for the manager groups.The perceived SQ mean scores for the shipper group are different from those for the forwarder group. In the shipper group, the perceived SQ mean scores relating to reliability, responsiveness and assurance (A1 to C4) for the subordinates are higher than those for the managers except for items B4, C3 and C4. However, the perceived SQ mean scores relating to empathy and tangibles (D1 to E4) for the subordinates are lower than those for the managers except for items D4 and E3. This is the source of the interactive effect of job status and customer type.

Table 6 MANOVA statistics Source

D.F.

Wilks’ Lambda

F value

Pr > F

Forwarder–shipper Manager–subordinate Interaction

1 1 1

0.2297 0.7439 0.6154

59.29 6.15 11.16

<0.0001 <0.0001 <0.0001

Table 7 Perceived SQ mean scores and t-values of 22 service items Abbreviation of item

Forwarder–subordinate

A1 A2 A3 A4 A5 B1 B2 B3 B4 C1 C2 C3 C4 D1 D2 D3 D4 D5 E1 E2 E3 E4

1.869 1.972 2.084 1.981 2.149 2.289 2.280 2.130 2.177 2.046 2.140 1.850 2.074 2.345 2.448 2.504 2.485 2.345 2.074 2.242 2.130 2.130

(2.005) (1.861) (2.082) (2.126) (2.195) (2.741) (2.146) (1.895) (2.085) (1.871) (1.857) (1.578) (2.367) (2.281) (2.315) (2.570) (2.238) (2.301) (1.988) (2.219) (1.789) (2.018)

Forwarder–manager

Shipper–subordinate

1.861 (2.939) 1.870 (3.219) 1.768 (2.720) 1.898 (3.195) 1.888 (3.141) 2.027 (3.658) 1.888 (3.528) 1.712 (3.276) 1.898 (2.782) 1.870 (3.936) 1.722 (3.273) 1.518 (2.658) 1.731 (3.205) 3.574 (3.165) 3.592 (3.310) 3.583 (3.176) 3.000 (3.201) 2.666 (3.985) 2.537 (3.921) 2.574 (4.802) 2.388 (3.826) 2.638 (4.761)

0.519 0.905 0.018 0.141 0.094 0.367 0.141 0.075 0.896 0.207 0.141 0.613 0.349 0.905 0.716 0.613 0.933 1.141 1.084 0.877 0.481 0.943

F, forwarder; I, shipper; S, subordinate; M, manager; In, interaction. The effects are significant at 10%.

*

(0.315) (0.668) (0.016) (0.136) (0.086) (0.317) (0.130) (0.077) (0.726) (0.232) (0.167) (0.590) (0.343) (0.173) (0.701) (0.575) (1.028) (1.038) (1.039) (1.009) (0.668) (1.826)

Shipper–manager 0.546 0.979 0.412 0.670 0.422 0.422 0.494 0.484 0.134 0.649 0.515 0.319 0.195 0.639 0.938 0.505 0.773 0.659 0.885 0.536 0.515 0.536

(0.364) (0.789) (0.335) (0.634) (0.333) (0.310) (0.445) (0.460) (0.112) (0.509) (0.423) (0.392) (0.168) (0.472) (0.789) (0.396) (0.752) (0.478) (0.700) (0.500) (0.499) (0.402)

Effects* Fa < Ib F M; F < I; ; In F M; F < I; S > M; F < I; S > M; F < I; S > M; F < I; ; In F < I; ; In F < I; ; In F < I; ; In F < I; ; In

In

In

In In In In In In

232

K.-K. Chen et al. / Transportation Research Part E 45 (2009) 222–237

0.5 0 A1 A2 A3 A4 A5 B1 B2 B3 B4 C1 C2 C3 C4 D1 D2 D3 D4 D5 E1 E2 E3 E4 -0.5 Forwarder subordinates

-1

Forwarder managers

-1.5

Shipper subordinates

-2

Shipper managers

-2.5 -3 -3.5 -4 Fig. 5. Profiles of SQ mean scores for four samples.

(3) The greatest difference in mean scores of the perceived SQ is seen in the construct empathy which includes service items D1 (XYZ company gives you individual attention), D2 (Employees in XYZ company give you personal attention), D3 (XYZ company has your best interests at heart) and D4 (Employees in XYZ company understand your specific needs) between the forwarder managers and their subordinates (see Fig. 5). The corresponding factor loadings of D1, D2 and D3 in the reduced measurement model (Table 5) are 0.8957, 0.9364 and 0.8888, respectively; and the correlation coefficient of the construct reflecting to these items and the overall satisfaction score are 0.57234 in the reduced measurement model and 0.39267 in the full measurement model (Table 4), respectively. These two significant correlation coefficients validated the nomological criterion for the empathy construct in these two measurement models. This result confirms our argument on different SQ perceived due to the employee statuses. Comparing to the forwarders’ subordinates, the responsibility of pursuing profitability should make the forwarder managers to rate their perceived SQ low. It is clearly that for closing the low SQ gap the liner should give more empathy to the forwarder managers. The reason why the perceived SQ mean scores of the shipper group is all lower than those of the forwarder group. In other words, the impact of the perceived SQ on the shippers’ profit is smaller than that the counterpart of the forwarders. In summary, the existence of the SQ gap between forwarders and shippers and the perceived SQ mean scores of the 22 service items for the forwarders being lower than those for the shippers are supported by statistical evidence. While the existence of the SQ gap among employee statuses of business customers is also supported by the data, the magnitudes of the perceived SQ mean scores of the service items for the managers and subordinates are not consistent. 6. Discussion This paper examines two issues related to the field of SQ. One is the validity of applying the SERVQUAL scale to the shipping industry of Taiwan is tested. The other is the SQ gap between types of business customers (forwarders and shippers) and the SQ gap among employee statuses (managers and subordinates) are examined empirically by a sample of customers of an international line company of Taiwan. Our results strongly support the existence of the two SQ gaps and that the magnitudes of all perceived SQ mean scores rated by the forwarders for the 22 service items are lower than those rated by the shippers. However, the validity of applying SERVQUAL to the shipping industry of Taiwan has been rejected. This study offers additional evidence to the controversy on the validity of SERVQUAL. 6.1. Measurement model Since the validity of applying SERVQUAL to the shipping industry of Taiwan is rejected by the evidence, we re-examine the possible constructs which could be extracted from the 22 service items of the SERVQUAL scale using the exploratory factor analysis (EFA) method. The forwarder model and the shipper model are examined separately. In the forwarder model, three factors with eigenvalues exceeding 1.0 are extracted from these 22 items. Only the first factor is meaningful because all factor loadings on the factor are positive and significant (>0.5) as shown in Table 8. All factor loadings on the second and third factors are insignificant. This result is similar to the finding of Cronin and Taylor (1992).

233

K.-K. Chen et al. / Transportation Research Part E 45 (2009) 222–237 Table 8 Eigenvalues and factor loadings for samples of forwarders and shippersa Forwarder model

Eigenvalues Cumulative proportion Factor loading Reliability A1 A2 A3 A4 A5 Responsiveness B1 B2 B3 B4 Assurance C1 C2 C3 C4 Empathy D1 D2 D3 D4 D5 Tangibles E1 E2 E3 E4 a

Shipper model

Factor 1

Factor 1

Factor 2

Factor 3

Factor 4

Factor 5

11.35 0.52

2.712 0.1233

2.375 0.2313

1.989 0.3217

1.714 0.3997

1.598 0.4723

0.68085 0.80051 0.72747 0.79590 0.74801

0.53428 0.59658 0.51868 0.57912

0.70210 0.77364 0.80692 0.74532

0.57973 0.51466

0.77653 0.79411 0.73307 0.60415 0.69161 0.83705 0.54765 0.61072 0.62974 0.67428 0.80448 0.79141 0.76912

0.58713 0.65814 0.56416 0.58553 0.68806 0.65073

0.54966 0.62468 0.55585 0.60136

Insignificant factor loadings are not listed.

Cronin and Taylor (1992) explored SQ using data gathered from two firms in each of four industries and performed a factor analysis of the SERVQUAL scale. All of the items loaded on a single factor with the exception of item ‘XYZ company has modern-looking equipment’. The result is much like the finding presented in Table 8. They suggested that the five-component structure proposed by Parasuraman et al. (1988,1991) for their SERVQUAL scale was not confirmed in any of the research sample. In fact, the number of constructs being fewer than 5 can be predicted from the invalidity of the discriminant test. The discriminant validity of a measurement model is examined by comparing the VE percentages for any two constructs with the square of the correlation estimate between these two constructs in the model. The correlation coefficients of all items or most items will be large if the customers have strong perceptions to the services delivered by the service providers. For example, the average partial correlation coefficient of the 22 measured variables for the forwarder sample is 0.6524. These large correlation coefficients between the measured variables result from the fact that the SQ perception of the forwarders is strong. Hence, the correlation coefficients of many constructs are greater than their VEs, the discriminant validity fails and the measurement model is invalid. As a result, only two constructs of SERVQUAL have passed the discriminant validity using the data of this paper. That evidence is support only for the two-construct model is the same as the result of Durvasula et al. (1999). These authors also concluded a two-factor model by the testing of discriminant validity for ocean freight shipping companies in Singapore. The other source of invalidity of SERVQUAL as applied to the shipping industry of Taiwan results from insufficient nomological criterion. This means that the predictive power for the customer’s behavior of the SERVQUAL model is insufficient. It is believed that some of critical service attributes are not included in the SERVQUAL. For example, the forwarders occasionally require lines to offer extra container spaces to deal with their emergent business while they usually contracted a fixed container space with lines. Thus, a liner with flexible loading capacity will be preferred by the forwarders. Table 8 also shows the measurement model estimated by the exploratory factor analysis for the shippers. There are five factors with eigenvalues exceeding 1.0. However, only 16 factor loadings on the five factors are significant (>0.5). The cumulative explained total variance of these five factors is only 47.23%. The reliability of this model is even poorer than that of the forwarder model. This also implies the need to develop a scale for measuring the perceived SQ of the forwarders and the shippers in the shipping industry of Taiwan.

234

K.-K. Chen et al. / Transportation Research Part E 45 (2009) 222–237

Another challenge of SERVQUAL posed by Cronin and Taylor (1992) is the competition from the SERVPERF model. A study of Durvasula et al. (1999) also showed support for using SERVQUAL performance scores instead of gap scores when assessing service quality. When using ‘‘performance-only (perception)” SQ measurements defined by the SERVPERF model, the factor loadings on the five constructs of the three measurement models – manager model (combining data sets of the forwarder– managers and the shipper–managers into a sample), subordinate model (combining data sets of the forwarder–subordinates and the shipper–subordinates into a sample), and pooling model (pooling all data together) are shown in Table 9. The reliability coefficients of all three models are not large enough (0.4729, 0.0648 and 0.2210; respectively). Many offending estimates are also found in this table. For example, the factor loadings of E3 in both the manager model and the pooling model are greater than 1.0, and those of A2 and E2 in the manager model are negative. In contrast to the conclusions of Brady et al. (2002), Cronin and Taylor (1992) and Durvasula et al. (1999) the quality of estimation obtained by the SERVPERF model is not superior to that of the SERVQUAL model. These findings will offer additional evidence (or material) to clarify the debates on the subject of SQ measurement. 6.2. SQ gaps and managerial implications The conclusion of lower SQ perception (perceived SQ mean scores < 0) of the forwarders than that of the shippers suggests that the line should review the strategy of resource allocation between forwarders and shippers. At present, many forwarders attract shippers by offering extra services such as offering shippers with a long period of account receivable or with loosing credit terms. A large part of volume of container shipment of lines in the world market comes from the contracts made with the forwarders. For instance, more than 80% volume of container shipment delivered by the company studied was employed by the forwarders. It is believed that the improvement of the forwarders’ SQ perception can enhance liner–forwarder partnering relationship, and hence, the market share of container shipping volume of the lines will rise. The managers of the line should examine the gap deliberately when they make marketing decisions or propose a SQ improvement project for their customers; otherwise, the effectiveness of various marketing efforts will be undermined. As mentioned above, for closing the forwarders’ SQ gap the lines should give more empathy to the forwarder managers.

Table 9 Factor loadings and Cronbach’s alpha of three SERVPERF models

Factor loading Reliability

Responsiveness

Assurance

Empathy

Tangibles

A1 A2 A3 A4 A5 B1 B2 B3 B4 C1 C2 C3 C4 D1 D2 D3 D4 D5 E1 E2 E3 E4

Manager model

Subordinate model

Pooling model

1.0000 (NAa)b 0.5253 (7.4822)* 0.6687 (9.7284) 0.0099 (0.1367) 0.7647 (11.3033) 1.0000 (NA) 0.3415 (4.8938) 0.0348 (0.4935) 0.7600 (11.3240) 1.0000 (NA) 0.1206 (1.6900) 0.1158 (2.1768) 0.1028 (1.4430) 1.0000 (NA) 0.7765 (8.1821) 0.4550 (5.5846) 0.5273 (6.3284) 0.0415 (0.5117) 1.0000 (NA) 0.3129 (4.7204)* 1.2328 (26.2461)** 0.2981 (4.4892)

1.0000 (NA) 0.0658 (0.9761) 0.1344 (1.9725)* 0.0166 (0.2462) 0.0297 (0.4402) 1.0000 (NA) 0.0704 (1.0402) 0.0163 (0.2413) 0.0103 (0.1532) 1.0000 (NA) 0.0081 (0.1201) 0.1211 (1.7755) 0.0161 (0.2374) 1.0000 (NA) 0.0002 (0.2267) 0.0018 (2.3255) 0.0010 (1.3174) 0.0019 (2.4719) 1.0000 (NA) 0.1936 (2.8086) 0.0676 (0.9961) 0.0549 (0.8099)

1.0000 (NA) 0.0255 (0.3751) 0.1194 (1.7440) 0.0242 (0.3564) 0.1159 (1.6942) 1.0000 (NA) 0.0344 (0.5070) 0.1217 (1.7792) 0.0323 (0.4762) 1.0000 (NA) 0.0843 (1.2322) 0.1072 (1.5624) 0.0940 (1.3724) 1.0000 (NA) 0.2239 (2.6700) 0.3593 (4.2360) 0.7560 (6.9268) 0.2268 (2.6647) 1.0000 (NA) 0.0536 (0.8044) 1.1404 (17.8972)** 0.8014 (12.688)

ac Reliability Responsibility Assurance Empathy Tangibles Total model a

0.2414 0.3355 0.1268 0.3368 0.3022 0.4729

NA, not available. Value in parentheses is t-value. c Cronbach’s alpha. * Negative value. ** Value > 1.0.

b

0.0514 0.0896 0.0890 0.1050 0.1012 0.0648

0.1172 0.2070 0.0950 0.2303 0.5661 0.2210

K.-K. Chen et al. / Transportation Research Part E 45 (2009) 222–237

235

This strategy will raise the marketing expenses of the lines. Instead of improving the partnering relationship with the forwarders, the higher perceived SQ revealed in the shippers makes the market concentration strategy possible. Concentrating on the shipper market will cut down the cost of the lines. Building up a supply chain with carriage firms to provide door-to-door service and serving shippers directly are other strategies for the lines. To lines the effect of this strategy acts as a forwarder by themselves. This might be a good alternative for a line whose total transaction cost (Williamson, 1985) could be reduced by the adoption of the strategy. The finding of the SQ gap between the managers and their subordinates extends the gaps model of Zeithaml et al. (1990) to the service reception side when the customer is a business whose service flows involve many steps and many employees with different statuses. Employees with different statuses have different personal needs and different responsibilities; hence, they will see the job with different views. Thus, they will emphasize on different job performances. Service providers should note the differences in perceived SQ among customers’ employees, particularly, the opinion leaders who will influence the behavioral intention of the business customer. Raising the perceived SQ of these opinion leaders is a tactic to improve the service image or market reputation of the service providers. 6.3. Future research There are many approaches mentioned in the extant literature for developing a SQ scale (Carr, 2002; Lehtinen and Lehtinen, 1985; Sureshchandar et al., 2002). Among them, the process perspective would be an adequate approach for the shipping industry (Chen and Chang, 2005) because under the review of practitioners or exporters, this approach will find every possible item out in each step involved in the whole process; hence, the possibility of missing important service items will be low. The core service prospect is another promising alternative (Sureshchandar et al., 2002). Lu (2003) and Brooks (1983) had mentioned many service attributes provided by the carriers. These service attributes belong to the category of the core service prospect. However, the service attributes mentioned in Lu (2003) included some price-related items that are not considered to be the measured variables in SQ studies (Caruana et al., 2000; Cronin et al., 2000; Lin et al., 2005; Parasuraman et al., 1988, 1991, 1994). The wave of the service marketing research has been shifted from the study of SQ solitary to the investigation of the relationships among SQ, service satisfaction, service value and the behavioral intention since 2000. The lack of a scale for measuring SQ will be an obstacle for continuing service marketing study in the shipping industry. More efforts should be put in the development of a scale for measuring the perceived SQ appropriate for the shipping industry of Taiwan in the near future. 7. Conclusions The validity of applying SERVQUAL to the shipping industry of Taiwan and the SQ gaps between business customer types and between employee statuses of business customers are examined using the data of perceived SQ of the forwarders and the shippers in the shipping industry of Taiwan. The implications for the management and future research are also discussed in this paper. Our results do not support the validity of applying SERVQUAL to the shipping industry of Taiwan due to the failures of discriminant and nomological validity. According to the customers’ perceptions (performance-only), the SERVPERF measurement model is not validated either. This result highlights the necessity of developing an appropriate SQ scale for measuring the perceived SQ of the forwarders and the shippers in the shipping industry of Taiwan. There is strong evidence supporting the existence of SQ between types of business customers and between the employee statuses of business customers. These findings extend the gaps model of Zeithaml et al. (1990) to the service reception side for the business customers. This extension of the SQ gaps model will provide a clear profile for the managers to make decisions on improving the SQ of their business customers. This finding also complements the scant SQ literature in the shipping industry. Durvasula et al. (1999) concluded that the psychometric properties of the SERVQUAL scale are not superior to those of the SERVPERF scale in the case of ocean freight shipping service. Lu (2003) studied the importance of the impact of carriers’ service attributes on the partnering relationships between the carriers and shippers. Neither the literature mentioned the complicated relationship between the shippers, forwarders and carriers, nor mentioned the need to considering the differentiation of managerial strategies for different customers. This paper discovers the existence of the SQ gaps between the forwarders and shippers and also between the managers and their subordinates of the forwarders and the shippers in the shipping industry of Taiwan. This will urge the managers in the industry to re-consider what their optimal resource allocation decisions are. The main contributions of the paper to the literature are the findings of the SQ gaps between types of business customers and among employee statuses of business customers. These findings extend the gaps model developed by Zeithaml et al. (1990), and also suggest a direction for further research on SQ for the shipping industry of Taiwan. Acknowledgements The authors are grateful to Prof. Talley (Editor-in-Chief) and an anonymous referee for their useful and valuable comments that helped to improve this manuscript significantly.

236

K.-K. Chen et al. / Transportation Research Part E 45 (2009) 222–237

Appendix The. SERVQUAL scale Directions: This survey deals with your feelings about the following service items delivered by XYZ Company. For each statement, please show the extent to which you believe XYZ has the feature (perception) and the extent to which you expect XYZ offers the feature (expectation) described by the statement. Do this by picking one of the seven numbers in each column next to each statement. If you strongly agree that XYZ should possess such a feature, circle the number 7. If you strongly disagree that XYZ should possess such a feature, circle 1. If your feelings are not strong or extreme, circle one of the numbers in the middle. Perception When XYZ promises to do something by a certain time, it does so. When you have a problem, XYZ shows a sincere interest in solving it. XYZ performs the service right the first time. XYZ provides its services at the time it promises to do so. XYZ insists on error-free records XYZ keeps customers informed about when services will be performed. Employees in XYZ provide you prompt service. Employees in XYZ are always willing to help you. Employees in XYZ are never too busy to respond to your request. The behavior of employees in XYZ instills confidence in you. You feel safe when doing transactions with XYZ. Employees in XYZ are always courteous to you. Employees in XYZ have the knowledge to answer your questions. XYZ gives you individual attention. Employees in XYZ give you personal attention. XYZ has your best interests at heart. Employees in XYZ understand your specific needs. XYZ has operating hours that are convenient to all its customers. XYZ has modern-looking equipment (ships or container). Facilities of XYZ are visually appealing (offices, berths and cranes). Employees in XYZ appear neat. Materials associated with the service (such as pamphlets or statements) at XYZ are visually appealing

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2

3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3

Expectation 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4

5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5

6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6

7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2

3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3

4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4

5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5

6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6

7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7

Direction: The following two statements describe your overall satisfaction in the SQ provided by XYZ and your long-term relationship with XYZ. Once again, circle the appropriate number next to the statement. 1. The overall SQ delivered by XYZ is: 1. 2. 3. 4. 5. 6. 7.

extremely poor, every poor, poor, so far so good, good, fairly excellent, excellent.

2. In the past 3 years, the container volume of your company delivered by XYZ is: 1. decreased, 2. the same as before, 3. increased.

References Ahire, S.L., Golhar, D.Y., Waller, M.A., 1996. Development and validation of TQM implementation constructs. Decision Sciences 27, 23–56. Anderson, J.C., Gerbing, D.W., 1991. Predicting the performance of measures in a confirmatory factor analysis with a pretest assessment of their substantive validities. Journal of Applied Psychology 76 (5), 732–740. Asubonteng, P., McCleary, K.J., Swan, J.E., 1996. SERVQUAL revisited: a critical review of service quality. The Journal of Services Marketing 10 (6), 62–81. Badri, M.A., Abdulla, M., Al-Madani, A., 2005. Information technology center service quality: assessment and application of SERVQUAL. The International Journal of Quality and Reliability Management 22 (8), 819–848. Bitner, M.J., Booms, B.H., Tetreault, M.S., 1990. The service encounter: diagnosing favorable and unfavorable incidents. Journal of Marketing 54 (1), 71–84.

K.-K. Chen et al. / Transportation Research Part E 45 (2009) 222–237

237

Bohrnstedt, G., 1983. Measurement. In: Rossi, P., Wright, J., Anderson, A. (Eds.), A Hand Book of Survey Research. Academy Press, San Diego, CA. Bollen, K.A., 1989. Structural Equations with Latent Variables. Wiley, New York. Brady, M.K., Cronin, J.J., Brand, R.R., 2002. Performance-only measurement of service quality: a replication and extension. Journal of Business Research 55 (1), 17–31. Brooks, M.R., 1983. Determinants of shipper’s choice of container carrier: a study of eastern Canadian exporters. Ph.D. Dissertation, Department of Maritime Studies and International Transport, University of Wales, College of Cardiff, UK. Browne, M.W., 1984. Asymptotically distribution-free methods for the analysis of covariance structure. British Journal of Mathematics and Statistical Psychology 37, 62–83. Buttle, F., 1996. SERVQUAL: review, critique, research agenda. European Journal of Marketing 30 (1), 8–32. Carr, C.L., 2002. A psychometric evaluation of the expectations, perceptions, and difference scores generated by the IS-adapted SERVQUAL instrument. Decision Sciences 33 (2), 281–296. Caruana, A., Money, A.H., Berthod, P.R., 2000. Service quality and satisfaction – the moderating role of value. European Journal of Marketing 34 (12), 1338– 1353. Chen, F.Y., Chang, Y.H., 2005. Examining airline service quality from a process perspective. Journal of Air Transport Management 11, 79–87. Cronin Jr., J.J., Taylor, S.A., 1992. Measuring service quality: a reexamination and extension. Journal of Marketing 56 (July), 55–68. Cronin Jr., J.J., Brady, M.K., Hult, G.T.M., 2000. Assessing the effects of quality, value, and customer satisfaction on customer behavioral intentions in service environments. Journal of Retailing 76 (2), 193–218. Dabholkar, P.A., Shepherd, C.D., Thorpe, D.I., 2000. A comprehensive framework for service quality: an investigation of critical conceptual and measurement issues through a longitudinal study. Journal of Retailing 76 (2), 139–173. Davis, B.R., Mentzer, J.T., 2006. Logistics service-driven loyalty: an exploratory study. Journal of Business Logistics 27 (2), 53–74. Durvasula, S., Lysonski, S., Mehta, S.C., 1999. Testing the SERVQUAL scale in the business-to-business sector: the case of ocean freight shipping service. The Journal of Services Marketing 13 (2), 132–150. Fornell, C., Larcker, D.F., 1981. Evaluating structural equations models with unobservable variables and measurement error. Journal of Market Research 18, 39–50. Gerbing, D.W., Anderson, J.C., 1992. Monte Carlo evaluations of goodness-of-fit indices for structural equations models. Sociological Methods and Research 21 (November), 132–160. Gilmore, J.H., Pine II, B.J., 2002. The Experience is the Marking. Report from Strategic Horizons LLP. Gounaris, S., 2005. An alternative measure for assessing perceived quality of software house services. The Service Industries Journal 25 (6), 803–823. Hair Jr., J.F., Black, W.C., Babin, B.J., Anderson, R.E., Tatham, R.L., 2006. Multivariate Data Analysis, sixth ed. Prentice Hall, New Jersey. Hogan, R., Nicholson, R.A., 1988. The meaning of personality test scores. American Psychologist 43 (August), 621–626. Kaplan, R.M., Sacuzzo, D.P., 1993. Psychological Testing: Principles, Applications and Issues, third ed. Brooks Cole, Pacific Grove, CA. Johnston, R., Clark, G., 2001. Service Operations Management. Prentice-Hall, London. Lehtinen, U., Lehtinen, J.R., 1985. Service quality: a study of quality dimensions. Working Paper, Service Management Institute, Helsinki. Lin, C.H., Sher, P.J., Shih, H.Y., 2005. Past progress and future directions in conceptualizing customer-perceived value. International Journal of Service Industry Management 16 (4), 318–336. Lu, C.S., 2003. The impact of carriers service attributes on the shipper–carrier partnering relationships: a shipper’s perceptive. Transportation Research Part E: Logistics and Transportation Review 39 (5), 399–415. Mardia, K.V., Foster, K., 1983. Omnibus tests of multinormality based on skewness and kurtosis. Communication in Statistics 12, 207–222. Mohsin, A., 2005. Service quality perceptions: an assessment of restaurant and café visitors in Hamilton, New Zealand. The Business Review 3 (2), 51–57. Nadiri, H., Hussain, K., 2005. Diagnosing the zone of tolerance for hotel services. Managing Service Quality 15 (3), 259–277. Nunnally, J.C., 1988. Psychometric Theory. McGraw-Hill Book Company, Englewood-Cliffs, NJ. Parasuraman, A., Berry, L., Zeithaml, V., 1991. Refinement and reassessment of the SERVQUAL scale. Journal of Retailing 67 (4), 420–450. Parasuraman, A., Zeithaml, V.A., Berry, L.L., 1985. A conceptual model of service quality and its implications for future research. Journal of Marketing 49 (4), 41–50. Parasuraman, A., Zeithaml, V.A., Berry, L.L., 1988. SERVQUAL: a multiple item scale for measuring consumer perception of service quality. Journal of Retailing 64 (1), 12–37. Parasuraman, A., Zeithaml, V.A., Berry, L.L., 1994. Reassessment of expectations as a comparison standard in measuring service quality: implications for future research. Journal of Marketing 58 (January), 111–124. Peiro, J.M., Vicente, M.T., Ramos, J., 2005. Employees’ overestimation of functional and relational service quality: a gap analysis. The Service Industries Journal 25, 773–788. Peter, J.P., Churchill Jr., G.A., 1986. Relationships among research design choices and psychometric properties of rating scales: a meta-analysis. Journal of Marketing Research 23 (February), 1–10. Robertson, T.S., 1971. Innovative Behavior and Communication. Holt, Rinehart & Winston, New York. Rohini, R., Mahadevappa, B., 2006. Service quality in Bangalore hospitals – an empirical study. Journal of Services Research 6 (1), 59–68. Seth, N., Deshmukh, S.G., Vrat, P., 2006. A conceptual model for quality of service in the supply chain. International Journal of Physical Distribution and Logistics Management: 3PL, 4PL and Reverse Logistics – Part 1 36 (7), 547–575. Sureshchandar, G.S., Rajendran, Chandrasekharan, Anantharaman, R.N., 2002. Determinants of customer-perceived service quality: a confirmatory factor analysis approach. The Journal of Services Marketing 16 (1), 9–34. Teas, K.R., 1993. Expectations, performance evaluation, and consumers’ perceptions of quality. Journal of Marketing 57 (October), 18–34. West, S.G., Finch, J.F., Curran, P.J., 1995. Structural equation models with non-normal variables: problems and remedies. In: Hoyle, R.H. (Ed.), Structural Equation Modeling: Concepts, Issues and Applications. Sage, Thousand Oaks, CA. Williamson, O.E., 1985. The Economic Institutions of Capitalism. Free Press, New York. Zeithaml, V.A., Bitner, M.J., Gremler, D.D., 2006. Services Marketing: Integrating Customer Focus Across the Firm, fourth ed. McGraw-Hill, New York. Zeithaml, V.A., Parasuraman, A., Berry, L., 1990. Delivering Quality Service: Balancing Customer Perceptions and Expectation. The Free Press, New York.