上海海事大学：《Supply Chain Management》课程教学资源（文献资料）measure profermace

团购合买资源类别：文库，文档格式：PDF，文档页数：18，文件大小：245.87KB

Measures for evaluating supply chain performance in transport logistics Kee-hung Lai a,*, E.W.T. Ngai b , T.C.E. Cheng b a Department of Shipping and Transport Logistics, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong b Department of Management, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong Received 26 November 2001; received in revised form 13 March 2002; accepted 5 April 2002 Abstract This study aims to investigate the construct of, and develop a measurement instrument for, supply chain performance (SCP) in transport logistics. Based on the supply chain operations reference model and various established measures, a measurement model and a measurement instrument for SCP in transport logistics are developed. A 26-item SCP measurement instrument was constructed, reflecting service effectiveness for shippers, operations efficiency for transport logistics service providers, and service effectiveness for consignees. The empirical findings suggest that the measurement instrument is reliable and valid for evaluating SCP in transport logistics. 2002 Elsevier Science Ltd. All rights reserved. Keywords: Transport logistics; Supply chain; Construct; Performance measurement; Confirmatory factor analysis 1. Introduction The emergence of the global economy and intensified competition have led many firms to recognize the importance of managing their supply chains for fast product introduction and service innovations to the markets. For improved competitiveness, many firms have embraced supply chain management (SCM) to increase organizational effectiveness and achieve such organizational goals as improved customer value, better utilization of resources, and increased profitability (Lee, 2000). Transportation Research Part E 38 (2002) 439–456 www.elsevier.com/locate/tre * Corresponding author. Tel.: +852-2766-7920; fax: +852-2330-2704. E-mail address: stlmlai@polyu.edu.hk (K.H. Lai). 1366-5545/02/$ - see front matter 2002 Elsevier Science Ltd. All rights reserved. PII: S 1 3 6 6 - 5 5 4 5 ( 0 2 ) 0 0 0 1 9 - 4

440K.H.Lai et al. /Transportation Research Part E38(2002)439-456In his seminal work on competitiveness of firms, Porter (1985) identifies customer values andcoststo customersascritical elementstogain competitiveadvantagesforafirm.Themanagementof a supply chain encompasses these two elements, which together emphasize the importance ofgetting goods/services to customers at the right time, in the right place, under the right conditions,in the right quantities, and at the lowest possible costs. Porter (1985) emphasizes that, differen-tiation, one type of competitive advantage for a firm, is closely linked to the customer values ofthe product/service that can be delivered. Low cost, another type of competitive advantage, isreflected in the costs of theproduct/servicetothe customers.Christopher (1998)adds that a firmwould achievea competitiveadvantageby striving forexcellence inboth serviceand cost lead-ership.To this end, making proper performance measurement of a supply chain is necessary as itcultivates understanding between member firms in the supply chain for performance improvement(Dreyer,2000;FawcettandCooper,1998)Traditionally,thefocus of performance measurement has been on process operations withintheorganizational boundaries of a firm (ShortandVenkatraman,1992).In thecontextof SCM,performance measurement involves not only the internal processes, but also requires an under-standing of the performance expectation of other member firms in the supply chain, backwardfrom the suppliers and forward to the customers (Normann and Ramirez, 1993).Coordinationbetween thevariousparties in the supply chain iskeyto its effective implementation (Frohlich andWestbrook,2001).As SCM focuses on process management beyond organizational boundaries, there is a need tomeasure performance for the effective management of a supply chain.Harrington (1991,p.164)states that'If you cannot measure it, you cannot control it. If you cannot control it, you cannotmanage it. If you cannot manage it, you cannot improve it'. In fact, the lack of relevant per-formance measures has been recognized as one of the major problems in process management(Davenport et al., 1996) and the management of a supply chain (Dreyer, 2000). Because of thedifferent views on what should constitute supply chain performance (SCP), many firms havefound it dificult topracticeSCM (Beamon,1999).Amajor contributingfactor to this problem isthat, with multiple parties having different interests, it is difficult for firms to effectively evaluatethe performance of their activities on a supply chain-wide basis (Cooper et al., 1997).Conse-quently, firms in different parts of the supply chain tend to work to improve performance in thoseareas within their interest.To overcomethisproblem,they need a comprehensive overview oftheir supply chain activities and full appreciation of the impact of their performance on othermemberfirms in the supplychain.The objective of this study is to investigate the construct of, and develop a measurementinstrument for, SCP with a focus on the intermediary component, i.e., transport logistics, in asupply chain process. A measurement instrument is a collection of measuring items appliedcollectivelyto reveal a theoretical construct,e.g.SCP in transport logistics, which cannot beassesseddirectly(DeVellis.1991).Given the ambiguityin the literatureand the lackof em-pirically validated measurement instruments for SCP, this research objective is well justifiedwith the aim to extend SCM research to the transport logistics context.Weidentifythecomponents of sCP in transport logistics, develop the measurement items and instrument,evaluate their validity using empirical data, discuss the implications of the SCP construct, andprovide suggestions for using the validated measures in substantive research and practice intransportlogistics

In his seminal work on competitiveness of firms, Porter (1985) identifies customer values and costs to customers as critical elements to gain competitive advantages for a firm. The management of a supply chain encompasses these two elements, which together emphasize the importance of getting goods/services to customers at the right time, in the right place, under the right conditions, in the right quantities, and at the lowest possible costs. Porter (1985) emphasizes that, differentiation, one type of competitive advantage for a firm, is closely linked to the customer values of the product/service that can be delivered. Low cost, another type of competitive advantage, is reflected in the costs of the product/service to the customers. Christopher (1998) adds that a firm would achieve a competitive advantage by striving for excellence in both service and cost leadership. To this end, making proper performance measurement of a supply chain is necessary as it cultivates understanding between member firms in the supply chain for performance improvement (Dreyer, 2000; Fawcett and Cooper, 1998). Traditionally, the focus of performance measurement has been on process operations within the organizational boundaries of a firm (Short and Venkatraman, 1992). In the context of SCM, performance measurement involves not only the internal processes, but also requires an understanding of the performance expectation of other member firms in the supply chain, backward from the suppliers and forward to the customers (Normann and Ramırez, 1993). Coordination between the various parties in the supply chain is key to its effective implementation (Frohlich and Westbrook, 2001). As SCM focuses on process management beyond organizational boundaries, there is a need to measure performance for the effective management of a supply chain. Harrington (1991, p. 164) states that ‘If you cannot measure it, you cannot control it. If you cannot control it, you cannot manage it. If you cannot manage it, you cannot improve it’. In fact, the lack of relevant performance measures has been recognized as one of the major problems in process management (Davenport et al., 1996) and the management of a supply chain (Dreyer, 2000). Because of the different views on what should constitute supply chain performance (SCP), many firms have found it difficult to practice SCM (Beamon, 1999). A major contributing factor to this problem is that, with multiple parties having different interests, it is difficult for firms to effectively evaluate the performance of their activities on a supply chain-wide basis (Cooper et al., 1997). Consequently, firms in different parts of the supply chain tend to work to improve performance in those areas within their interest. To overcome this problem, they need a comprehensive overview of their supply chain activities and full appreciation of the impact of their performance on other member firms in the supply chain. The objective of this study is to investigate the construct of, and develop a measurement instrument for, SCP with a focus on the intermediary component, i.e., transport logistics, in a supply chain process. A measurement instrument is a collection of measuring items applied collectively to reveal a theoretical construct, e.g. SCP in transport logistics, which cannot be assessed directly (DeVellis, 1991). Given the ambiguity in the literature and the lack of empirically validated measurement instruments for SCP, this research objective is well justified with the aim to extend SCM research to the transport logistics context. We identify the components of SCP in transport logistics, develop the measurement items and instrument, evaluate their validity using empirical data, discuss the implications of the SCP construct, and provide suggestions for using the validated measures in substantive research and practice in transport logistics. 440 K.H. Lai et al. / Transportation Research Part E38 (2002) 439–456

441K.H.Lai etal./TransportationResearchPartE38 (2002)439-4562.Conceptual backgroundSCM is concerned with managing the upstream and downstream relationships with suppliersand customers to deliver superior customer value at the least cost to the chain as a whole(Christopher, 1998).Implementation of SCM requires that the internal perspective of perfor-mance measures be expanded to include both“interfunctional"and“partnership"perspectivesand avoid inward-looking and self-focused attitudes in the management approach (Holmberg,2000).This isto beachieved by closely integrating the internal functions within afirm and ef-fectively linking them with the external operations of member firms in the chain.To this end, anappropriateperformancemeasurement is conduciveto successful sCMimplementation(Lee andBillington, 1992).Mentzer and Konrad (1991)define performance measurement as effectiveness and efficiency inaccomplishing a given task in relation to how well a goal is met. In the logistics and supply chaincontext, effectiveness is concerned with the extent to which goals are accomplished and they mayincludelead-time, stockoutprobability,andfill rate.Efficiencymeasures how well theresourcesare utilized, for which themeasures may include inventory costs and operating costs. While manyfirms recognize both aspects of performance, they fail to understand them from a perspective of abalanced frameworkforperformance measurement (Brewerand Speh,2000).This could bedisruptive for performance management in a supply chain.For instance, one firmmay concen-trateon operationalefficiency,whiletheothersaremoreoncernedwithserviceeffectivenessinthe supply chain.The differences in the views of SCP would lead to inconsistency in the perfor-mancemeasures used across memberfirms in a supplychain andconsequently suboptimizesupplychain-wide performance (Bechtel and Jayaram, 1997; Caplice and Sheffi, 1995; Gunasekaranet al., 2001).Traditional performance measures such as profitability are less relevant for measuring SCPbecause they tend to have an “individual focus"and fail to consider chain-wide areas for per-formance improvement.Bechtel and Jayaram (1997)advocate the use of integratedmeasures, inaddition to non-integrated measures, that motivate firms to consider chain-wide performance,rather than their own individual performance measures.An exampleof an integrated measure iscash-to-cash cycle that spans functional and organizational boundaries to show all member firmshowthechainisperforming,andfostersincentivesforfirmstoworkwithothersinthechain.Incontrast, non-integrated measures only provide insights into potential problems within individualfirms in a supply chain.Otherthan integrated performancemeasures,there are conceptual frameworks on SCP.New(1996)presents a taxonomy for the classification of supply chain improvement.van Hoek (1998)proposes a framework at the firm's level of integration in the supply chain and the strategyadopted. Beamon (1999)develops a performance evaluation framework for manufacturingsupply chains, where resources, output, and flexibility are considered necessary components forSCP.Shah and Singh (2001)provide a framework for benchmarking internal SCP.Gunasek-aran et al. (2001) develop a conceptual model for SCP at three management levels. Even thoughthereexist a varietyof frameworks for SCPmeasurement,many companies still managetheirsupply chain in a way different from what their member firms desire. The main reason is thatthey lack agreement of goals and performance measures in their supply chain activities (Tanet al., 1999)

2. Conceptual background SCM is concerned with managing the upstream and downstream relationships with suppliers and customers to deliver superior customer value at the least cost to the chain as a whole (Christopher, 1998). Implementation of SCM requires that the internal perspective of performance measures be expanded to include both ‘‘interfunctional’’ and ‘‘partnership’’ perspectives and avoid inward-looking and self-focused attitudes in the management approach (Holmberg, 2000). This is to be achieved by closely integrating the internal functions within a firm and effectively linking them with the external operations of member firms in the chain. To this end, an appropriate performance measurement is conducive to successful SCM implementation (Lee and Billington, 1992). Mentzer and Konrad (1991) define performance measurement as effectiveness and efficiency in accomplishing a given task in relation to how well a goal is met. In the logistics and supply chain context, effectiveness is concerned with the extent to which goals are accomplished and they may include lead-time, stockout probability, and fill rate. Efficiency measures how well the resources are utilized, for which the measures may include inventory costs and operating costs. While many firms recognize both aspects of performance, they fail to understand them from a perspective of a balanced framework for performance measurement (Brewer and Speh, 2000). This could be disruptive for performance management in a supply chain. For instance, one firm may concentrate on operational efficiency, while the others are more concerned with service effectiveness in the supply chain. The differences in the views of SCP would lead to inconsistency in the performance measures used across member firms in a supply chain and consequently suboptimize supply chain-wide performance (Bechtel and Jayaram, 1997; Caplice and Sheffi, 1995; Gunasekaran et al., 2001). Traditional performance measures such as profitability are less relevant for measuring SCP because they tend to have an ‘‘individual focus’’ and fail to consider chain-wide areas for performance improvement. Bechtel and Jayaram (1997) advocate the use of integrated measures, in addition to non-integrated measures, that motivate firms to consider chain-wide performance, rather than their own individual performance measures. An example of an integrated measure is cash-to-cash cycle that spans functional and organizational boundaries to show all member firms how the chain is performing, and fosters incentives for firms to work with others in the chain. In contrast, non-integrated measures only provide insights into potential problems within individual firms in a supply chain. Other than integrated performance measures, there are conceptual frameworks on SCP. New (1996) presents a taxonomy for the classification of supply chain improvement. van Hoek (1998) proposes a framework at the firm’s level of integration in the supply chain and the strategy adopted. Beamon (1999) develops a performance evaluation framework for manufacturing supply chains, where resources, output, and flexibility are considered necessary components for SCP. Shah and Singh (2001) provide a framework for benchmarking internal SCP. Gunasekaran et al. (2001) develop a conceptual model for SCP at three management levels. Even though there exist a variety of frameworks for SCP measurement, many companies still manage their supply chain in a way different from what their member firms desire. The main reason is that they lack agreement of goals and performance measures in their supply chain activities (Tan et al., 1999). K.H. Lai et al. / Transportation Research Part E38 (2002) 439–456 441

442K.H.Laietal./TransportationResearchPartE38(2002)439-456Table 1ScORperformance measures fora supplychainSupply chain processMeasurement criteriaPerformanceindicatorsCustomerfacingSupply chain reliabilityDelivery performanceOrderfulfillmentperformancePerfect order fulfillmentFlexibility and ResponsivenessSupply chain response timeProduction flexibilityInternal facingCostsTotal logistics management costsValue added productivityReturn processing costAssetsCash-to-cash cycle timeInventory days of supplyAsset turnsAmong the extant sCP conceptualizations, the supply chain operations reference model(SCOR) developed by the Supply Chain Council (cf. Stewart, 1995)provides a useful frameworkthat considers the performance requirements of member firms in a supply chain. The SCORmodel views activities in the supply chain as a series of interlocking interorganizational processeswith each individual organization comprising four components: plan, source, make, and deliver.Each of these components is considered a critical intraorganizational process in the supply chainwithfourmeasurement criteria:(1)supply chain reliability,(2)responsiveness/flexibility,(3)costs,and (4) assets.The first two criteria deal with effectiveness-related (customer-facing)performancemeasures, while the other two are efficiency-related (internal-facing) performance measures of afirm. Customer-facing measures are concerned with how well a supply chain delivers products/services to customers, e.g.delivery performance.Internal-facing measures are concerned with theefficiency with which a supply chain operates, e.g. cash-to-cash cycle time (cf. Geary, 2001).In line with MentzerandKonrad (1991),the SCOR model provides an indication as to howeffective a firm uses resources in creating customer value. It considers the performance expecta-tions of member firms on both input and output sides of supply chain activities.The measurementcriteriaandindicatorsofperformancemeasurementinSCORacrosssupplychainmembers(cf.Stephens,2000), shown in Table l,providea useful framework for developing a construct and thecorrespondinginstrumentforSCPmeasurementinthetransportlogisticscontext.3. SCP in transport logisticsTransport logistics in a supply chain is usually an intermediary that facilitates the physical flowsof goods from a point of origin, i.e., shipper, to a point of destination, i.e., consignee.Firms intransport logistics perform the physical distribution function to move goods from one place toanother (Coyle et al., 1996) and the business process spans organizational boundaries, encom-passing shippers and consignees.Under this conception, SCP in transport logistics involves shippers on the input side andconsignees on the output side. The goal of a transport logistics service provider is to satisfy the

Among the extant SCP conceptualizations, the supply chain operations reference model (SCOR) developed by the Supply Chain Council (cf. Stewart, 1995) provides a useful framework that considers the performance requirements of member firms in a supply chain. The SCOR model views activities in the supply chain as a series of interlocking interorganizational processes with each individual organization comprising four components: plan, source, make, and deliver. Each of these components is considered a critical intraorganizational process in the supply chain with four measurement criteria: (1) supply chain reliability, (2) responsiveness/flexibility, (3) costs, and (4) assets. The first two criteria deal with effectiveness-related (customer-facing) performance measures, while the other two are efficiency-related (internal-facing) performance measures of a firm. Customer-facing measures are concerned with how well a supply chain delivers products/ services to customers, e.g. delivery performance. Internal-facing measures are concerned with the efficiency with which a supply chain operates, e.g. cash-to-cash cycle time (cf. Geary, 2001). In line with Mentzer and Konrad (1991), the SCORmodel provides an indication as to how effective a firm uses resources in creating customer value. It considers the performance expectations of member firms on both input and output sides of supply chain activities. The measurement criteria and indicators of performance measurement in SCORacross supply chain members (cf. Stephens, 2000), shown in Table 1, provide a useful framework for developing a construct and the corresponding instrument for SCP measurement in the transport logistics context. 3. SCP in transport logistics Transport logistics in a supply chain is usually an intermediary that facilitates the physical flows of goods from a point of origin, i.e., shipper, to a point of destination, i.e., consignee. Firms in transport logistics perform the physical distribution function to move goods from one place to another (Coyle et al., 1996) and the business process spans organizational boundaries, encompassing shippers and consignees. Under this conception, SCP in transport logistics involves shippers on the input side and consignees on the output side. The goal of a transport logistics service provider is to satisfy the Table 1 SCORperformance measures for a supply chain Supply chain process Measurement criteria Performance indicators Customer facing Supply chain reliability Delivery performance Order fulfillment performance Perfect order fulfillment Flexibility and Responsiveness Supply chain response time Production flexibility Internal facing Costs Total logistics management costs Value added productivity Return processing cost Assets Cash-to-cash cycle time Inventory days of supply Asset turns 442 K.H. Lai et al. / Transportation Research Part E38 (2002) 439–456

443K.H.Lai etal./Transportation ResearchPartE38 (2002)439-456customers (both upstream and downstream) in the chain with greater effectiveness and efficiencythan the competitors. The measurement of sCP in transport logistics needs to incorporate theseperformance aspects to be successful. For example, cost efficiency in providing the services mightbe an important performance measure for a transport logistics service provider. However, thismight not be desired by shippers and consignees. They would instead demand high quality andlow-pricedelivery of shipments conforming totheirrequirements.Anotherexampleis that de-laying shipments until carriage infull truck loads is possible may reduce the costs for organizingthedeliveryandimproveefficiencymeasuresforthetransportlogistics serviceprovider.However,this would lead to a reduction in the service effectiveness provided to shippers and consigneesNeither performance measures alone, effectiveness and eficiency,can fully reflect sCP in trans-portlogistics.In this regard, sCP in transport logistics should encompass not only operations eficiencyparameters, but also measures of service effectiveness (Kleinsorge et al.,1991) to meet the goals ofall parties,i.e., shipper, service provider and consignee. It must not be centered only on individualfunctional areas, but rather on the different parties involved in the transport logistics processesandtheoverallSCP(Cavinato.1992:Lee,2000)To this end, the SCORmodel in Table1 provides a useful framework.It represents a systematicapproach tomeasuringperformancewith inputs from,and outputs to,member firms in the supplychain and considers performanceassessment ona supply chain-widebasis,not just on that of anindividual component, e.g.providers of transport logistics services, in the chain. This is an im-portant point because it not only identifies both the effectiveness and efficiency aspects of per-formance, but also recognizes that there can be internal as well as customer-related reasons forperformance measurement. Based on this, three dimensions of SCP in transport logistics areidentified.Theseare: Service effectiveness for shippers (SES);.Operations efficiencyfortransport logistics serviceproviders (OE);.Serviceeffectivenessforconsignees (SEC).SES and SEC measure how well the activities are performed to meet the requirements ofshippers and consignees,respectively.OErefersto the efficiency of a transportlogisticsserviceprovider in the use of resources to perform its service activities. These three dimensions of SCP intransport logistics are congruent with the critical components for supply chain success postulatedin the SCOR model. In this study, the three-factor structure of the SCP construct is tested in afirst-order model, where SES, OE and SEC correlate among themselves in measuring the sameconstruct,i.e.,SCPintransportlogistics,andinasecond-ordermodel,wheretheSCPconstructistreated as a higher order model governing the covariance of the three dimensions of SES, OE andSEC.4.MethodologyFollowing Churchill's (1979)paradigmfor construct measurement, we first define the domainof a SCP construct in transport logistics, then operationalize the construct by developing a

customers (both upstream and downstream) in the chain with greater effectiveness and efficiency than the competitors. The measurement of SCP in transport logistics needs to incorporate these performance aspects to be successful. For example, cost efficiency in providing the services might be an important performance measure for a transport logistics service provider. However, this might not be desired by shippers and consignees. They would instead demand high quality and low-price delivery of shipments conforming to their requirements. Another example is that delaying shipments until carriage in full truck loads is possible may reduce the costs for organizing the delivery and improve efficiency measures for the transport logistics service provider. However, this would lead to a reduction in the service effectiveness provided to shippers and consignees. Neither performance measures alone, effectiveness and efficiency, can fully reflect SCP in transport logistics. In this regard, SCP in transport logistics should encompass not only operations efficiency parameters, but also measures of service effectiveness (Kleinsorge et al., 1991) to meet the goals of all parties, i.e., shipper, service provider and consignee. It must not be centered only on individual functional areas, but rather on the different parties involved in the transport logistics processes and the overall SCP (Cavinato, 1992; Lee, 2000). To this end, the SCORmodel in Table 1 provides a useful framework. It represents a systematic approach to measuring performance with inputs from, and outputs to, member firms in the supply chain and considers performance assessment on a supply chain-wide basis, not just on that of an individual component, e.g. providers of transport logistics services, in the chain. This is an important point because it not only identifies both the effectiveness and efficiency aspects of performance, but also recognizes that there can be internal as well as customer-related reasons for performance measurement. Based on this, three dimensions of SCP in transport logistics are identified. These are • Service effectiveness for shippers (SES); • Operations efficiency for transport logistics service providers (OE); • Service effectiveness for consignees (SEC). SES and SEC measure how well the activities are performed to meet the requirements of shippers and consignees, respectively. OE refers to the efficiency of a transport logistics service provider in the use of resources to perform its service activities. These three dimensions of SCP in transport logistics are congruent with the critical components for supply chain success postulated in the SCORmodel. In this study, the three-factor structure of the SCP construct is tested in a first-order model, where SES, OE and SEC correlate among themselves in measuring the same construct, i.e., SCP in transport logistics, and in a second-order model, where the SCP construct is treated as a higher order model governing the covariance of the three dimensions of SES, OE and SEC. 4. Methodology Following Churchill’s (1979) paradigm for construct measurement, we first define the domain of a SCP construct in transport logistics, then operationalize the construct by developing a K.H. Lai et al. / Transportation Research Part E38 (2002) 439–456 443

444K.H.Lai etal. /TransportationResearch PartE38 (2002)439-456measurement instrument. The instrument is pre-tested, modified, and used to capture data in across-sectional survey of transport logistics service providers.The following paragraphs describethese processes in detail.4.1.Domainspecificationand instrumentdevelopmentIntheprevious discussion,SCPintransportlogistics isidentifiedasa three-factormodel.Inline with SCOR, SES and SEC are customer-facing measures and concerned withthereliability(REL)andresponsiveness(RES)of a supplychainprocessperformedfor shippersandcon-signees,respectively.These two service-oriented components are operationalized bymodifying thereliability and responsivenessdimensions of the SERVQUAL instrumentdevelopedbyPara-suraman et al. (1988). I The modified measures gauge the service effectiveness performed re-spectively for shippers (SES-REL and SES-RES) and consignees (SEC-REL and SEC-RES).OE is concerned with the eficient use of resources in performing transport logistics services.InSCOR, there are two aspects of OE: cost-related and asset-related. In line with Mentzer andKonrad (1991),the cost-related aspect of OE (OE-COST)is operationalized by five broad cate-gories of logistics performance:transportation, warehousing, costs associated with thefacilitiesand manpowerused in providing the services, orderprocessing, and logistics administration.Theasset-related aspectofOE(OE-ASSET)is developed on thebasis of thethreemeasures suggestedin SCOR: cash-to-cash cycle time, utilization of facilities and manpower in providing the services,and assetturns.A total of 26 measurement items are generated for the measurement instrument: nine for SES,eight for OE and nine for SEC as shown in Appendix A. An example is added to each item toenrich the content and improve the comprehensiveness of the item in the instrument.2 Contentvalidity is concerned with the extent to which a specific set of items reflects a content domain(DeVellis, 1991).Assessing content validity helps to ensure that the items used to operationalizethe construct actually measure what they are supposed to measure (Churchill, 1979).We per-formed a content validation test by inviting some experts to review the measuring items to ensurethat they are representative of our SCP conceptualization in transport logistics.3Several changesin the wording were made and the items were subject to further refinement in a pilot test.'SERVQUAL is a widely accepted instrument to measure service quality across a wide variety of service domains,see Parasuraman et al. (1988,1994)for details.There are five dimensions in SERVQUAL:reliability,responsivenessassurances,empathy,and tangibility.The service-oriented component of the SCP construct regarding reliability andresponsiveness in this studyare developed on the basis of thefirsttwodimensions in SERVQUALbecauseof theirwideacceptanceand robustness in the literature.2The measurement items are measured on a five-point scale, ranging from an anchor Imuch worse than thecompetition,2-worse than thecompetition,3-same as the competition,4better than the competition, andsuperiorto the competition.Respondents wereinvited to evaluatetheperformance of theircompanies withrespectto the items on the five-point scale.The measurement items were included in a structured questionnaire for contentvalidation and refinement.Two neutral academics in the transport logisticsfield and two industry practitioners were invited toreviewtheitemsto ensure the relevance and clarity of the wording for the items in the instrument. Each of the reviewers was briefed onthe purpose of the study and asked to criticallyreview the items for completeness, understandability,terminology,andambiguity

measurement instrument. The instrument is pre-tested, modified, and used to capture data in a cross-sectional survey of transport logistics service providers. The following paragraphs describe these processes in detail. 4.1. Domain specification and instrument development In the previous discussion, SCP in transport logistics is identified as a three-factor model. In line with SCOR, SES and SEC are customer-facing measures and concerned with the reliability (REL) and responsiveness (RES) of a supply chain process performed for shippers and consignees, respectively. These two service-oriented components are operationalized by modifying the reliability and responsiveness dimensions of the SERVQUAL instrument developed by Parasuraman et al. (1988). 1 The modified measures gauge the service effectiveness performed respectively for shippers (SES-REL and SES-RES) and consignees (SEC-REL and SEC-RES). OE is concerned with the efficient use of resources in performing transport logistics services. In SCOR, there are two aspects of OE: cost-related and asset-related. In line with Mentzer and Konrad (1991), the cost-related aspect of OE (OE-COST) is operationalized by five broad categories of logistics performance: transportation, warehousing, costs associated with the facilities and manpower used in providing the services, order processing, and logistics administration. The asset-related aspect of OE (OE-ASSET) is developed on the basis of the three measures suggested in SCOR: cash-to-cash cycle time, utilization of facilities and manpower in providing the services, and asset turns. A total of 26 measurement items are generated for the measurement instrument: nine for SES, eight for OE and nine for SEC as shown in Appendix A. An example is added to each item to enrich the content and improve the comprehensiveness of the item in the instrument. 2 Content validity is concerned with the extent to which a specific set of items reflects a content domain (DeVellis, 1991). Assessing content validity helps to ensure that the items used to operationalize the construct actually measure what they are supposed to measure (Churchill, 1979). We performed a content validation test by inviting some experts to review the measuring items to ensure that they are representative of our SCP conceptualization in transport logistics. 3 Several changes in the wording were made and the items were subject to further refinement in a pilot test. 1 SERVQUAL is a widely accepted instrument to measure service quality across a wide variety of service domains, see Parasuraman et al. (1988, 1994) for details. There are five dimensions in SERVQUAL: reliability, responsiveness, assurances, empathy, and tangibility. The service-oriented component of the SCP construct regarding reliability and responsiveness in this study are developed on the basis of the first two dimensions in SERVQUAL because of their wide acceptance and robustness in the literature. 2 The measurement items are measured on a five-point scale, ranging from an anchor 1––much worse than the competition, 2––worse than the competition, 3––same as the competition, 4––better than the competition, and 5––superior to the competition. Respondents were invited to evaluate the performance of their companies with respect to the items on the five-point scale. The measurement items were included in a structured questionnaire for content validation and refinement. 3 Two neutral academics in the transport logistics field and two industry practitioners were invited to review the items to ensure the relevance and clarity of the wording for the items in the instrument. Each of the reviewers was briefed on the purpose of the study and asked to critically review the items for completeness, understandability, terminology, and ambiguity. 444 K.H. Lai et al. / Transportation Research Part E38 (2002) 439–456

445K.H.Laietal./TransportationResearchPartE38(2002)439-456Table 2Summary measurement resultsS.D.Number of itemsMeanAlphaFactorsRange of item-to-total correlations545SES-REL4.12 (3.80)0.52 (0.49)0.74 (0.73)0.450.57 (0.360.64)SES-RES4.04 (3.92)0.48 (0.53)0.76 (0.77)0.460.63 (0.450.68)OE-COST3.65 (3.69)0.73 (0.49)0.87 (0.70)0.620.77 (0.420.55)354OE-ASSET3.74 (3.65)0.41 (0.62)0.80 (0.79)0.560.72 (0.580.74)SEC-REL4.03 (3.87)0.63 (0.42)0.86 (0.66)0.57-0.75 (0.380.52)SEC-RES4.01 (3.84)0.52 (0.46)0.83 (0.60)0.61-0.70 (0.300.61)Note: Entries in theparentheses arepilottest results.4.2. Pilot testA pilot test was carried out to further test and refine the instrument. The pilot test was con-ducted with 30 postgraduate students studying a part-time Master's degree in InternationalShipping and Transport Logistics at The Hong Kong Polytechnic University (who were full-timetransport logistics practitioners) and a convenient sample of 20 practitioners in the field. 4 A totalof 32valid responses were collected inthepilot test.Based on the32responses,preliminaryvalidity of the instrument was established on the basis of two criteria: content validity, andconstruct validity from an item-to-total correlation analysis and reliability test. 6 The results ofthe pilot test are given in Table 2.4.3.Data collectionTo further explorethe SCP construct, thefinal version ofthequestionnaire was mailed, with acoveringletterand a self-addressedprepaid return envelope, to thecomplete sampleof all 924companies in the Schednet Asian Logistics Directory (Schednet, 2001), in which all the companiesinvolved in transport logistics in Hong Kong are listed.7Weused thekey informant strategy tocarry out the survey research (Phillips and Bagozzi, 1986). Target respondents were generalmanagers or logistics managers of the sampled companies.The questionnaire was mailed twice:one month after the first mailing, the questionnaire was again mailed to the non-respondents.4The pilot test samples were asked to complete the questionnaire and to offer suggestionsfor improvement of themeasurement instrument.The pilot test resulted in minormodifications to the wording and examples provided in somemeasurementitems5Content validity is ensured because themeasurement items were derived andmodifiedfrom established measures,aswell as from suggestions from academics and practitioners in the field.Moreover, the pilot test respondents indicatedthat the content of SCP in transport logistics is well represented by the items included in the measurement instrument.These procedures are entirely consistent with those required for attaining high content validity.The construct validity of the SCP scalewas examined using a reliabilitytest with the coefficient alpha computed foreach of the sub-dimensions, e.g.SES-REL, and item-to-total correlation analysis.These procedures resulted in a set ofitems with coefficient alpha values all higher than 0.60 and all item loadings in item-to-total correlation analysis weregreater than 0.307The sample represents four broad categories of companies in the industry: sea transport,freightforwarding,airtransport, third-party logistics services.8 These executives were targeted because they possess expert knowledge of SCP in transport logistics in theircompanies

4.2. Pilot test A pilot test was carried out to further test and refine the instrument. The pilot test was conducted with 30 postgraduate students studying a part-time Master’s degree in International Shipping and Transport Logistics at The Hong Kong Polytechnic University (who were full-time transport logistics practitioners) and a convenient sample of 20 practitioners in the field. 4 A total of 32 valid responses were collected in the pilot test. Based on the 32 responses, preliminary validity of the instrument was established on the basis of two criteria: content validity, 5 and construct validity from an item-to-total correlation analysis and reliability test. 6 The results of the pilot test are given in Table 2. 4.3. Data collection To further explore the SCP construct, the final version of the questionnaire was mailed, with a covering letter and a self-addressed prepaid return envelope, to the complete sample of all 924 companies in the Schednet Asian Logistics Directory (Schednet, 2001), in which all the companies involved in transport logistics in Hong Kong are listed. 7 We used the key informant strategy to carry out the survey research (Phillips and Bagozzi, 1986). Target respondents were general managers or logistics managers of the sampled companies. 8 The questionnaire was mailed twice: one month after the first mailing, the questionnaire was again mailed to the non-respondents. 4 The pilot test samples were asked to complete the questionnaire and to offer suggestions for improvement of the measurement instrument. The pilot test resulted in minor modifications to the wording and examples provided in some measurement items. 5 Content validity is ensured because the measurement items were derived and modified from established measures, as well as from suggestions from academics and practitioners in the field. Moreover, the pilot test respondents indicated that the content of SCP in transport logistics is well represented by the items included in the measurement instrument. These procedures are entirely consistent with those required for attaining high content validity. 6 The construct validity of the SCP scale was examined using a reliability test with the coefficient alpha computed for each of the sub-dimensions, e.g. SES-REL, and item-to-total correlation analysis. These procedures resulted in a set of items with coefficient alpha values all higher than 0.60 and all item loadings in item-to-total correlation analysis were greater than 0.30. Table 2 Summary measurement results Factors Number of items Mean S.D. Alpha Range of item-to-total correlations SES-REL 5 4.12 (3.80) 0.52 (0.49) 0.74 (0.73) 0.45–0.57 (0.36–0.64) SES-RES 4 4.04 (3.92) 0.48 (0.53) 0.76 (0.77) 0.46–0.63 (0.45–0.68) OE-COST 5 3.65 (3.69) 0.73 (0.49) 0.87 (0.70) 0.62–0.77 (0.42–0.55) OE-ASSET 3 3.74 (3.65) 0.41 (0.62) 0.80 (0.79) 0.56–0.72 (0.58–0.74) SEC-REL 5 4.03 (3.87) 0.63 (0.42) 0.86 (0.66) 0.57–0.75 (0.38–0.52) SEC-RES 4 4.01 (3.84) 0.52 (0.46) 0.83 (0.60) 0.61–0.70 (0.30–0.61) Note: Entries in the parentheses are pilot test results. 7 The sample represents four broad categories of companies in the industry: sea transport, freight forwarding, air transport, third-party logistics services. 8 These executives were targeted because they possess expert knowledge of SCP in transport logistics in their companies. K.H. Lai et al. / Transportation Research Part E38 (2002) 439–456 445

446K.H.Laietal./TransportationResearchPartE38(2002)439-456Table 3Profile of the respondent companies (n = 134)Natureof business30 (22.4%)Sea transport49 (36.6%)Freight forwardingAirtransport2 (1.5%)Third party logistics services53 (39.5%)Numberof employeesBelow 100102 (76.1%)100-49923 (17.2%)500-9991 (0.7%)Over10007 (5.2%)Unknown1 (0.7%)Level of turnover (HKS)17 (12.7%)Below1million1-<10 million40 (29.9%)10-100million45 (33.6%)Over100million28 (20.9%)Unknown4 (3.0%)A total of 139 questionnaires were returned, but five of them were not useable because ofsignificant data missing and incompleteness.The remaining 134 responses-97in thefirstmailingand37inthe secondmailing-representaneffectiveresponserateof14.5%.Theprofilesoftherespondent companies and their characteristics are displayed in Table 3.A comparison of early (those responding to thefirst mailing)and late (those responding to thesecond mailing) respondents was carried out to test for non-response bias (Armstrong andOverton, 1977).9The26 measurement items in this study were randomly selected for a non-response bias test. We divided the 134 survey respondents into two groups based on their responseswave (first and second) and performed t-tests on the responses of the two groups. At the 5% level,therearenosignificantdifferencesbetweenthetwogroupsinthemeasurementitems.Althoughtheresults do not rule out the possibility of non-response bias, they suggest that non-response may notbe a problem to the extent that the late respondents represent the opinions of non-respondents.5. Results5.1. Validity and reliabilityWe first tested the measurement properties of the sub-dimensions of the SCP construct usingreliability test and item-to-total correlation analysis, followed by confirmatory factor analysis(CFA)(Anderson, 1987; Gerbing and Anderson, 1988; Joreskog,1993).10 In this study,we first9This method is based on theassumption that the opinions of late respondents are somewhat representative of theopinions of non-respondents10 The CFA was conducted using Maximum Likelihood Estimation in AMOS 4.0 (Arbuckle and Wothke, 1999)

A total of 139 questionnaires were returned, but five of them were not useable because of significant data missing and incompleteness. The remaining 134 responses––97 in the first mailing and 37 in the second mailing––represent an effective response rate of 14.5%. The profiles of the respondent companies and their characteristics are displayed in Table 3. A comparison of early (those responding to the first mailing) and late (those responding to the second mailing) respondents was carried out to test for non-response bias (Armstrong and Overton, 1977). 9 The 26 measurement items in this study were randomly selected for a nonresponse bias test. We divided the 134 survey respondents into two groups based on their responses wave (first and second) and performed t-tests on the responses of the two groups. At the 5% level, there are no significant differences between the two groups in the measurement items. Although the results do not rule out the possibility of non-response bias, they suggest that non-response may not be a problem to the extent that the late respondents represent the opinions of non-respondents. 5. Results 5.1. Validity and reliability We first tested the measurement properties of the sub-dimensions of the SCP construct using reliability test and item-to-total correlation analysis, followed by confirmatory factor analysis (CFA) (Anderson, 1987; Gerbing and Anderson, 1988; Joreskog, 1993). € 10 In this study, we first Table 3 Profile of the respondent companies (n ¼ 134) Nature of business Sea transport 30 (22.4%) Freight forwarding 49 (36.6%) Air transport 2 (1.5%) Third party logistics services 53 (39.5%) Number of employees Below 100 102 (76.1%) 100–499 23 (17.2%) 500–999 1 (0.7%) Over 1000 7 (5.2%) Unknown 1 (0.7%) Level of turnover (HK$) Below 1 million 17 (12.7%) 1–<10 million 40 (29.9%) 10–100 million 45 (33.6%) Over 100 million 28 (20.9%) Unknown 4 (3.0%) 9 This method is based on the assumption that the opinions of late respondents are somewhat representative of the opinions of non-respondents. 10 The CFA was conducted using Maximum Likelihood Estimation in AMOS 4.0 (Arbuckle and Wothke, 1999). 446 K.H. Lai et al. / Transportation Research Part E38 (2002) 439–456

developed measures based on theory and previous research (Lai et al., 2001). CFA was used to assess how well the observed variables, i.e., measurement items, reflect unobserved or latent variables, i.e., the sub-dimensions, in the hypothesized structure. A strong a priori basis warrants the use of CFA instead of exploratory factor analysis (EFA). The reliability test and item-to-total correlation analysis results provided in Table 2 suggest a reasonable fit of the latent factors to the data. Cronbach alpha values for all six factors, i.e., subdimensions, are all greater than 0.70 and the item loadings on the factors are all acceptable, i.e., >0.40. These tests, however, do not allow for unidimensionality, 11 convergent validity, 11 nor discriminant validity. 12 We proceeded to test them using CFA. The CFA results for SES, OE, and SEC are provided in Table 4. A series of goodness-of-fit indices, i.e., CFI > 0:90, GFI > 0:90, NFI > 0:90 and RMR 0:4 and t > 2). A series of pairwise CFAs were conducted to assess the discriminant validity of the subdimensions using v2 difference tests (Anderson and Gerbing, 1988). 13 This test was performed on all possible pairs of the factors and Table 5 reports the results of the 15 pairwise tests of the factors. Discriminant validity is not achieved in some cases (SES-REL and SES-RES, SES-RES 11 Unidimensionality and convergent validity refers to the existence of one latent trait or construct underlying a set of measures (Gerbing and Anderson, 1988). In CFA, the measurement items are restricted to load on their respective subdimensions in the SCP and the sub-dimensions are allowed to be correlated between themselves in their respective measurement models. 12 Discriminant validity is the degree to which a dimension in a theoretical system differs from other dimensions in the same theoretical system (Churchill, 1979). Table 4 Results from confirmatory factor analysis model for SES, OE and SEC Measurement models Range of standardized loadings Range of t-values CFI GFI NFI RMR v2 (d.f., prob.) SES 0.99 0.96 0.93 0.03 27.72 (26, p > 0:10) SES-REL 0.52–0.69 4.89–7.47 SES-RES 0.59–0.74 6.11–7.47 OE 0.88 0.86 0.86 0.05 85.45 (19, p < 0:01) OE-COST 0.68–0.85 7.64–9.73 OE-ASSET 0.71–0.82 7.89–8.22 SEC 0.95 0.91 0.92 0.03 57.29 (26, p < 0:01) SEC-REL 0.64–0.81 6.91–7.75 SEC-RES 0.73–0.77 8.25–8.73 Note: For standardized loading of individual measurement items, see Appendix A. 13 This was conducted by forcing measurement items of each pair of factors (sub-dimensions) into a single underlying factor, leading to a significant deterioration of model fit relative to a two-factor model. Such a result, this implies the presence of discriminant validity between the pair of factors (Bagozzi and Phillips, 1982). K.H. Lai et al. / Transportation Research Part E38 (2002) 439–456 447

448K.H.Laietal./TransportationResearchPartE38(2002)439-456Table5Discriminant validity checks:differences12345Factors1.SES-REL1.802. SES-RES25.1147.853.OE-COST20.9443.5128.414.OE-ASSET2.5248.935.SEC-REL20.8362.3840.696.9374.9552.745.706. SEC-RESNote:differencebetween the separate latentfactors measurement modelanda onelatent factormeasurement model(alltests=1df):×>11,p6.7.p3.85.p<0.05and SEC-REL). This was expected as they are the sub-dimensions of the SCP construct and aremeasuringahigherorderlatentfactor,i.e.,SCPintransportlogistics.Thesignificantresultsof thedifference tests (13 out of 15)attest to the presence of discriminant validity between any twofactors (Anderson and Gerbing, 1988). Upon obtaining satisfactory reliability and validity testresults,we averaged thevalues ofthemeasurement itemsfor each sub-dimension and usethesearithmetic means as single-indicator constructs to measure SCP in transport logistics in subse-quent stages. 145.2.Testingfirst-order and second-ordermodelsIn the previous discussion, SES, OE and SEC are specified as a priori factors of SCP intransport logistics.In the first-order model, SES,OE and SEC are correlated measures for SCP intransportlogistics.Alternatively,SCP in transport logistics may be operationalized as a second-order model, 15 where the three dimensions are governed by a higher order factor, i.e., SCP intransport logistics.The results of themodel estimation are shown in Figs.1 and 2.The first-order model for testing the existence of SCP in transport logistics implies that SES,OE and SEC are correlated but not governed by a common latent factor.Although the statisticis significant (= 25.08; df = 6; p< 0.001), other fit indices suggest good fits for the first-ordermodel.The GFI is 0.94, which is greater than 0.90 as recommended by Joreskog (1993), sug-14By using summaryconstructsacomplex modelissimplified,andtheconceptofamultipleindicator ismaintained(Garver and Mentzer, 1999).It also reduces the model's complexity, identification problems, and the variables tosample size ratio (Marsh and Hau, 1999).This method also allows us to test the SCP construct based on a sample size of134 respondents.Another advantage of using a summary construct is that it provides more meaningful informationsince it signals where potential problems in SCPmayexist.For example, if SCP is not performing up to expectation, it iseasier to identify the problem in one of these six sub-dimensions and to indicate where more effort should be put.Instead of concentrating on individual measurement items, this approach allows the examination of the overalltheoreticalSCPconstructatahigherlevelofabstractionIs In the second-order model, the correlations between the factors are denoted by a second-order factor. Thisalternative model explains the covariation in an alternative way (three paths in contrast to three correlations).Comparingthetwomodelscanprovidefurthermeasurementefficacy(Joreskog,1993)

and SEC-REL). This was expected as they are the sub-dimensions of the SCP construct and are measuring a higher order latent factor, i.e., SCP in transport logistics. The significant results of the v2 difference tests (13 out of 15) attest to the presence of discriminant validity between any two factors (Anderson and Gerbing, 1988). Upon obtaining satisfactory reliability and validity test results, we averaged the values of the measurement items for each sub-dimension and use these arithmetic means as single-indicator constructs to measure SCP in transport logistics in subsequent stages. 14 5.2. Testing first-order and second-order models In the previous discussion, SES, OE and SEC are specified as a priori factors of SCP in transport logistics. In the first-order model, SES, OE and SEC are correlated measures for SCP in transport logistics. Alternatively, SCP in transport logistics may be operationalized as a secondorder model, 15 where the three dimensions are governed by a higher order factor, i.e., SCP in transport logistics. The results of the model estimation are shown in Figs. 1 and 2. The first-order model for testing the existence of SCP in transport logistics implies that SES, OE and SEC are correlated but not governed by a common latent factor. Although the v2 statistic is significant (v2 ¼ 25:08; df ¼ 6; p 11, p 6:7, p 3:85, p < 0:05. 14 By using summary constructs, a complex model is simplified, and the concept of a multiple indicator is maintained (Garver and Mentzer, 1999). It also reduces the model’s complexity, identification problems, and the variables to sample size ratio (Marsh and Hau, 1999). This method also allows us to test the SCP construct based on a sample size of 134 respondents. Another advantage of using a summary construct is that it provides more meaningful information since it signals where potential problems in SCP may exist. For example, if SCP is not performing up to expectation, it is easier to identify the problem in one of these six sub-dimensions and to indicate where more effort should be put. Instead of concentrating on individual measurement items, this approach allows the examination of the overall theoretical SCP construct at a higher level of abstraction. 15 In the second-order model, the correlations between the factors are denoted by a second-order factor. This alternative model explains the covariation in an alternative way (three paths in contrast to three correlations). Comparing the two models can provide further measurement efficacy (Joreskog, 1993). € 448 K.H. Lai et al. / Transportation Research Part E38 (2002) 439–456

点击进入文档下载页（PDF格式）

共18页，试读已结束，阅读完整版请下载

点击下载（PDF格式）

浏览记录