EXPLORING THE LINKAGES BETWEEN DEMING’S PRINCIPLE, …

205

Ciptono—Exploring the Linkages between Deming’s Principle, World-Class Company, ...

Gadjah Mada International Journal of BusinessMay-August 2005, Vol. 7, No. 2, pp. 205—239

EXPLORING THE LINKAGES BETWEENDEMING’S PRINCIPLE, WORLD-CLASS COMPANY,

OPERATIONAL EXCELLENCE, AND COMPANYPERFORMANCE IN AN OIL AND

GAS INDUSTRY SETTING*

Wakhid Slamet Ciptono

This study explores the linkages between Deming’s Principle,World-Class Company, Operational Excellence, and Company Per-formance in the Indonesia’s oil and gas industry. The aim of this studyis to examine the causal relationships model between the Deming’sPrinciple (DP), World-Class Company (WCC), Operational Excel-lence (OE), and Company Performance (Monetary Gain Perfor-mance or MGP and Value Gain Performance or VGP). The authorused 140 strategic business units (SBUs) in 49 oil and gas companiesin Indonesia. The survey was administered to every level of manage-ment at each SBU (Top, Middle, and Low Level Management). Amultiple informant sampling unit is used to ensure a balanced view ofthe relationships between the research constructs, and to collect datafrom the most informed respondents on different levels of manage-ment. A total of 1,332 individual usable questionnaires were returnedthus qualified for analysis, representing an effective response rate of50.19 percent.

Path analysis and structural equation modeling (SEM) are usedto analyze the effect of Deming’s principle on company performance

* Acknowledgements:

The author would like to acknowledge the work of three supervisors—Assoc. Prof. Dr. AbdulRazak Ibrahim, Assoc. Prof. Dr. Ainin Sulaiman, and Assoc. Prof. Dr. Tan Lee Peng and twoanonymous referees. Their advices and constructive comments on an earlier draft of this paper resultedin significant changes and improvements. I am thankful for their dedication to the field devotion tohelping others. “Promote the good and prevent the bad.”

206

Gadjah Mada International Journal of Business, May-August 2005, Vol. 7, No. 2

Introduction

Total quality management (TQM)as articulated by the quality manage-ment gurus —such as Juran (1994),Ishikawa (1985), Deming (1986), andCrosby (1989)— is a set of powerfulinterventions wrapped in a highly at-tractive package. When implementedwell, TQM can help an organizationimproves itself and, in the process,better serves its community and itsown stakeholders (Hackman andWageman 1995). Researchers like Imai(1986) Usilaner (1992), Shetty (1993),Smith and Sibler (1994), Krumwiedeet al. (1996), and Shin et al. (1998)suggested that top management plays

such a key role in establishing andmaintaining any type of corporate cul-ture, its leadership behavior and per-sonality become critical to successfulimplementation of the TQM philoso-phy.

The characteristic of successfulTQM implementation program encour-ages organizations to address qualityon a broad range of issues (i.e., world-class standards, operational excellence,and company performance). Not all ofthe implementations of TQM have beensuccessful as expected. Researcherslike Harari (1993; 1997), Becker(1993), Russel (1993), Usilaner (1992),Eskildson (1994), Kolesar (1995),Krumwiede et al. (1996), Tatikonda

Keywords: company performance; Deming’s principles; operational excellence; world-class company

and to investigate the interrelationships between Deming’s principle,world-class company, operational excellence, and company perfor-mance. The results show that Deming’s Principle has significantpositive and indirect effect on company performance (monetary gainperformance and value gain performance). Although the Deming’sPrinciple has no significant direct effects on company performance,the Deming’s Principle has significant positive effects on the interven-ing variables (world-class company and operational excellence). Theresult also shows that a complete model fit and the acceptableparameter level that indicate the overall parameter are good fitbetween the hypothesized model and the observed data. By concen-trating on a single industry (oil and gas), SEM specification of thecausal relationship model between five constructs can be morecomplete and specific because unique characteristics of the oil and gasindustry can be included (upstream and downstream chain activities).Finally, the particular design of the research and the findings suggestthat the structural model of the study has a great potential forreplication to manufacturing as well as service operations.

207


and Tatikonda (1996) and Anderson etal. (1998), suggested that TQM is fail-ing at a rate of two out of three.

Based on the lessons from suc-cessful and unsuccessful TQM imple-mentations that have already been doneby worldwide organizations in the pasttwo decades, Hackman and Wageman(1995) stated that the challenge in re-search on TQM is to specify and col-lect data about those processes thatwould be expected to result from TQMand that should, all else being equal,contribute to company performance.The major emphasis of this study is onan empirical research orientation ofthe development of the causal rela-tionship of Deming’s principle, world-class company, operational excellence,and company performance in theIndonesia’s oil and gas industry. Thisresearch orientation is consistent withthe fact that TQM is still recognizedonly recently by companies as a power-ful competitive strategy in order toachieve world-class company success-fully (Madu and Kuei 1993 in Ahire etal. 1995).

Indonesia’s oil and gas compa-nies realize that they must make somevery fundamental changes to competein unpredictable oil and gas market.With substantial reserves of naturalgas, coal and geothermal, Indonesiacould remain as a net energy exporterfor a longer period than current fore-casts (Embassy of the USA 2004). Todo so, the government has implementedlegislation and policies that will attractnew private direct investment and ra-tionalize use of Indonesia’s energy

resources (i.e. oil and gas). Energypolicy reform is necessary in order tomaintain Indonesia’s status as a net oiland gas exporter and enhance efficientusage of energy resources.

In addition, the Indonesian Par-liament passed the oil and gas bill intothe new law concerning oil and gas onOctober 23, 2001. The new law (Lawof the Republic of Indonesia number22/2001) authorizes the establishmentof an implementation agency for oiland gas upstream operations and adownstream regulatory agency to as-sume state oil and gas company’s roles.The implementation agency has re-placed the state oil and gas company inmanaging Production Sharing Con-tract (PSC) with private oil and gascompanies. The new law also removedthe state oil and company’s monopolyin the downstream sector with the regu-latory agency assuming the responsi-bility for managing natural gas anddomestic fuel distribution and supply(Embassy of the USA 2004; Law of theRepublic of Indonesia number 22/2001).

In order, to meet the future chal-lenges of discovering new reserves,the oil and gas companies have re-structured their companies into a Stra-tegic Business Unit (SBU) organiza-tion towards achieving the world-classvision (to be recognized as world-classcompanies committed to operationalexcellence). The SBUs have alreadyimplemented TQM program to achievethis vision. Although TQM programhad been, and will continue to be vitalpart of companies’ operations, compa-

208


nies must fundamentally rethink theirways of conducting business and havethe courage to implement Deming’sprinciple where necessary to achievethe real changes in company perfor-mance (monetary gain and value gainperformances) through world-classcompany and operational excellence.

The rest of this paper is organizedas follows. The next sections of thisstudy discuss objectives of the study,related literature and asses TQM prac-tices, review of methodology, and in-terpretation of results. The conclusionsand contributions are provided at thelast session.

Objectives of the Study

To validate the oil and companies’claim to TQM practices, the causalrelationships of Deming’s principle,world-class company, operational ex-cellence, and company performancehave been developed through thisstudy. In light of these research con-structs, the objectives for this study aretwofold:

(1) To develop a path analytical frame-work or structural equation modelwhich includes the interrelation-ships between five research con-structs: Deming’s principles as theindependent variable, world-classcompany, and operational excel-lence constructs as the mediatingvariables and further, the impor-tance of considering two distincttypes of company performance(monetary gain performance andvalue gain performance) as the de-pendent variables;

(2) To establish the relationshipsamong five research constructsthat are both substantively mean-ingful and statistically well-fittingfor the Indonesia’s oil and gas in-dustry in order to fulfill its vision tobe recognized as World-Class Com-pany committed to operational ex-cellence based on the oil and gaspolicy reform.

A testable model in this studywould be beneficial to conduct an em-pirical study that more systematicallyinvestigates the extent to which TQMpractices (based on Deming’s Prin-ciple) and success as connections ofdiffering contemporary managementpractices, such as world-class com-pany, operational excellence, and com-pany performance. Such study couldbe done not only within the same indus-try, but also across different firms anddifferent industries. The researchcould verify, or refute, a testable modelin this study might suggest more pre-cisely how Deming’s Principle varydepending upon the other change ini-tiatives or situational variables. Theapproach of the study is to conductempirical research guided by existingtheories.

Related Literature andAssessing TQM Practices

According to Hackman andWageman (1995), a full-fledged evalu-ation of a TQM program would in-clude three distinct type of assess-ment. First is empirical study that TQMactually has been implemented. Sec-

209


ond is determination of whether TQMalters the degree to which the improve-ments in organizational functions thatare expected actually are observed.Third comes assessment of the degreeto which improvements in bottom-lineorganizational effectiveness are found.It is important to examine both process(Deming’s principle, world-class com-pany, operational excellence) and out-come criteria (company performance)in measuring and assessing TQM prac-tices. In addition, empirical evaluationof TQM program presents a signifi-cant challenge to researchers becausewhat must be done to accomplish thethree assessments involves very dif-ferent methods and analytical strate-gies.

To assess how TQM is actuallypracticed in the Indonesia’s oil and gasindustry, the author addresses two ques-tions. First, in what ways are contem-porary implementations of TQM con-sistent with the founder’s tenets(Deming’s principle)? Second, in whatways do current practices differ fromthe Deming’s prescriptions, and dothose differences enrich the core ideasof TQM or diverge from them? An-swering the two questions will con-tribute to a deeper understanding ofthe relationship between the role ofeach of the elements of TQM(Deming’s principles) and companyperformance. This would help manag-ers with the allocation of resources tothose categories that have the mostsignificant effect on company perfor-mance (Samson and Terziovski 1999).Samson and Terziovski stated that only

a small percentage of over 1000 ar-ticles on the TQM philosophy andmethods attempted to test the strengthof the relationship between TQM andcompany performance.

In addition, attempts are neededto realize that TQM practices neednot to operate in isolation from otherchange initiative programs, such asoperational excellence and world-classcompany —it could be integrated. Thisstudy investigates the rational of link-age (a marriage) between five con-structs (Deming’s Principles, world-class company, operational excellence,and two distinct types of companyperformance— monetary and valuegains performance). Dave andBuschmann (1998) stated the goal ofworld-class company and operationalexcellence is to reduce total operatingcosts, to improve productivity in analready well-understood critical suc-cess factors such as critical successfactors of TQM implementation, andto enable previously unavailable world-class strategic capabilities. SuccessfulTQM implementation model (basedon Deming’s Principle) should con-sider the other contemporary manage-ment such as World-Class Company,Operational Excellence, and CompanyPerformance as an integrated network(connections) management system.The study wants to determine that withconnections, TQM practices could berepositioned as a more impactive busi-ness improvement program. As a fur-ther effort to renew interest in TQMpractices, all SBUs along the supply –and demand – chains in the oil and gas

210


industry begin to realize that they de-pended on each other and that poorquality from one SBU partner mush-roomed to affect others.

A number of research studies ofTQM and quality award systems havebeen conducted, and led to a debateabout the effectiveness of such awardsand of the various TQM elements. Thesestudies have been either perceptual stud-ies or small-scale empirical works(Samson and Terziovski 1999). Thepresent study is a large cross-sectionalexamination of 140 SBUs in oil and gascompanies with 1332 respondents (mul-tiple informant). In this section theauthor examines some of the key exist-ing empirical studies.

Tamimi (1998) in his study of asecond-order factor analysis of criticalTQM factors stated that the effectivetransformation to the total quality man-agement (TQM) organization has beenlinked to the extent to which firmssuccessfully implement certain criticalTQM practices. Tamimi (1998) devel-oped a second-order factor model totest whether a set of critical qualitymanagement factors load on an overallconstruct that may be termed TotalQuality Management. Using surveydata collected from 173 manufacturingand service firms, the LISREL VIIcomputer program is used to estimateand validate the proposed model. Theresults provide the eight factors collec-tively load on a single factor calledTQM. The eight critical factors of TQMwere top management commitment,supervisory leadership, education,cross-functional communications to

improve quality, supplier management,quality training, product/service inno-vation, and providing assurance foremployees.

Saraph et al. (1989) provided asynthesis of the quality literature byidentifying eight critical factors of qual-ity management in a business unit. Theeight critical factors of quality man-agement based on Saraph et al. (1989)are: the role of management leadershipand quality policy, role of the qualitydepartment, training, product/servicedesign, supplier quality management,process management, quality data andreporting, and employee relations. Op-erational measures of these factors aredeveloped using data collected from162 general managers and quality man-agers of 89 business units or 20 compa-nies. Researchers can use such mea-sures to better understand quality man-agement practice and to build theoriesand models that relate the critical fac-tors of quality management to theorganization’s quality environment andquality performance.

Black and Porter (1996) extracteda series of items from the BaldrigeModel and established literature. Theseitems formed the basis of a question-naire sent to over 200 managers. Datawas examined using several well-es-tablished analytical techniques thatidentified 10 critical factors of TQM.These ten critical factors were peopleand customer management, supplierpartnership, communication of im-provement information, customer sat-isfaction orientation, external interfacemanagement, strategic quality man-

211


agement, teamwork structures for im-provement, operational quality plan-ning, quality improvement measure-ment systems, and corporate qualityculture. The findings can be used toimprove existing self-assessmentframeworks such as the Baldrige andEuropean Quality Award criteria.

Samson and Terziovski (1999)examined the TQM practices and op-erational performance of a large num-ber of manufacturing companies (1200companies) in Australia and NewZealand in order to determine the rela-tionships between these practices, in-dividually and collectively, and firmperformance. The study showed thatthe relationship between TQM prac-tice (leadership, management ofpeople, and customer focus) and orga-nizational operational performance issignificant in a cross-sectional sense.

Antony et al. (2002) provided anempirical study on the identificationof the critical success factors (CSFs ofTQM) implementation in Hong Kongindustries. Through a thorough anddetailed analysis of the literature, 11success factors with 72 elements wereidentified to develop a questionnaire.These items were empirically testedby data collected from 32 companiesin Hong Kong. A factor analysis wascarried out and identified seven CSFswith 38 elements of the implementa-tion of TQM. These CSFs were train-ing and education, quality data andreporting, management commitment,customer satisfaction orientation, roleof the quality department, communi-cation to improve quality, and con-tinuous improvement.

From the above key existing em-pirical studies of critical success fac-tors of TQM, there is certainly not aclear agreement as to what the “real”factors of TQM are, and there willalways the disagreements as to ‘how tobest cut the TQM cake’ into factors orelements (Samson and Terziovski1999). Nothing that the differencesexist and are generally not major, theauthor prefers to use a well establishedfactor set, the Deming’s 14 points, andmove on to test the correspondence ofthese factors, individually and collec-tively, with world-class company, op-erational excellence and company per-formance measures, which is an addi-tional step from Tamimi’s contribu-tion.

The author’s motivation is basedon wishing to contribute to knowledgeabout ‘What works?’ This means us-ing not only methods similar to thoseof Saraph et al. (1989), Black andPorter (1996), Tamimi (1998), andSamson and Teziovski (1999) but alsousing matched world-class company,operational excellence, and companyperformance, to test linkages of theestablished Deming’s principles tocompany performance through world-class company and operational excel-lence.

Review of Methodology

Population and Sampling

The Indonesia’s oil and gas indus-try is an oil and gas industry consistingof seven independent group compa-nies based on the contract types: State-

212


Owned Company, Production SharingContract (PSC) Onshore, PSC Off-shore, Joint Operating Body (JOB)Onshore, JOB Offshore, TechnicalAssistance Contract (TAC) Onshore,TAC Offshore (based on the new lawof oil and gas number 22/2001). Thisindustry contains 60 oil and gas com-panies and 200 strategic business units(SBUs). This oil and gas industry wasselected because it has been a leader inimplementing progressive quality man-agement strategies in Indonesia (Hakim1996).

The primary unit of analysis forempirical validation is the individualStrategic Business Unit (SBU) level.The SBU organizational structure ischosen in this study because of threereasons. First, the Indonesia’s oil andgas companies have realized that SBUsallow corporate management to del-egate authority for the strategic man-agement of distinct business entities—the SBUs (Hakim 1996; Pearce andRobinson 2005). In addition the SBUsare profit centers, which facilitates ac-curate assessment of profit and loss.The second important reason of choos-ing the SBU as unit of analysis is thatthe advantage of the SBU to meet theincreased coordination and decision-making requirements that result fromincreased diversity and size (Pearceand Robinson 2005). Third, the SBU isthe level of implementation for mostquality management program. Further-more, studies have shown (e.g. Saraphet al. 1989) that quality investmentsvary between plants or SBUs withinthe same firm, indicating that a more

aggregated unit of analysis, such as theparent firm level, would likely obscureimportant differences (Curkovic 2000et al.).

Two thousand and eight hundred(2800) questionnaires were distributedto the participating (49) oil and gascompanies in a qualified sample of 140SBUs. Each qualified sample of 140SBUs from 49 oil and gas companiesreceived 20 questionnaires. The surveywas administered to every level of man-ager (top, middle, low levels or mul-tiple informant) of oil and gas compa-nies at the SBU level. Only responsesthat answered completely on of theresearch constructs were used.

An assessment of non responsebias was made by using the extrapola-tion approach recommended byArmstrong (1979). Tests revealed nosignificant differences between earlyresponders (the first wave of responses;n= 442) and late responders (the sec-ond wave of responses; n= 890) on anyof the constructs.

In terms of number of cases(SBUs), some people argued the mini-mum cases to run structural equationanalysis was 200 and/or ten times thenumber of observed variables in themost complex construct. However, therecent article by Gefen et al. (2000)demonstrated that the required mini-mum sample size was around 100-150cases (SBUs). Thus, this study with140 cases (a qualified of 140 SBUs)was considered appropriate to run theAMOS program (Achjari and Quaddus2004).

213


Choice of Research Method

According to Yin (2003), thereare three conditions that determine theapplicability of certain research strat-egies. The three conditions consist of(1) the type of research question posed,(2) the extent of control an investigatorhas over actual behavioral events, and(3) the degree of focus on contempo-rary as apposed to historical events.Yin suggests that various strategiesare not mutually exclusive, but thatcertain situations exist in which a spe-cific strategy has a distinct advantage.For the case study approach (i.e. asingle industry and/or multiple casestudy) to have a distinct advantage, a“how” and “why” questions (an ex-planatory research approach) shouldbe asked about a contemporary set ofevents over which the investigator haslittle or no control. Smith and Reece(1999) used this criterion in their ar-ticle of the relationship of strategy, fit,productivity, and business performancein a service setting.

The particulars of this study, interms of the condition suggested byYin (2003) and implemented by Smithand Reece (1999), strongly suggest thecase study (a single industry and/ormultiple case study) as the most appro-priate research methodology. A singleindustry was chosen for the study be-cause the restriction permitted the con-trol of several potential confoundingvariables that often differ between in-dustries, including the scope and com-plexity of quality issues (Curkovic etal. 2000). The most important contri-bution of the present investigation (the

causal relationships) is in the analysisof a sample of companies at the strate-gic business units (SBUs) level fromthe same industry (a single industry —oil and gas industry). The advantage ofconcentrating on a single industry isthat the structural equation model(SEM) of the relationships betweenfive research constructs can be morecomplete because unique characteris-tics of the industry can be included(Simpson and Kohers 2002).

Data Collection Methods

Empirical data for the cross-sec-tional study was collected fromIndonesia’s oil and gas companies un-der Directorate of oil and gas’s con-tract system and procedure. Many de-cision contexts using cross-sectionalanalysis compare entities that are “simi-lar” in at least one attribute. The studyconsidered the following alternativeapproaches to define “similar” entities(Foster 1998): similarity on supply-side, similarity on demand-side, simi-larity in capital market attributes, andsimilarity in legal ownership.

Yin (2003) lists six sources ofevidence that can be served as thefocus of data collection for case stud-ies. These sources are documentation,archival records, interviews, direct ob-servations, participant observation, andphysical artifacts. Three of thesesources are used in this study. Struc-tured interviews of a majority of themanagers at each SBU are used for thefive research constructs. Archivalrecords in the form of operating re-ports are used for all five variables.

214


Finally, direct observation is used forthe independent variable (Deming’sprinciple), primarily to corroborate andvalidate the findings from the struc-tured interviews. The use of multipleinformant helps with both the validityand reliability of the study. In addition,a database containing the various fieldnotes, documents, and narratives col-lected over the course of the study wasmaintained to improve reliability(Smith and Reece 1999).

The structured interviews (thequestionnaire development) was basedon a compilation of previous assessingstudies: the Deming’s principle(Tamimi 1995 and 1998), world-classmanufacturing practices (Flynn et al.1999), operational excellence (Chev-ron Texaco 2003), company perfor-mance (Cook and Verma 2002). Thequestionnaire contained five broadtypes of questions. The first sectionasked the managers at the SBU level torespond to a set of fifty quality man-agement practices, synthesized fromDeming’s philosophy of their estab-lishments based on a five-point inter-val scale (1= not at all true; 2= slightlytrue; 3= somewhat true; 4= mostlytrue; and 5= completely true). Thesecond section contained a series ofquestion measuring world-class com-pany practices on five-point scale. Thethird section asked the managers torate their relative reliability, efficiency,health, environmental, and safety di-mensions of operational excellence;also on five-point scale. Sections fourand five contained company perfor-mance dimensions (financial or mon-

etary gain performance and non finan-cial or value gain performance). Reli-ability and convergent validity assess-ment were performed after the surveyhas been accomplished by examiningitem-to-total correlation and employ-ing confirmatory factor analysis, whereseveral items were dropped for furtheranalysis (see e.g. Tamimi 1995 and1998; Flynn et al. 1999).

Definitions andOperationalization of Variables

The five constructs found in Ap-pendix 1 are discussed in the followingsection. Deming’s Principles (DP), wasoperationalized using a set of 50 qual-ity management practices. These fiftyquality management practices can beexplained by Deming’s 14 points (DP1-14). Deming’s 14 points can be sum-marized as follows: (1) create con-stancy of purpose for improvement ofproduct and service; (2) adopt newphilosophy; (3) cease dependence onmass inspection; (4) end the practice ofawarding business on price tag alone;(5) improve constantly and perpetuallythe system of production and service;(6) institute training; (7) institute lead-ership; (8) drive of fear to ask or sug-gest; (9) break down barriers betweenstaff areas; (10) eliminate slogans, ex-hortations, and targets for theworkforce; (11) eliminate numericalquotas; (12) remove barriers to prideartisanship; (13) institute a vigorousprogram of education and training; (14)take action to accomplish the transfor-mation (Goetsch and Davis 2000).Specifically, exploratory factor analy-

215


sis is employed to determine whetherDeming’s 14 points (DP1-14) collec-tively has positive and significant ef-fect on Deming’s Principles. The mea-sures were developed based onTamimi’s studies (1995, 1998).

World-Class Company (WCC),was operationalized using sixty sevenHayes and Wheelwright dimensions.Hayes and Wheelwright (1984) devel-oped their concept of world-classmanufacturing based on six principles(HWP 1-6). Specifically, confirma-tory factor analysis is employed todetermine whether Hayes andWheelwright’s 67 dimensions has posi-tive and significant effect on the sixprinciples of world-class manufactur-ing. The measure was developed byFlynn et.al. 1999. The term world-class company was used because thesefirms were associated with outstand-ing performance in the global oil andgas industry.

Operational Excellence (OE),was operationalized using five dimen-sions of operational excellence. Themeasure was adapted from ChevronTexaco 2003. In order to succed,ChevronTexaco must achieve world-

class performance and exceed the ca-pability of the strongest competitor.To do so, the management of the Chev-ron Texaco company develops theseven dimensions of tracking for suc-cess: quality of work life, world-classin safety, efficiency, health, reliabil-ity, environmental stewardship, speedand agility, partner of choice, qualityof life in the communities, earninggroup and superior long-term stock-holder returns.

Company Performance, was con-ceptualized by Cook and Verma’sstudy (2002) according to two dimen-sions. The first dimension is relating tofinancial or monetary gain or MGP(i.e., profit, market share enhancement,and cost reduction). The second di-mension is relating to non financial orvalue gain or VGP (i.e., product/ser-vice quality enhancement, deliveryperformance, customer and employeesatisfaction, and community develop-ment impacts). Based on Cook andVerma (2002), the study described thecompany performance into two con-structs: value gain performance con-struct and monetary gain performanceconstruct.

Key Research Questions

This study addresses fourteen key research questions that enable us to understand the relationships ofDeming’s principle, world-class company, operational excellence, monetary gain performance, andvalue gain performance.

1. Do Deming’s 14 Points individually and collectively comprise reliable and valid instrument formeasuring Deming’s Principle (DP)?

2. Do Hayes and Wheelwright Practices (HWP) elements individually and collectively comprisereliable and valid instrument for measuring World-Class Company (WCC)?

3. Do Operational Excellence (OE) elements individually and collectively comprise reliable and validinstrument for measuring Operational Excellence (OE)?

216


Hypotheses Development

The research framework (Appen-dix 1) which identifies a nine-stagepath analytic model (structural model)delineating the factors involved in therelationships between five researchconstructs. On the basis of a review ofthe diffusion of distinctive literatures,the researcher posits fourteen quantita-tive-deductive research hypotheses totest the effect of Deming’s Principle onCompany Performance (Monetary GainPerformance and Value Gain Perfor-mance) through World-class Companyand Operational Excellence and to in-vestigate the interrelationships between

Deming’s Principle, World-Class Com-pany, Operational Excellence, andCompany Performance.

The case study (explanatory) re-search hypotheses include two groupsof hypotheses. The first group of hy-pothesis is concerned with the mea-surement analysis of the Deming’s Prin-ciple (DP), World-Class Company(WCC), Operational Excellence (OE),Monetary Gain Performance (MGP),and Value Gain Performance (VGP)and their elements. The second groupof hypothesis deals with the predictivepower of these elements (the structuralanalysis).

4. Do Monetary Gain Performance (MGP) elements individually and collectively comprise reliableand valid instrument for measuring Monetary Gain Performance (MGP)?

5. Do Value Gain Performance (VGP) elements individually and collectively comprise reliable andvalid instrument for measuring Value Gain Performance (VGP)?

6. Does Deming’s Principles (DP) have a direct and significant effect on Monetary Gain Performance(MGP)?

7. Does Deming’s Principles (DP) have a direct and significant effect on Value Gain Performance(VGP)?

8. Does Deming’s Principles (DP) have a direct and significant effect on World-class Company(WCC)?

9. Does Deming’s Principles (DP) have a direct and significant effect on Operational Excellence (OE)?

10. Does Deming’s Principles (DP) have an indirect and significant effect on Monetary GainPerformance (MGP) through its direct effect on World-Class Company (WCC)?

11. Does Deming’s Principles (DP) have an indirect and significant effect on Monetary GainPerformance (MGP) through its direct effect on Operational Excellence (OE)?

12. Does Deming’s Principles (DP) have an indirect and significant effect on Value Gain Performance(VGP) through its direct effect on World-Class Company (WCC)?

13. Does Deming’s Principles (DP) have an indirect and significant effect on Value Gain Performance(VGP) through its direct effect on Operational Excellence (OE)?

14. Does Value Gain Performance (VGP) have a direct and significant effect on Monetary GainPerformance (MGP) for the first scenario of structural analysis?

217


The first group of hypotheses of measurement analysis consists of five hypotheses:

1. H1: Deming’s 14 Points individually and collectively comprise reliable andvalid instrument for measuring Deming’s Principle (DP).

2. H2: Hayes and Wheelwright Practices (HWP) elements individually andcollectively comprise reliable and valid instrument for measuring World-ClassCompany (WCC).

3. H3: Operational Excellence (OE) elements individually and collectively comprisereliable and valid instrument for measuring Operational Excellence (OE).

4. H4: Monetary Gain Performance (MGP) elements individually and collectivelycomprise reliable and valid instrument for measuring Monetary GainPerformance (MGP).

5. H5: Value Gain Performance (VGP) elements individually and collectivelycomprise reliable and valid instrument for measuring Value Gain Performance(VGP).

The second group of hypotheses of structural analysis consists of nine hypotheses:

1. H6: Deming’s Principles (DP) has a direct and significant effect on MonetaryGain Performance (MGP).

2. H7: Deming’s Principles (DP) has a direct and significant effect on Value GainPerformance (VGP).

3. H8: Deming’s Principles (DP) has a direct and significant effect on World-classCompany (WCC).

4. H9: Deming’s Principles (DP) has a direct and significant effect on OperationalExcellence (OE).

5. H10: Deming’s Principles (DP) has an indirect and significant effect onMonetary Gain Performance (MGP) through its direct effect on World-ClassCompany (WCC).

6. H11: Deming’s Principles (DP) has an indirect and significant effect onMonetary Gain Performance (MGP) through its direct effect on OperationalExcellence (OE).

7. H12: Deming’s Principles (DP) has an indirect and significant effect on ValueGain Performance (VGP) through its direct effect on World-Class Company(WCC).

8. H13: Deming’s Principles (DP) has an indirect and significant effect on ValueGain Performance (VGP) through its direct effect on Operational Excellence(OE).

9. H14: Value Gain Performance (VGP) has a direct and significant effect onMonetary Gain Performance (MGP) for the first scenario of structural analysis.

218


Data Analyzes Methods

The data (quantitative) generatedin the preceding stages were analyzedusing factor analysis (confirmatoryfactor analysis or CFA and explor-atory factor analysis or EFA), struc-tural equation modeling or SEM andpath analysis. These statistical meth-ods were selected because they weremost appropriate for this study, givenits objectives.

Factor Analysis is a data reduc-tion technique used to reduce a largenumber of variables to a smaller set ofunderlying factors that summarize theessential information contained in thevariables. Structural equation model-ing (SEM) is a statistical methodologythat take a confirmatory (i.e., hypoth-esis-testing) approach to the analysisof a structural theory bearing on somephenomenon. Typically, this theoryrepresents causal processes that gener-ate observations on multiple variables(Bentler 1988 in Byrne 2001). Pathanalysis is used to refer to the set ofstructural equations representing thepostulated causal and non causal rela-tionships among the variables underconsideration (Land 1969).

The software programs AMOS(Analysis of Moment Structures orAnalysis of Mean and CovarianceStructures) version 4.01 (Arbuckle1999), SPSS 11.0 (Coakes and Stead2003) were used for the quantitativedata analyses. As one of the SEM(Structural Equation Modeling) soft-ware program, AMOS facilitates thespecification process by automatically

incorporating the estimation of vari-ances by default for all independentfactors (Byrne 2001).

Data Screening

The first step in the factor analysisprocess is to explore the characteristicsof the data. It is often useful to be ableto conduct normality and multi-col-linearity. Analysis on subsets of thedata and to make conditional transfor-mations of variable (Coakes and Steed2003). These can be achieved using thenormality and multicollinearity assump-tions and practical considerations un-derlying the application of principleaxis factoring (PAF) and principle com-ponents (PC).

According to Coakes and Stead(2003), the anti-image correlation ma-trix is used to assess the samplingadequacy each variable. The measuresof sampling adequacy are displayed onthe diagonal of the anti-image correla-tion matrix. Variables with a measureof sampling accuracy that falls belowthe acceptable level of 0.5 should beexcluded from the analysis. Bartlett’stest of sphericity and the Kaiser-Meyer-Olkin (KMO) measure of samplingadequacy are both tests that can beused to determine the factorability ofthe matrix as a whole. If Bartlett’s testof sphericity is large and significant,and if the Kaiser-Meyer-Olkin mea-sure is greater than 0.6, thenfactorability is assumed. An examina-tion of the correlation matrix indicatesthat considerable number of correla-tions exceed 0.3 and thus the matrix issuitable for factoring.

219


Interpretation of Results

A total of 1,332 individual usablequestionnaires were returned thusqualified for analysis, representing aneffective response rate of 50,19 per-cent. All 140 SBUs returned question-naires from their high level manager,middle level manager, and low levelmanager. According to Black (1994),the typical response rate for a researchsurvey is of the order of 15-20 percent.

One hundred and thirty one (131)questionnaire items of the relationshipof Deming’s Principle, World-ClassCompany, Operational Excellence,Monetary Gain Performance, andValue Gain Performance. After re-versed scores were adjusted, items rep-

resenting the constructs and dimen-sions were subjected to reliability andvalidity tests. Rigorous statisticalanalysis is required in order to meetprofessional standards of validity andreliability.

a. Reliability and Validity of Mea-sures. Cronbach’s alpha coefficientswere computed to estimate the reli-ability of each scale (observed vari-able or indicator). The Cronbach’salpha of the measures are rangingfrom 0.8119 to 0.9147, which, ac-cording to DeVellis (1991), are re-spectable to very good. Table 2shows the reliability of the mea-sures. Table 3 summarizes the num-ber of items retained of the researchconstructs.

Table 1. Normality and Multicollinearity Statistics Tests

Variable VIF) Correlation Anti- KMO Bartlett's Normally

(Construct) (< 10) Matrix Image Measure of Sphericity Distributed

(>0.3) Corelation Sampling (Large and

Matrix Adequacy Significant)

Diagonal (>0.6)

Value)

(>0.5)

DP 3.347 0.736-0.822 > 0.882 0.903 5919.358; df 28 Standard deviation 0.61

(significant, p<0.000) mean 3.53

WCC 3.269 0.557-0.651 > 0.680 0.931 114895.300; df 8515 Standard deviation 0.44

(significant, p<0.000) Mean 3.03

OE 1.990 0.511-0.648 > 0.791 0.830 3105.893; df 10 Standard deviation 0.71


MGP 1.812 0.387-0.741 > 0.678 0.712 1023.040; df 3 Standard deviation 0.55


VGP 2.108 0.494-0.741 > 0.832 0.864 3187.373; df 15 Standard deviation 0.48


220


Table 2. Reliability Coefficients (Cronbach’s Alpha) of the Constructs

Construct Number of Items Number of Items Cronbach’s

in the Questionnaire Retained Alpha

DP 14 Items 8 Items 0.9147

WCC 6 Items 4 Items 0.8475

OE 5 Items 3 Items 0.9106

MGP 3 Items 3 Items 0.8119

VGP 6 Items 6 Items 0.8702

Table 3. Number of Items Retained of The Research Constructs

Construct’s Name Construct Sub Construct Item CodeItem Code

Deming’s DP1 QMP 1, 2, 3, 4Principles (DP) DP2 QMP 5, 6, 7

DP3 QMP 8, 9, 10DP4 QMP 11, 12, 13DP5 QMP 15, 16, 17DP6 QMP 18, 19, 20, 21DP7 QMP 22, 23, 24, 25DP8 QMP 26, 27, 28, 29

World-class HWP1 Hayes-Wheelwright dimensions:51, 54, 55, 56, 57, 58Company (WCC) HWP2 Hayes-Wheelwright dimensions: 59, 60, 62, 63, 64, 65,

66, 67, 68HWP3 Hayes-Wheelwright dimensions: 72, 73, 74, 75, 76, 78HWP4 Hayes-Wheelwright dimensions: 80, 82, 83, 84, 85, 88,

89, 90, 91

Operational OE1 OE (OE1, OE2, OE3)Excellence (OE) OE2

OE3

Monetary Gain MGP1 MGP (MGP1, MGP2, MGP3)Performance (MGP) MGP2

MGP3

Value Gain VGP1 VGP (VGP1, VGP2, VGP3, VGP4, VGP5, VGP6)Performance (VGP) VGP2

VGP3VGP4VGP5VGP6

221


After the scales had met the neces-sary levels of reliability, the scaleswere assessed for validity. Confir-matory factor analysis was to assessthe validity of each scales, whichconsisted of the retained items ormanifest indicators. All loadings(path coefficients or regressionweights) from a latent construct totheir corresponding manifest indica-tors were significant (critical ratiovalues >1.96). Thus provided evi-dence of convergent validity.

This study also assessed the dis-criminant validity of the latent con-structs. Discriminant validity is thedegree to which two conceptuallysimilar constructs are distinct. Ac-cording to Anderson and Gerbing(1988), when the confidence inter-val of ± two standard errors arounda correlation estimate between twofactors (constructs) does not includethe value 1, that is evidence of dis-criminant validity for the two con-structs. None of the confidence in-tervals in this study included one.

b. Construct Reliability (). The com-posite reliability of each latent con-struct () measures the internal con-sistency of the construct indicators,depicting degree to which they indi-cate the common latent (unobserved)construct. High reliability of mea-sures provides the researcher withgreater confidence that the indi-vidual indicators consistently mea-sure the same measurements (seeTable 5). The threshold value foracceptable reliability is 0.70. (Hairet al. 1998).

c. Fixing the Error Terms and theLambdas. Single indicators mea-sured latent constructs of this study;however, in each case, the indicatorwas a multiple-item scale. It is un-likely that a single indicator per-fectly measures a construct; there-fore, this study estimated the mea-surement error terms. The measure-ment error terms were fixed at (1- )2 and the corresponding lambdas –the loading from a latent construct toits corresponding indicator– were

Table 4. Construct Reliability

Construct

DP (Before deleted components 9-14) 0.0600 0.8190 0.9910

DP (After deleted components 9-14) 0.0270 0.8190 0.9620

WCC (Before deleted components 5 and 6) 0.0217 0.7866 0.9661

WCC ( After deleted components 5 and 6) 0.0379 0.8186 0.9465

OE 0.1387 0.5999 0.7218

MGP 0.0567 0.4976 0.8136

VGP 0.0248 0.4508 0.8912

222


fixed at 1/2 (Howell 1987). Forthe non-latent (observed) variables,the error terms were fixed at 0 andthe corresponding lambdas werefixed at 1.

The measure of this study consistsof indicators five latent variablesmeasured based on a 5 point scale.Therefore, before fixing the errorterms and the lambdas for thesamples, the study converted thoselatent variables into standard scores(Z scores) by subtracting the meanand dividing by the standard devia-tion for each variable. Using stan-dardized variables eliminates the ef-fects due to scale differences (Hair

et.al. 1998). Table 4 provides thereliability of the constructs, lamb-das, and error terms.

d. EFA for Deming’s Principle (DP)Construct. An exploratory principlecomponents factor analysis was con-ducted to determine whether theDeming’s 14 points load collectivelyan overall constructs that may betermed as Deming’s Principles(DP)—hypothesis 1 was supported.All Deming’s 14 points were ex-tracted that accounted for 58.768percent of the total variation in theobserved variable and all of the 14component matrix value above 0.5.All of Deming’s 14 points collec-

Table 5. The Result of Exploratory Factor Analysis: Deming’s Principle

Initial Eigenvalues Extraction Sums Component

of Squared Loading Matrix

% of Comulative Total % of Cumulativre Component 1

Component Total Variance % Variance %

1 8.227 58.768 58.768 8.227 58.768 58.768 0.747

2 0.896 6.402 65.169 0.757

3 0.735 5.247 70.417 0.779

4 0.622 4.443 74.860 0.688

5 0.548 3.911 78.771 0.749

6 0.485 3.463 82.234 0.779

7 0.427 3.052 85.287 0.808

8 0.399 2.853 88.140 0.788

9 0.345 2.465 90.604 0.749

10 0.307 2.191 92.795 0.790

11 0.283 2.024 94.820 0.786

12 0.261 1.864 96.684 0.755

13 0.240 1.717 98.402 0.780

14 0.224 1.598 100.000 0.773

Extraction Method: Principal Component Analysis

1. Component Extracted: only one component was extracted. The solution cannot be rotated

Components 9-14 should be deleted, because the total initial eigenvalues are < 0.4 (Nunnally 1967)

223


tively have positive impact on“Deming’s Principles. Table 5shows the result of Exploratory Fac-tor Analysis: Deming’s principle.Components with a measure of totalinitial Eigen values that falls belowthe acceptable level of 0.4 should beexcluded from the analysis (i.e.,components 9 - 14) (Nunnally 1967)–there are 8 components ofDeming’s principle should be con-sidered in this study (components 1-8).

EFA for World-Class Company(WCC) Construct

An exploratory principle compo-nents factor analysis was conducted todetermine whether the Hayes andWheelwright’s six dimensions (HWP

1-6) load collectively an overall con-structs that may be termed as World-Class Company (WCC). All Hayesand Wheelwright’s six dimensions(HWP 1-6) were extracted that ac-counted for 60.436 percent of the totalvariation in the observed variable andall of the 6 component matrixes valuedabove 0.5. All Hayes and Wheelwright’ssix dimensions (HWP 1-6) collectivelyhave positive impact on World-ClassCompany (WCC) –hypothesis 2 wassupported. Table 6 shows the result ofExploratory Factor Analysis: World-Class Company or WCC. Componentswith a measure of total initial Eigenvalues that falls below the acceptablelevel of 0.4 should be excluded from theanalysis (i.e., components 5 and 6)(Nunnally 1967).

Table 6. The Result of Exploratory Factor Analysis: World-class Company(WCC) Total Variance Explained

Initial Eigenvalues Extraction Sums Component

of Squared Loading Matrix

% of Comulative Total % of Cumulativre Component 1

Component Total Variance % Variance %

1 3.626 60.436 60.436 3.626 60.436 60.436 0.7922 0.710 11.835 72.272 0.8593 0.558 9.298 81.570 0.7014 0.494 8.226 89.796 0.8005 0.317 5.279 95.075 0.8226 0.295 4.925 100.000 0.674

Extraction Method: Principal Component Analysis1. Component Extracted: only one component was extracted. The solution cannot be rotatedComponents 5 and 6 should be deleted, because the total initial Eigen values are < 0.4 (Nunnally1967)

224


CFA for Operational Excellence(OE) Construct

The result of Confirmatory FactorAnalysis (CFA) for Operational Ex-cellence (OE) construct shows thatonly three factors have significant andpositive impact on Operational Excel-lence (OE) —H3 was supported by thethree factors of OE. The Cronbach’salpha for the overall scale as equal to0.893. An examination of the itemcompressing the Operational Excel-lence construct indicates that item fourand five have the highest correcteditems-total correlations. If these itemswere removed from the scale, the Al-pha if Item Deleted column shows asthat overall reliability would increaseslightly. When the researcher removesthese items and recalculated the reli-ability coefficient, Cronbach’ alpha isa raised slightly to 0.911. Therefore,deletion of these items may be consid-ered appropriate.

CFA for Monetary GainPerformance (MGP) Construct

The result of Confirmatory FactorAnalysis (CFA) for Monetary GainPerformance (MGP) construct showsthat the three factors have significantand positive impact on Monetary GainPerformance (MGP)—H4 was sup-ported. The result shows the reliabilityof three dimensions of the MonetaryGain Performance (MGP) constructcomputed by Cronbach’s alpha, variedfrom 0.718 to 0.770, suggesting anacceptable internal consistency, espe-cially given the number of item

(Nunnally 1978). The Cronbach’s al-pha for the overall scale as equal to0.812.

CFA for Value GainPerformance (VGP) Construct

The result of Confirmatory FactorAnalysis (CFA) for Value Gain Perfor-mance (VGP) construct shows that thesix factors have significant and posi-tive impact on Value Gain Performance(VGP) —H5 was supported. Constructscomputed by Cronbach’s alpha, variedfrom 0.831 to 0.865, suggesting anacceptable internal consistency, espe-cially given the number of item(Nunnally 1978). The Cronbach’s al-pha for the overall scale as equal to0.870.

Assessment of Structural ModelFit

Table 7 presented the initial (origi-nal) structural model and the good-ness-of –fit statistics results. The esti-mation of parameters in the model wasdetermined using maximum likelihood(ML) estimation (Bollen 1989; Bentler1990, Joreskog and Sorbom 1989).Eight criteria for accessing for overallfit were used: X2, X2/df, GFI, AGFI,CFI, RMR, RMSEA, and p-value basedon the acceptable or desirable param-eter level (Hair et al. 1998). A sum-mary of selected fit indices for theEQS analysis (initial model) is pro-vided in Table 7. While this model fitsperfectly (GFI= 1, CFI= 1, and RMR=0), but X2/df, AGFI, and p-value arenot available. Interpretation of the con-

225


Table 7. SEM Results (The Initial/Original Model)

Structural StandardizedRelation Regression CR error () Residual ()

(Causal Paths) Weights

(y)

MGP <—— DP -0.048 -1.380 DP = 0.027 WCC = 0.458

VGP <—— DP -0.059 -1.674 WCC = 0.022 OE = 0.771

WCC <—— DP 0.736 36.624 OE = 0.139 MGP = 0.346

OE <—— DP 0.478 15.812 MGP = 0.057 VGP = 0.492

MGP <—— WCC 0.104 2.386 VGP = 0.025

VGP <—— WCC 0.513 12.488

MGP <—— OE 0.177 4.540

VGP <—— OE 0.319 8.730

MGP <—— VGP 0.637 17.523

Acceptable Parameter Level Desirable

Goodness of Fit Measures (Hair et al. 1998) Parameter Level

(Hair et al. 1998)

Chi-Square Statistic (X2 ) 0.000

Degree of Freedom (df) 0

Normed Chi-Square (X2/df) N/A 1 < x < 5 1 < x < 2

GFI 1.000 Close to 1 is better

AGFI N/A > 0.90

CFI 1.000 Close to 1 is better

RMR 0.000 Close to 0 is better

RMSEA 0.471 < 0.10 < 0.05

P N/A > 0.05 > 0.15

ECVI 0.023 < 0.05

fident interval indicates that the authorcan be 90 percent confident that thetrue RMSEA value (0.471) in the popu-lation will greater than 0.10 indicatepoor fit.

The hypothesized (initial) modeldepicted in Appendix 2. The initialstructural model supported hypothesesH8-H14, although the model fit showedthe marginal satisfactory goodness-of-fit indices. Unfortunately, hypoth-eses H6 and H7 were not confirmed

since the paths from Deming’s Prin-ciple (DP) to Monetary Gain Perfor-mance (MGP) and from Deming’sPrinciple (DP) to Value Gain Perfor-mance (VGP) were not significant andnegative effect—the critical ratio val-ues (CR) were negative (these indicatestatistically insignificant). Therefore,the path DP to VGP was eliminatedand the model was revised.

Table 8 shows the direct effectand indirect effects of the one exog-

226


enous variable (DP) and the two inter-vening variables (WCC and OE) oncompany performance (MGP andVGP). This result provided by the pathanalysis do allow for some meaningfulinterpretation and potential contribu-tion. For example the direct effects ofDP on MGP and DP on VGP are nega-tive. However, DP has positive indi-rect and significant effects on MGPand VGP through its positive directeffects on WCC and OE in light of thepositive total effects.

Table 9 shows the revised struc-tural model 1. After eliminating thepath from DP to VGP, the GFI, AGFI,and CFI were iteratively used to deter-mine whether the structural model fit-ted the data well. Taken together, theseresults suggest the hypothesized modelis well-fitting and represents a reason-able approximation to the population(X2/df < 3; GFI, AGFI, CFI > 0.90;RMR and RMSEA < 0.05; and p-value> 0.05). Turning to Table 9, again wesee that the path from DP to MGP is

not significant in this reduced model —CR value was negative. This pathshould be eliminated and further modi-fication was needed to improve modelfit to acceptable levels.

The fully revised structural modelis presented in Table 10. All paths inthe fully revised structural model aresignificant, so no further model reduc-tions were attempted. The final modelsurpasses the hypothesized model onall fit criteria, which confirms that themodifications were meaningful. Thefinal model is tenable from a contentand theoretical standpoint. Althoughthe final model (without the directpaths from DP to MGP and DP toVGP) appears to fit better, the revisedmodel 1 also has adequate fit. The pathanalyses indicate that the DP affectsMGP and VGP only through its directeffects on World-Class Company(WCC) and Operational Excellence(OE).

Most of the overall model fit indi-ces (the goodness-of-fit-index or GFI=

Table 8.The Direct and Indirect Effects of the One Exogenous Variable andthe Intervening Variables on Company Performance

Dependent Predetermined Total Indirect Effects Direct

Variable Variable Effect Via Effect

WCC OE

WCC DP 0.736 - - 0.736

OE DP 0.478 - - 0.478

MGP DP 0.414 0.379 0.083 -0.048

MGP VGP 0.637 - - 0.637

VGP DP 0.471 0.337 0.193 -0.059

t-value significance at the p = 0.000

227


0.998; the goodness-of-fit-index ad-justed for degrees of freedom orAGFI=0.988; the root mean squareerror of approximation or RMSEA=0.034; the root mean square residual orRMR= 0.004; the comparative fit in-dex or CFI= 0.999); and p-value= 0.077thus suggesting that the contendingmodel represented a superior fit to thedata. Because of the goodness-of-fitstatistics resulting from this analysis isa well-fitting model, this model is ac-

cepted. This attempt has improved theNormed Chi-square from 2.802 to 2.569as well as the RMSEA from 0.037 to0.034.

It means that the Indonesia’s oiland gas companies should try to inter-pret of such guidelines to help manag-ers at the SBU level to measure orimplement Deming’s principle. Thus,reducing Deming’s 14 points into asmaller set of meaningful factors (i.e.,eight factors) should aid the Indonesia’s

Table 9. SEM Results (Revised Model 1)

Structural StandardizedRelation Regression CR error () Residual ()

(Causal Paths) Weights

(y)

MGP <—— DP -0.052 -1.532 DP = 0.027 WCC = 0.459

VGP <—— DP (Deleted) - - WCC = 0.022 OE = 0.776

WCC <—— DP 0.735 36.564 OE = 0.139 MGP = 0.346

OE <—— DP 0.473 15.693 MGP = 0.057 VGP = 0.495

MGP <—— WCC 0.108 2.500 VGP = 0.025

VGP <—— WCC 0.470 14.598

MGP <—— OE 0.177 4.549

VGP <—— OE 0.317 8.664

MGP <—— VGP 0.638 17.635



(Hair et al. 1998)

Chi-Square Statistic (X2) 2.802

Degree of Freedom (df) 1.000

Normed Chi-Square (X2/df) 2.802 1 < x < 5 1 < x < 2


AGFI 0.987 > 0.90



RMSEA 0.037 < 0.10 < 0.05

P 0.094 > 0.05 > 0.15

ECVI 0.023 < 0.05

228


oil and gas managers in their“Demingization” efforts. It should bepointed out, however, that by reducingDeming’s 14 points into a smaller set ofquality principles (eight factors), thisstudy is not diminishing the importanceof other factors. Rather, this study istesting whether the 14 points may becollapsed into a smaller set of qualitypractices for easier implementation(Tamimi 1995).

This study also provides impor-tant insights into the consistent andsmallest all ECVI values (ExpectedCross-Validation Index) from the ini-tial model, revised model 1, and finalmodel (ECVI= 0.023). According toByrne (2001) the structural model hav-ing the smallest ECVI values exhibitsthe greatest potential for replication. Inassessing the hypothesized for the struc-tural model we compare its ECVI valueof 0.023 with that of both the saturatedmodel (ECVI= 0.023) and the indepen-dence model (ECVI= 2.234). Given thelower ECVI value for the hypothesizedmodel, compared with both the inde-pendence and saturated models, weconclude that it represents the best fit tothe data.

Contrary to the proposed model,the Deming’s Principle has no directeffects on monetary gain performanceand value gain performance. The hy-pothesized relationship betweenDeming’s Principle and monetary gainperformance and value gain perfor-mance are not supported. It is impor-tant to note that the quality manage-ment practices (Deming’s Principle)does not have its own set of measured

indicators. Rather, it is linked indi-rectly to those measuring the world-class company and operational excel-lence factors. However, the results sug-gest that the effects of Deming’s Prin-ciple on World-Class Company andValue Gain Performance to MonetaryGain Performance have relativelygreater positive effects.

In finalizing this revision on struc-tural model assessment, Table 10 sum-marizes the results of testing each hy-pothesis and the associated causal path(Appendix 3). Appendix 3 is the bestpredictors of company performance(MGP and VGP) based on the good-ness of fit measures (X2/df= 2.569;GFI= 0.998, AGFI= 0.988, CFI=0.999, RMR= 0.004, RMSEA= 0.034and p-value= 0.077), even though theCR value of the path from WCC toMGP d” 1.96 (CR value= 1.943).Therefore, the deletion of this path maynot be considered appropriate. Theprobability level (p-value) cannot beimproved (the p-value would be de-crease tremendously from 0.077 to0.031), because all of good-of-fit sta-tistic represent a well-fitting and a rea-sonable approximation for the popula-tion. In summary, seven of nine causalpaths specified in the hypothesizedmodel were found to be positive andstatistically significant with small er-ror () of five constructs (close to zero)and small residual () of mediating anddependent variables (<2.58).

These results are remarkable, butit does lend some credence to the datacollection and measurement methods.With two modifications, the findings

229


of the research support some degree ofconfidence to the study that usedDeming’s principle as surrogate mea-sure for company performance (mon-etary gain performance and value gainperformance) through world-classcompany and operational excellence.The contending model shown in Ap-pendix 3 should be compared to theinitial model in Appendix 2 and sug-gested that the contending model beaccepted over the initial model, as the

contending (final) model was more par-simonious. Identifying a final modelwithout condition codes or signs ofmisspecification allowed the author totest hypotheses using the theoreticalmodel (Johnson and Greening 1999).

Table 10 and Appendix 3 presentresults of the hypothesis tests. Overall,seven of nine hypotheses (H8-H14)were statistically significant in the pre-dicted direction at a significance levelof less than 0.05 or better. These are

Table 10. SEM Results (Revised Model 2, The Final/Contending Model)

Structural StandardizedHypothesis Relation Regression Hypothesis

(Causal Weights CR error () Residual () Supported?

Paths) (y)

H6 MGP <— DP (Deleted) - - DP = 0.027 WCC= 0.459 No

H7 VGP <— DP (Deleted) - - WCC = 0.022 OE = 0.779 No

H8 WCC <— DP 0.735 36.550 OE = 0.139 MGP = 0.347 Yes

H9 OE <— DP 0.470 15.612 MGP = 0.057 VGP = 0.495 Yes

H10 MGP <— WCC 0.067 1.943 VGP = 0.025 Yes

H11 VGP <— WCC 0.470 14.597 Yes

H12 MGP <— OE 0.173 4.466 Yes

H13 VGP <— OE 0.317 8.664 Yes

H14 MGP <— VGP 0.641 17.727



(Hair et al. 1998)

Chi-Square Statistic (X2) 5.137

Degree of Freedom (df) 2

Normed Chi-Square (X2/df) 2.569 1 < x < 5 1 < x < 2


AGFI 0.988 > 0.90



RMSEA 0.034 < 0.10 < 0.05

P 0.077 > 0.05 > 0.15

ECVI 0.023 < 0.05

230


Table 11. A Complete Model Fit: Initial Vs. Final Model

Model Fit for Acceptable Parameter Level for

Goodness-of- Fit Initial Final The Criteria of Initial Final Causal

Statistics Test Model Causal Hair et al. Model Model

Model 1998

Chi-Square Statistic (X2) 0.000 5.137

Degree of Freedom (df) 0 2

Normed Chi-Square (X2/df) N/A 2.569 1 < x < 5 No Yes

GFI 1.000 0.998 Close to 1 is better Yes Yes

AGFI N/A 0.988 > 0.90 No Yes

CFI 1.000 0.999 Close to 1 is better Yes Yes

RMR 0.000 0.004 Close to 0 is better Yes Yes

RMSEA 0.471 0.034 < 0.05 No Yes

P N/A 0.077 > 0.05 No Yes

ECVI 0.023 0.023 < 0.05 Yes Yes

within Deming’s principle (eight com-ponents): (1) create constancy of pur-pose for improvement of product andservice; (2) adopt new philosophy; (3)cease dependence on mass inspection;(4) end the practice of awarding busi-ness on price tag alone; (5) improveconstantly and perpetually the systemof production and service; (6) institutetraining; (7) institute leadership; and(8) drive of fear to ask or suggest) forcompany performance (monetary gainperformance and value gain perfor-mance) were statistically significantthrough world-class company and op-erational excellence.

Table 11 shows a complete modelfit and the acceptable parameter levelof the research constructs which indi-cate the overall parameter of final orcontending model are good fit between

the hypothesized model and the ob-served data. The author compares themodel fit for original and final, theresults of the final model suggest thatthe hypothesized models are well-fit-ting and represent reasonable approxi-mation to the population. The exami-nations to find these goodness-of-fitstatistics of the final model with respectto the acceptable parameter level haveencouraged the author to seek somemodifications of the hypothesized mod-els in the efforts to attain adequate fit tothe data.

Conclusions andContributions

Given that reliable and validmeasures were needed for the researchconstructs, which are in turn important

231


for theory building, this study hasdeveloped and validated an appropriatemeasurement instrument for the causalrelationships between Deming’sprinciple, world-class company,operational excellence, and companyperformance (monetary gain perfor-mance and value gain performance).The ability of structural equationmodeling (SEM) techniques to assessrelationships comprehensively hasprovided a transition from exploratoryto confirmatory analysis (Bollen 1989in Samson and Terziovski 1999). Thestrong statistically significant for thecausal relationships between Deming’sprinciple, world-class company,operational excellence, and companyperformance (monetary gain perfor-mance and value gain performance)suggest that the findings are quitereliable.

By developing these causal rela-tionships, the managers will be able tomake rational judgments about howthe journey of TQM (Deming’s prin-ciples) should be implemented. Thejourney into TQM will be accom-plished by implementing World-ClassCompany and Operational Excellenceto raise the company performance.Being so, it will be helpful to examinethe perspectives of people at variouslevels, and at various milestones alongthe way. This is presented not so muchas a warning, but more to raise thesensitivity level of the manager of atotal quality effort (Goetsch and Davis2000).

The study is a timely research inIndonesia, so that it is important to

explore and to show the relationship ofworld-class company and operationalexcellence as mediating variables be-tween the independent variable ofDeming’s principle and the dependentvariables of company performances(monetary gain performance and valuegain performance) in an oil and gasindustry setting (a single industry orcase study research). The importanceand the benefit of the research findingsare to support the Indonesia’s oil andgas companies to fulfill their vision tobe recognized as World-Class Com-pany committed to operational excel-lence based on the oil and gas policyreform.

The particular design of the re-search and the findings suggest thatthe structural model of the study has agreat potential for replication to manu-facturing as well as service operations.To do this, future research will need torecognize how service businesses aresimilar and different from manufactur-ing businesses in term of the causalrelationship between Deming’s prin-ciple, world-class company, opera-tional excellence, and company per-formance. This translation of manu-facturing (oil and gas industry) TQM-based systems to service organizationscannot be presumed and would have tobe carefully thought out (Curkovic etal. 2000)

This study also provides severalcontributions to the field of TQM. First,the most contribution of the presentinvestigation is the analysis of a largemultiple informant of firms from thesame industry (at the SBU level). The

232


advantage of concentrating on a singleindustry is that the model fit of struc-tural equation modeling can be morespecific because unique characteristicof the oil and gas industry can beincluded.

Second, this study demonstratesthat Deming’s Principle has significantpositive and indirect effects on com-pany performance through world-classcompany and operational excellence.This result may lend support toDeming’s notion that improvementsfocus on world-class company andoperational excellence, though oftenhard to quantify, are ultimately benefi-cial to the long-term health of the orga-nization (Deming 1986).

Third, this study has utilized casestudy research method (field-based re-search) to examine concepts that seemreasonable, yet lack empirical verifica-tion. The particular design of the re-search and the findings suggest thatmuch of the conceptual work in qualitymanagement practices may be appli-cable to manufacturing as well as ser-vice operation.

Finally, the implication of thesefindings for managers involved in anintegrated oil and gas chain is highlysignificant. As a result of its amend-ment of law of the Republic of Indone-sia concerning oil and natural gas, theIndonesia’s oil and gas industry has aprogram under way to improve refin-ery level management, operations, andmaintenance. The name of this pro-gram is Indonesia oil and gas refineriesaiming for world-class operations. Inaddition the strong statistically sig-

nificant results these relationships(Deming’s Principle and World-ClassCompany, Value Gain Performanceand Monetary Gain Performance) sug-gest that the findings are quite reliable.These findings provide evidence andsupport for other programs to buildupon in trying to ascertain emergingpatterns within the oil and gas field.

Limitation of the Study

The findings and conclusions ofthis study should be interpreted whilekeeping in mind the following limita-tions. It is important to note that thefirst potential limitation of this studystems from the use of a cross sectionalanalysis. Cross sectional analysis onlygive us portrayed at a particular pointof time. The researcher can not exam-ine the dynamic nature of trade-offwhich is changing over time (Silveiraand Slack 2001). In addition the re-searcher encourages to think aboutwhether the model of the study effectsvary over time, either because othertime the constructs are theoreticallyimportant or because the theoreticaleffect is unstable for some reason. Nextresearch should be conducted longitu-dinally to observe the progress of im-provement efforts (i.e., by developingAntecedents, Behavioral, Conse-quences analysis; or by using triangu-lation method).

A second limitation relates to thegeneralizability of the sample of singleindustry (the Indonesia’s oil and gasindustry; five digit of SIC Codes) tothe larger population of wide variety

233


industries (two digit of SIC Codes)employing the successful quality man-agement implementation for World-class Performance in Operations. Toaddress this problem, next researchshould not emphasize on a single in-dustry only, but the companies comefrom a wide ranges of industries (e.g.the classification of responding com-panies based on the two or three digitof SIC Codes) using a longitudinalstudy.

Third, one must be cautious ininterpreting the findings of this studydue to the companies restructuringpolicy into Strategic Business Units(SBUs) was relatively new. The poten-tial problem with respect to the new

policy implementation is a probabilitythat SBUs lack of strategic consensusbetween policy maker (top level man-ager), middle level manager, and lowlevel manager in the upstream, anddownstream of oil and gas chains. As aresult, the research findings are in-tended to represent the types of issuesfaced by strategic business units (SBUs)inexperienced in the implementation ofDeming’s Principle but nonethelesschanged with the necessity of attainingsuccessful TQM practices (Deming’sprinciples) in order to develop world-class company and operational excel-lence while also rising company per-formance.

References

Ahire, S.L., R. Landeros, and D. Y. Golhar. 1995. Total quality management: Aliterature review and an agenda for future research. Journal of Production andOperations Management 4 (3): 277- 306.

Achjari, D., and M. A. Quaddus. 2004. Electronic commerce success model: A searchfor multiple criteria. Gadjah Mada International Journal of Business 6 (1) (January):1-27.

Anderson, J. C., and D. W. Gerbing. 1988. Structural equation modeling in practice: Areview and recommended two-step approach. Psychological Bulletin 103 (3): 411-423.

Anderson, J. C., M. Rungtusanatham, and R. G. Schroeder. 1998. A theory of qualitymanagement underlying the Deming management method. Academy of Manage-ment Review 19 (3): 472-509.

Antony, J., K. Leung, G. Knowles, and S. Gosh. 2002. Critical success factors of TQMimplementation in Hong Kong industries. International Journal of Quality andReliability Management 19 (5): 551-566.

Arbukle, J. L., and W. Wothke. 1999. Amos 4.0. User’s Guide. Chicago: Smallwaters.

Becker, S. W. 1993. TQM does work: Ten reasons why misguided efforts fail. Manage-ment Review 82 (5): 30-34.

Armstrong, J. S. 1979. Advocacy and objectivity in Science. Management ScienceJournal 25 (5) (May): 423-428.

234


Bentler, P. M. 1990. Comparative fit indexes in structural models. Psychological Bulletin107 (2): 238-246.

Black, S. A. and L. J. Porter. 1996. Identification of the critical factors of TQM. DecisionSciences 27 (1): 1-21.

Bollen, K. A. 1989. Structural Equations with Latent Variables. New York: Wiley.

Bollen, K. A. 1989. A new incremental fit index for general structural models.Sociological Methods and Research 17: 303-316.

Brislin, R. W. 1986. The wording and translation of research instruments. In W. J.Looner and J. W. Berry (Eds.) Field Methods in Cross-Cultural Research. BeverlyHills, CA: Sage

Bryman, A., and E. Bell. 2003. Business Research Methods. New York: OxfordUniversity Press, Inc.

Byrne, B. M.2001. Structural Equation Modeling with AMOS. New Jersey, LawrenceErlbaum Associates, Inc., Publishers.

Chevron Texaco. 2003. Operational Excellence: Framework and Process DevelopmentGuidance (January). Jakarta.

Coakes, S. J., and L. H. Steed. 2003. SPPS: Analysis Without Anguish (Version 11.0 forWindows). Australia: John Wiley and Son, Ltd.

Cook, L. S., and R. Verma, R. 2002. Exploring the linkages between quality systems,service quality, and performance excellence: Service providers’ perspectives.Quality management Journal 9 (2). http://www.asq.org/pub/qmj/past/vol9/issue2/index.html.

Crosby, P. B. 1989. Quality is Free. New York: McGraw-Hill.

Curkovic, S., S. Melnyk, R. Calantone, and R. Handfield. 2000. Validating the MalcolmBaldrige National Quality Award framework through structural equation model-ling. International Journal of Production Research 38 (4): 765-791.

Deming, W. E. 1986. Out of the Crisis. Cambridge, MA: MIT Press.

Embassy of the United States of America. 2004. Petroleum Report Indonesia 2002 – 2003(March). Jakarta: USA Embassy.

Eskildson, L. 1994. Improving the odds of TQM’s success. Quality Progress 27 (4): 61-63.

Flynn, B. B., R. G. Schroeder, and E. J. Flynn. 1999. World-class manufacturing: Aninvestigation of Hayes and Wheelwright’s foundation. Journal of Operation Man-agement 17: 249-269.

Foster, G. 1998. Financial Statement Analysis (2nd ed.). New York: Prentice-HallInternational, Inc.

Gefen, D., D. W. Straub, and M. C. Boudreau. 2000. Structural equation modeling andregression: Guidelines for research practice. Communication of the Association forInformation Systems 4 (7): 1-77.

Goetsch, D. L., and S. B. Davis. 2000. Quality Management: Introduction to TotalQuality Management for Production, Processing, and Services (3rd ). Upper SaddleRiver, New Jersey: Prentice Hall.

235


Hackman, J. R., and R. Wageman. 1995. Total quality management: Empirical,conceptual, and practical issues. Administrative Science Quarterly 40 (1): 309-342.

Hair, J. F, Jr., R. E. Anderson, R. L. Tatham, and. W. C. Black. 1998. Multivariate DataAnalysis. (5th ed.). Englewood Cliffs, New Jersey: A Simon and Schuster Company.

Hakim, B. H. 1996. Our bridge to world-class: PT. Caltex Pacific Indonesia’s total qualitymanagement practice. Training for Quality 4 (1): 40-42.

Harari, O. 1993. The eleventh reason why TQM doesn’t work. Management Review 83(5): 31-35.

Harari, O. 1997. Ten reasons TQM doesn’t work. Management Review 86 (1): 37-43.

Howell, R. D. 1987. Covariance structure modeling and measurement issues: A note oninterrelations among a channel entity’s power sources. Journal of MarketingResearch 24 (February): 119-126.

Imai, M. 1986. Kaizen. New York, McGraw-Hill Publishing Company.

Indonesia Business Unit (IBU). 2002. Operational Excellence Implementation: Safety,Environment, Health, Reliability and Efficiency (February). Jakarta, IBU.

Ishikawa, K. 1985. What is Total Quality Control? The Japanese Way. Englewood Cliffs,NJ, Prentice-Hall.

Johnson, R. A., and D. W. Greening. 1999. The effects of corporate governance andinstitutional ownership types on corporate social performance. Academy of Man-agement Journal 42 (5): 564-576.

Joreskog, K. G., and D. Sorbom, D. 1989. LISREL 7 User’s Reference Guide. Chicago:Scientific Software, Inc.

Juran, J. M. 1994. The quality trilogy: A universal approach for managing for quality.In H. Costin (ed.). Total Quality Management. New York: Dryden.

Kolesar, P. J. 1995. Partial quality management: An essay. Journal of Production andOperations Management. 4 (3): 195-200.

Krumwiede, D. W., C. Sheu, and J. Lavelle. 1996. Understanding the relationship of topmanagement personality to TQM implementation. Production and Inventory Man-agement Journal (Second Quarter): 6-10.

Land. 1969. Principles of path analysis. In Borgatta, E., and G. Bohrnstedt (Ed.)Sociological Methodology. San Francisco: Jossey Bass, Inc.

Madu, C. N., and C. Kuei. 1993. Introducing strategic quality management. Long-RangePlanning 26 (6): 121-131.

Morgan, N. A., and N. F. Piercy. 1998. Interaction between marketing and quality at theSBU level: Influences and outcomes. Journal of the Academy of Marketing Science26 (3): 190-208.

Nunnally, J. C. 1967. Psychometric Theory. New York: McGraw-Hill.

Nunally, J. C. 1978. Psychometric Theory. New York: McGraw-Hill.

Pearce II, J. A., and R. B. Robinson, Jr. 2005. Strategic Management: Formulation,Implementation, and Control. (9th Ed.). New York, NY. McGraw-Hill.

236


Russell, L. 1993. Beyond total quality management. Journal of Quality and Participation16 (2): 5-10.

Samson, D., and M. Terziovski. 1999. The relationship between total quality manage-ment practices and operational performance. Journal of Operations Management17: 393-409.

Saraph, J. V., P. G. Benson, and R. G. Schroeder. 1989. An instrument for measuringthe critical factors of quality management. Decision Sciences 20: 810-829.

Shetty, Y. K. 1993. The quest for quality excellence: Lessons from the Malcolm BaldrigeQuality Award. SAM Advanced Management Journal (Spring): 34-40.

Shin, D., J. G. Kalinowski, and G. A. El-Enein. 1998. Critical implementation issues intotal quality management. SAM Advanced Management Journal (Winter): 10-14.

Silveira, G. D., and N. Slack. 2001. Exploring the trade-off concept. InternationalJournal of Operation and Production Management 21 (7): 919-964.

Simpson, W. G., and T. Kohers. 2002. The link between corporate social and financialperformance: Evidance from the banking industry. Journal of Business Ethics 2 (2)(January): 97-109.

Smith, T. M. and J. S. Reece. 1999. The relationship of strategy, fit, productivity, andbusiness performance in a service setting. Journal of Operations Management 17:145-161.

Smith A. W., and J. M. Sibler. 1994. TQM success —or it’s the process, Stupid! Journalof Property Management (September-October ): 12-16.

Tamimi, N. 1995. An empirical investigation of critical TQM factors using exploratoryfactor analysis. International Journal of Production Research 33 (11): 3041-3051.

Tamimi, N. 1998. A second-order factor analysis of critical TQM factors. InternationalJournal of Quality Science 3 (1): 71-79.

Tamimi, N., and M. Gershon. 1995. A tool for assessing TQM practice versus theDeming philosophy. Journal of Production and Inventory Management 36 (1): 27-32.

Tamimi, N., M. Gershon, and S. Currall. 1995. Assessing the psychometric propertiesof Deming’s 14 principles. Quality management Journal 2 (3): 38-52.

Tatikonda, L. U. and R. J. Tatikonda. 1996. Top ten reasons your TQM effort is failingto improve profit. Production and Inventory Management Journal (Third Quarter):5-9.

The Government of Republic of Indonesia. 2001. Law of the Republic of IndonesiaNumber 22 Year 2001. Jakarta.

Usilaner, B. 1992. What’s the bottom line payback for TQM? The Journal for Quality andParticipation 15 (2): 82-90.

Yin, R. K. 2003. Case Study Research: Design and Method. Thousand Oaks, California,Sage Publications, Inc.

237


238


239


Documents

EXPLORING THE LINKAGES BETWEEN DEMING’S PRINCIPLE, …