Full Terms & Conditions of access and use can be found athttps://www.tandfonline.com/action/journalInformation?journalCode=ucis20

Journal of Computer Information Systems

ISSN: 0887-4417 (Print) 2380-2057 (Online) Journal homepage: https://www.tandfonline.com/loi/ucis20

Project Control, Coordination, and Performance inComplex Information Systems Outsourcing

Dong-Gil Ko, Gwanhoo Lee, Mark Keil & Weidong Xia

To cite this article: Dong-Gil Ko, Gwanhoo Lee, Mark Keil & Weidong Xia (2019): Project Control,Coordination, and Performance in Complex Information Systems Outsourcing, Journal of ComputerInformation Systems

To link to this article: https://doi.org/10.1080/08874417.2019.1606687

Published online: 20 May 2019.

Submit your article to this journal

View Crossmark data

Project Control, Coordination, and Performance in Complex Information SystemsOutsourcingDong-Gil Ko a, Gwanhoo Leeb, Mark Keilc, and Weidong Xiad

aUniversity of Cincinnati, Cincinnati, OH, USA; bAmerican University, Washington, DC, USA; cGeorgia State University, Atlanta, GA, USA; dFloridaInternational University, Miami, FL, USA

ABSTRACTInformation systems outsourcing has not only become prevalent but also involves an increasingly widerange of activities beyond the traditional programming, testing, documentation, and localization activ-ities to include complex business-focused project planning, process design, requirements determination,and logical and physical systems designs. Such complex IS outsourcing often requires interactionsamong various stakeholders across different locations and cultures and therefore is much more difficultto manage than the traditional IS outsourcing. Coordination and control have been viewed as especiallyimportant project management strategies for managing project activities; however, little research hasfocused on how they relate to each other and how they collectively affect the performance of complexIS outsourcing projects. In this study, we examine the differential performance effects of outcome andbehavior controls on coordination and show that the effect of outcome control on project performanceis mediated by coordination.

KEYWORDSCoordination; control;project performance;outsourcing; projectmanagement

Introduction

Information Systems (IS) outsourcing refers to the use ofa third-party vendor that provides IS services to the clientfirm. It has become a prevalent business practice for organi-zations of varying sizes and industries. IS outsourcing hasgrown over the past three decades not only in numbers butalso in scope.1 Most IS outsourcing activities in earlier per-iod involved traditional IS development activities such asprogramming, testing, documentation, and localization.2

However, IS outsourcing projects have become more com-plex with the inclusion of “higher-value activities” such asproject planning, process design, requirements determina-tion, and logical and physical systems designs.3,4

Traditional IS outsourcing projects typically involved activ-ities such as programming, testing, and documentation asso-ciated with well-defined client requirements. Over the years, ISoutsourcing projects have becomemore complex as the need forbusiness-based activities such as project planning, processdesign, requirement determination, logical and physical designshas grown so that firms can compete more effectively in theglobal market.5 Accordingly, firms engaged in complex IS out-sourcing activities tend to discover, define, and create clientrequirements. The nature of tasks and relationships betweenthe different stakeholders for complex IS outsourcing projectsare different from those of traditional IS outsourcing projects.6

Therefore, the management of complex IS outsourcing projectsrequires different project controls and coordination than that forthe traditional IS outsourcing projects. In this research, we focuson complex IS outsourcing project control, coordination andperformance relationships.

The many uncertainties that accompany complex IS out-sourcing project activities make it difficult to clearly specifythe detailed terms of the contract. Such low contractspecificity7 makes it difficult for vendors to carry out thesecomplex IS outsourcing project activities cost-effectively andwith certainty. A vendor needs to devote resources to under-stand a client’s needs, then formalize and translate them intorequirements and design documents that will enable thedevelopment of a functional system.8 With low contract spe-cificity, a client organization cannot rely solely on vendorcontracts to ensure project success, thus requiring greaterproject coordination and control.

While IS outsourcing continues to grow, outsourcing pro-jects have experienced high failure rates.9 These failures canbe attributed to “hidden costs” associated with the manage-ment of IS outsourcing projects.10 A national survey11 corro-borates those hidden cost issues and highlights the twogreatest challenges in IS outsourcing: project coordinationand control. It has long been recognized that “software devel-opment is inherently a complex activity that is embedded withinterdependencies…and entails significant coordination”12

and researchers have focused on “identifying the factors thatlead to the choice of controls and not the results that comefrom the use of controls”.13 Project coordination and controlchallenges are more pronounced in complex IS outsourcingpractices due to the greater level of uncertainties associatedwith the crossing of organizational, cultural, and geographicalboundaries, the need to engage in a sense-making process thatinvolves a new setting with a new set of business processes,and the difficulty of managing frequent changes that affect thefinal system. Thus, complex IS outsourcing provides a rich

CONTACT Dong-Gil Ko donggil.ko@uc.edu Operations, Business Analytics, & Information Systems, University of Cincinnati, Cincinnati, OH, USA

JOURNAL OF COMPUTER INFORMATION SYSTEMShttps://doi.org/10.1080/08874417.2019.1606687

© 2019 International Association for Computer Information Systems

context for studying project coordination and control andtheir effects on project performance.

This study makes two significant contributions resultingfrom a two-stage, time-phased survey of 204 complex out-sourcing IS projects. First, we extend prior research on projectcoordination and control by offering a research modelgrounded in project coordination and control theories. Wefocus on project coordination and two formal control modes(i.e., behavioral control and outcome control), theorizing andthen testing their relationships as well as differential effects onperformance of complex IS outsourcing projects. Second, weoffer a glimpse into a long-standing ambiguity in the litera-ture concerning the relationship between control and projectperformance in complex IS outsourcing projects by offeringempirical evidence that project coordination mediates therelationship between project control and project performance.

Literature review and research model

Project coordination

Researchers have recognized project coordination as a usefultheoretical lens and an important research issue for under-standing interdependency challenges that impact IS projectsin general14 and IS outsourcing project in particular.15–17

Unlike insourcing projects, IS outsourcing projects oftencross organizational boundaries, cultural settings, multiplegeographical boundaries and time zones, and thus pose addi-tional challenges. Consequently, project coordinationbecomes a significant challenge as project performanceincreasingly depends on the cooperation of project stake-holders with different professional cultures, work practices,expectations, geographies, and time zones.

In the context of outsourcing arrangements, a smoothcooperation between vendor and client teams ensures thatthe work of all team members contributes towards achievingproject goals.18 Project coordination thus involves knowledgeexchange activities between a vendor team and a client teamto ensure that project team members work towards meetingproject goals – i.e., satisfy functionality requirements, deliverthe system within budget and schedule, and meet qualitystandards. Such project coordination ensures that projectteam members agree to a common understanding of whatthey are analyzing, designing, and building. This suggests thatvendor teams and client teams, especially the respective pro-ject managers, will share knowledge with one another asneeded in order to deliver an information system that satisfiesrequirements and is completed within budget and schedule.Project coordination may be more prominent for managingcomplex IS outsourcing projects because the high uncertaintyassociated with planning, requirements, analysis and designactivities will require frequent knowledge exchanges to com-plete tasks within established project budget and schedule.19

Coordinating IS project activities between clients and ven-dors has been a topic of growing interest for both scholars andpractitioners, and the effect of project coordination on projectperformance has been proposed in the IS literature.20

Sabherwal and Becerra-Fernandez21 found that project coor-dination played an important role in bridging communication

gaps and facilitating the management of day-to-day activities.Gopal et al.22 found that coordination in a global softwaredelivery context had a positive impact on software develop-ment performance. However, Srikanth and Puranam’s23

examination of 60 global software projects provides evidencethat coordinating project activities across firm boundaries isprone to failure. Inconsistent findings observed in this areamay be due to a focus on project coordination involvingtraditional IS outsourcing project activities rather than com-plex IS outsourcing project activities.16 Therefore, additionalresearch on the effects of project coordination on projectperformance with an added focus on complex IS outsourcingproject activities could shed light on the inconsistent empiri-cal findings.

Project control

Client project managers are responsible for ensuring that theirprojects are progressing in conformance with the proposedbudget, schedule, and objectives. They can assume leadershipof IS projects through the exercise of control in dealing withthe vendor project team. Project control refers to the set ofmechanisms organizations use to motivate individuals towork in such a way that desired objectives are achieved.24,25

Project control has been recognized as important for mana-ging projects.26–28

According to control and agency theories29,30, there aretwo modes of formal control available to managers to moti-vate and guide the activities of vendors: outcome and beha-vior controls. Outcome control involves the definition ofspecific desired task outputs. A manager’s focus when usingoutcome control is on defining appropriate targets, allowingvendors to decide how to meet those output targets. Thus,outcome controls are used to evaluate performance based onthe extent to which the targets are met, rather than focusingon the processes used to achieve the targets.

In contrast, behavior control involves defining appropriatesteps and procedures for vendors to follow as they performtheir tasks. Thus, behavior control is used to evaluate perfor-mance based on the extent to which the vendor adheres to theprescribed procedures. Unlike outcome control, the under-lying assumption behind the exercise of behavior control isthat following prescribed steps and procedures will lead tofavorable project performance. Thus, exercising behavior con-trol minimizes deviation from the prescribed process andshould therefore maximize the probability that the projectwill lead to a favorable outcome.

Project control literature suggests that outcome controlaffects project performance. Henderson and Lee’s26 study ofIS design teams found that outcome control correlated posi-tively with a team’s performance. Nidumolu and Subramani28

showed that a standard set of performance criteria improvedproject performance in software development. Gopal andGosain’s31 study of IS outsourcing projects found that out-come control improved project performance. However,Tiwana and Keil32 found that outcome control did notenhance project performance for IS outsourcing projects.

Prior research has found a significant relationship betweenbehavior control and project performance for insourcing

2 KO ET AL.

software development (e.g.,33) However, this relationship was notfound to be significant when examining outsourcing softwareprojects from either the client’s32 or the vendor’s31 perspective.These mixed findings highlight a need to gain a better under-standing of project control and its impact on project performance,particularly in the context of complex IS outsourcing projectswhere high failure rates have been observed.

Research model

Our research model (Figure 1) proposes that outcome andbehavior controls affect performance and these relationshipsare partially mediated by project coordination. It has beendocumented in theories of organizational behavior that inte-gration and accomplishment of tasks or processes dependmainly on two constructs: coordination and control.34

Though scholars tend to examine either project coordination(e.g.,15,18) or project control (e.g.,12,35) in isolation from oneanother, prior literature suggests that project coordinationand control are interrelated in that control can support orundermine coordination36, but they represent two distinctconcepts.21,37 In software development projects, both projectcoordination and project control have been recognized as twosignificant and distinct activities (e.g.,38)

In the context of IS outsourcing projects, project control is themechanism through which the client ensures that the vendorperforms according to the agreed-upon specification of beha-viors and outcomes. Outsourcing projects are contract-basedengagements in which prescribed behaviors and outcome deli-verables are defined based on agreements between the client andthe vendor about how goals should be specified and measured,and how performance should be monitored. In contrast, projectcoordination is the exchange and communication of projectinformation between vendor and client teams to distributework among project members and to integrate dependent activ-ities aimed atmeeting project goals – a definition consistent witha summary of research on coordination which embodies (1)people working collectively, (2) interdependent tasks, and (3)achievement of a goal.39

Prior research has shown that control mechanisms areeffective for managing increased information processingrequirements40 and task interdependence.41 Given that thereoften are variances between actual task progress and originalproject schedule as well as potential misunderstandings

regarding task interdependencies, managers frequently relyon outcome and behavior controls to facilitate necessaryinformation gathering and to ensure that output targets aremet and prescribed procedures are followed. By managingand evaluating project deliverables and adherence to proce-dures, outcome and behavior controls would impact projectperformance. However, control should also affect projectcoordination. While outcome control provides informationnecessary to manage task interdependencies, behavior controlfacilitates the development of a shared mental model for taskprocedures, which enables the constant evaluation of projectgoals and compliance to procedures. The effects of outcomeand behavior control are particularly pronounced in complexIS outsourcing project where task uncertainties force signifi-cant search process to produce the information needed toexecute tasks.42

It is well documented that the management of interdepen-dent tasks in software development occurs through routineevaluation of project requirements and goals (i.e., projectcoordination) which effectively influences project perfor-mance. Indeed, the aim of project coordination is to improveperformance.43 Cray44 contends that “[t]he pattern of coordi-nation can be imposed through an act of control, but theresulting responsibilities are rooted in coordination.” Thus,we argue on the basis of prior research that outcome andbehavior controls are positively associated with project per-formance and project coordination, and that project coordi-nation mediates the relationships between outcome andbehavior controls and project performance.

Since the use of control mechanisms has been proposed asan effective approach for managing increased task interdepen-dencies and behavioral challenges45, clients rely on controlmechanisms to influence project coordination and perfor-mance. The use of outcome control requires a vendor toreport project progress to a client based on agreed-upon out-come measures. Consequently, a client becomes aware of theproject status which gives him/her the opportunity to evaluatewhether interdependent tasks are carried out properly and,therefore, is able to take corrective measures if necessary.Working together and evaluating task interdependencies aidsin making progress towards meeting project goals. For exam-ple, if the start of a task is dependent on the completion ofthree predecessor tasks, information about the status of thepredecessor tasks enables a client to manage interdependent

OutcomeControl

Control VariablesProject Size

Vendor Reputation

ProjectPerformance

ProjectCoordination

BehaviorControl

Figure 1. Research model.

JOURNAL OF COMPUTER INFORMATION SYSTEMS 3

activities to ensure that the successive task is on target and ontrajectory towards meeting project goals. Therefore,

Hypothesis 1: Outcome control positively impacts projectperformance, and project coordination mediates thisrelationship.

On the other hand, a client's use of behavior controlsuggests that a vendor is obligated to share informationregarding how they are conforming to predefined procedures.Such information and subsequent corrective actions force thevendor to maintain consistent and uniform processes throughthe development of a shared mental model for taskprocedures.46 Shared understanding among software teammembers about the process of completing a task improvescoordination activities and helps the team to meet projectspecifications.47 For example, if a time-boxed approach toSCRUM is the preferred development methodology asopposed to the traditional Waterfall approach, a client canredirect a vendor through education and development ofa common shared understanding should a vendor’s behaviorsdeviate from the iterative and incremental process prescribedby a client. Coordination of these activities is important toensure that project goals are achieved which impacts overallperformance.

Hypothesis 2: Behavior control positively impacts projectperformance, and project coordination mediates thisrelationship.

Research methodology

Data used for testing the hypotheses was collected througha field study based on surveys of managers of outsourced ISprojects. To reduce threats from common methods bias/var-iance, two separate, time-phased, web-based surveys wereadministered.48,49 In the first survey, IS project managerswere asked to provide responses about the antecedents (pro-ject coordination and project control). Two weeks followingcompletion of the initial survey, the same IS project managersresponded to a second survey regarding project performancemeasures.

Study sample

We surveyed the members of the Information SystemsSpecific Interest Group of the Project Management Institute(PMI-ISSIG), an international organization of IS project man-agers. To encourage participation, survey participants wereentered into a drawing for a 33% chance to receive a $25gift certificate of their choice selected from www.giftcertificates.com. Two reminders were sent two weeks apart afterthe initial PMI-ISSIG–sponsored e-mail was sent; similarly,two reminders were sent two weeks apart after the link tothe second questionnaire was sent out to only those whoresponded to the first survey. Email addresses and projecttitles were captured from those participating in the first

survey so that we could reliably send the second surveydirectly to the respondent with a reference to the project title.

Participation and project selection were based on threecriteria. First, a portion or all of the IS project work had tobe outsourced to a vendor. Second, outsourced IS projects hadto have been completed during the last 12 months.Completion was considered important so that the respondentscould respond to project performance questions. Twelvemonths was considered important to minimize recall biasand to ensure that respondents could accurately respond tothe survey questions. Third, the completion of both question-naires was required. This requirement also qualified respon-dents to participate in the drawing for a gift certificate.

For each project, the IS project manager was defined as themanager from the client’s organization responsible forthe day-to-day management of the IS outsourcing project.Among the 245 first questionnaires received, 204 respondents(representing a total of 204 firms) completed the second sur-vey (83.7% response rate).

The industry distribution of the 204 organizations is shownin Table 1. The size of the 204 projects varied in terms ofproject cost, duration, and project team size (see Table 2).Over 80% of projects in the sample involved packaged soft-ware implementation (33%), new software development(25%), or a major enhancement of existing software (24%).The primary development methodologies used were waterfall(N = 101), prototyping (N = 27), and object-oriented (N = 19)approaches with 128 of the projects focusing on Accounting/Finance, Sales, and Operations (distribution/logistics).

Among the 204 completed surveys, 198 respondents indi-cated the project lifecycle stage at which they outsourced their

Table 1. Participating organizations.

Industry No. of Firms

Agriculture 5Banking & Finance 32Communications 25Computers & Software 32Consulting 24Insurance 21Manufacturing 39Medicine & Health 19Oil & Petroleum 12Publishing & News 6Real Estate 3Restaurant & Hotel 4Transportation 18Utilities 10Wholesale & Retail 12Other 33TOTAL: 262a

a Organization conducts business in multiple industries.

Table 2. Sample demographics.

Range Median MeanStandardDeviation Nb

Project Duration(months)

1–75 11.0 14.44 11.23 187

Project Budget($)

5,000–650,000,000 590,000 6,633,489 49,457,272 175

IS TeamMembers

2–180 13.0 21.13 25.73 190

b Number of respondents.

4 KO ET AL.

project activities. Since our study focuses on complex IS out-sourcing projects, our final analysis was based on 189 out-sourced projects that involved both business-based andtechnology-based activities – i.e., Project Selection,Initiation, and Planning; Requirements Determination;Analysis and Design; Coding and Testing; Installation andTraining/Documentation; and Maintenance.

Job titles of the respondents included Project Manager,Technical Manager, and Project Director. These respondentshad 3 to 40 years of experience in IS with an average of 18.32years (standard deviation of 7.46 years). The number of yearsin their current positions (at their current organizations)ranged from 1 to 27, averaging 4.85 years (standard deviationof 4.11 years).

Survey development

Existing instruments were used or adapted to measure con-structs wherever possible; scales for the dependent and theindependent variables were derived from earlier work. Scaleitems and literature sources are shown in the appendix. Theconceptualization and measurement of project performancein this study is consistent with prior project managementliterature investigating outsourced IS project performance,and includes whether the project was completed within sche-dule and budget, and to the client’s satisfaction.8,22,28

Measures of behavior and outcome control were based onscales described in Kirsch25,38 and colleagues.27

Some measurement items for coordination have been pro-posed and used by prior researchers50,51; however, there is noagreed-upon, standard measurement for project coordination.Thus, in accordance with the literature, scale items for projectcoordination were adapted from the literature to capture thenotion of management of interdependencies between activ-ities that directly contribute to project goals – namely budget,schedule, and scope.21 The items focus on measuring theextent to which the client team and the vendor team arecoordinated with respect to project goals.

Consistent with prior IS outsourcing project research, pro-ject size, and vendor reputation are included in the model toaccount for the use of formal controls and project coordina-tion. Project size is commonly used as a control variable whenexamining project performance, and is measured by the totalnumber of client and vendor IS professionals working on theproject.31 We also included vendor reputation as a control.We interpreted vendors’ reputation and proven track recordto mean that they established themselves as better-than-average performers, and such reputation should influenceproject performance (e.g.,52)

We used a sorting procedure to qualitatively assess the facevalidity and the construct validity of the initial measurementitems for our constructs.53 Eight doctoral students enrolled inthe IS program of a large U.S. university participated in thesorting procedure as judges. Each item in the initial measure-ment pool was printed on a 3 × 5-in. index card. In thesorting procedure, each judge was asked to carefully readthe card and place it in one of the constructs used in thisstudy. An additional category, “too ambiguous/unclear,” wasincluded for the judges to put a card into if they felt it did notbelong to any of the predefined constructs. The overall hitratio was 79.2%, indicating that the judges placed most of theitems with their intended constructs. Based on the results andthe feedback from the judges, we changed the wording of twoitems designed to measure behavior control.

Instrument validation

Cronbach’s alpha was used as an indicator of internalconsistency.54 Nunnally55 suggests 0.70 as an acceptable cut-off for reliability. Each item was measured on a scale of one toseven. As shown in Table 3, all of the scales had Cronbach’salphas above the cut-off value of 0.70, suggesting that they arereliable.

To assess convergent and discriminant validity, we per-formed factor analysis. One indication of convergent validityis that all items of a scale load together on one factor; withloadings above 0.40. An indication of discriminant validity isrelatively low cross-loadings of items on other factors. Thisstudy exceeds the factor analysis requirement of a sample size-to-number of items ratio of 10:155 A Principal ComponentsAnalysis (PCA) with Varimax Rotation (with KaiserNormalization) technique was used, and components witheigenvalues greater than one were extracted.

Table 4 displays the results of the factor analysis; fivefactors were extracted. The six items measuring outcomecontrol did not load cleanly together; in particular, twoitems (1 and 3) had high cross-loadings and were judged tobe problematic. Similarly, the five items measuring behaviorcontrol did not load cleanly together; in particular, two items(4 and 5) had high cross-loadings and were judged to beproblematic. Therefore, these items were excluded from thedata analyses. All other items for other constructs loadedcleanly as intended without significant cross-loadings. Afterexcluding the four problematic items, all scales demonstratedboth convergent and discriminant validity. Cronbach’s alphafor the revised behavior and outcome control scales weresatisfactory (0.780 and 0.701, respectively). After completingreliability and validity tests, 189 completed matched-pair sur-veys were used in the final analysis.

Table 3. Reliability and descriptive statistics.

Construct $ of items N Cronbach’s Alpha Range Mean Standard Deviation

Project Performance 6 189 0.937 1.00–7.00 4.45 1.73Project Coordination 3 189 0.852 1.00–7.00 5.54 1.24Outcome Control 4 186 0.701 1.75–7.00 5.27 1.10Behavior Control 3 189 0.780 1.00–7.00 4.98 1.39Vendor Reputation 2 189 0.877 1.00–7.00 5.17 1.40

JOURNAL OF COMPUTER INFORMATION SYSTEMS 5

Analysis

Correlations among the variables included in the researchmodel are shown in Table 5. The results show a significantcorrelation between Project Coordination and ProjectPerformance as well as a significant correlation betweenOutcome Control and Project Performance. OutcomeControl also correlated significantly with Behavior Controland Project Coordination. Behavior Control, however, wasnot significantly correlated with Project Performance.

Before proceeding to test our research model, several testswere conducted to check the assumptions behind regressionmodeling. The first test assessed whether or not the errorterms were normally distributed. For the dependent variable,residuals were plotted against their expected values. Each plotproduced a reasonably straight line, suggesting that thisassumption holds true.56 A second test checked for outliers.Scatter plots of each independent variable against each depen-dent variable were generated and examined. There were nooutliers. Therefore, all data points were retained and weproceeded to test our research model.

In order to test our research model, we conducted a series ofregressions using the Hayes and Preacher57 approach for media-tion analysis. Specifically, we used the MEDIATEmacro for SPSSwhich was obtained from Hayes’ website (http://www.afhayes.com/spss-sas-and-mplus-macros-and-code.html). MEDIATE

was chosen for two reasons: (1) it has the ability to handlemultipleindependent variables simultaneously (and our research modelhas two independent variables), and (2) it uses a bootstrappingapproach for testing the significance of indirect effects.

We chose the bootstrapping approach58,59 for testing thesignificance of indirect effects over the more traditionalSobel’s test of indirect effects. The bootstrapping approach isviewed as superior because Sobel’s test assumes that the sam-pling distribution of the indirect effect (ab) is normal, when infact it is often skewed leading to biased estimates.Bootstrapping overcomes this problem. Bootstrappinginvolves repeatedly sampling with replacement from the data-set and estimating the indirect effect in each resampled data-set. By resampling thousands of times, the samplingdistribution for ab can be approximated and used to constructconfidence intervals for the indirect effect. In our analysis, weset the number of samples to 10,000.

Following the approach described above, we conducted threeregressions: (1) the independent variables (Behavior Controland Outcome Control) and control variables (Project Size andVendor Reputation) predicting the dependent variable (ProjectPerformance), (2) the independent variables and control vari-ables predicting the mediator (Project Coordination), and (3)the independent variables (IVs), the mediator, and the controlvariables predicting the dependent variable (DV). Mediation isestablished when the IVs are found to have significant effects onthe DV and the mediator, respectively, in the first two regres-sion equations. Further, if the IVs have significant effects on theDV in the presence of the mediator in the third regressionequation, and the effects of the IVs on the DV are less in thethird equation than in the first, then partial mediation is indi-cated (if these effects become insignificant in the third equation,full mediation is indicated). Table 6 shows the results of ourmediation analysis. Tolerance and variance inflation factorscores indicated no multicollinearity problems.

Results

As indicated in Table 6, outcome control had a significanteffect on both project performance (Model 1) and projectcoordination (Model 2), while behavior control did not havea significant effect on either. Project coordination hada significant effect on project performance (Model 3), andthe effect of outcome control on project performance wassignificant even in the presence of the mediator (projectcoordination), though its effect was weakened relative to thedirect effect observed in Model 1. These results suggest thatthe effect of outcome control on project performance is par-tially mediated by project coordination.

In order to test the significance of the indirect effects, weexamined bias-corrected bootstrap confidence intervals.Consistent with the results of Model 1 and Model 2, wefound that there was no statistically significant indirect effectof behavior control on project performance as the 95% con-fidence interval constructed for the indirect effect (−0.078,0.029) contained zero. The indirect effect of outcome controlon project performance, however, was 0.210, and this effectwas found to be statistically significant as the 95% confidenceinterval (0.109, 0.337) did not contain zero.

Table 4. Factor analysis.

Component

1 2 3 4 5

Project Performance1 .789 .189 .210 .058 .115Project Performance2 .795 .180 .131 .062 −.016Project Performance3 .818 .180 .131 .038 .111Project Performance4 .855 .181 .103 −.022 .187Project Performance5 .845 .171 .142 −.086 .253Project Performance6 .852 .142 .137 −.007 .314Project Coordination1 .218 .823 .089 −.006 .176Project Coordination2 .345 .761 .313 −.012 .005Project Coordination3 .247 .788 .171 .070 .144Outcome Control2 .096 .277 .669 .213 −.081Outcome Control4 .174 .311 .580 .191 .128Outcome Control5 .167 .246 .783 .043 .138Outcome Control6 .181 −.083 .699 .070 .086Behavior Control1 −.063 .080 −.096 .831 −.048Behavior Control2 −.018 .024 .266 .824 .130Behavior Control3 .117 −.056 .324 .771 .066Vendor Reputation1 .266 .168 .114 .059 .901Vendor Reputation2 .383 .132 .091 .079 .823Eigenvalue 6.976 2.486 1.547 1.211 1.051% Variance Explained 38.753 13.812 8.594 6.730 5.841

(1) Project Performance(2) Project Coordination(3) Outcome Control(4) Behavior Control(5) Vendor Reputation

Table 5. Correlations.

1 2 3 4 5 6

1. Project Performance 12. Project Coordination 0.517** 13. Outcome Control 0.410** 0.470** 14. Behavior Control 0.068 0.109 0.359** 15. Vendor Reputation 0.517** 0.373** 0.304** 0.154* 16. Project Size 0.075 −0.012 0.132 0.149* 0.102 1

*p < 0.05 **p < 0.01

6 KO ET AL.

The results provide partial support for the proposedhypotheses (see Table 6). Specifically, Hypothesis 1 was sup-ported; project coordination was found to partially mediatethe relationship between outcome control and project perfor-mance. The indirect effect of outcome control on projectperformance was 0.210 while the direct effect in the presenceof the mediator was 0.269. Based on our results, 44% of thetotal effect of outcome control on project performance wascarried indirectly through project coordination. Hypothesis 2was not supported; project coordination did not mediate theeffect of behavior control on project performance. Indeed,based on our findings, behavior control had no significanteffect on either project coordination or project performance.One of our control variables, vendor reputation, was found tobe significant, suggesting that when a credible vendor isinvolved in the project, project performance increases.

Discussions and implications

While project coordination and control have been viewed byprior studies as two important project management constructsthat can affect project performance, little is known about howproject coordination and control are related to each other andhow they collectively affect project performance especially forcomplex IS outsourcing projects. As IS project planning,requirements determination, and logical and physical systemsdesigns activities continue to be outsourced, managing suchcomplex IS outsourcing projects have become increasinglydifficult. There is a critical need for empirical research thatprovides insights and practical guidelines to the organizationon how to manage the high level of uncertainty involved incomplex IS outsourcing projects.

Our research supports a mediating role of project coordi-nation on the effect of outcome control on project perfor-mance. This finding extends the traditional view of outcomecontrol only as a direct predictor of project performance bydemonstrating an additional path through which outcomecontrol affects project performance. It also suggests that pro-ject coordination and control are related to each otherthrough a mediating path. Revealing the link between thetwo constructs is an important contribution to the extanttheories of project coordination and control. In the contextof project activities wrought with uncertainties, by constantlyevaluating progress and performance against the project’s

schedule, budget, and scope targets, the project team detectsperformance gaps and recognizes issues and problems thatneed to be resolved. As a result, outcome control facilitatesproject coordination to ensure that the project stays on tracktowards successful completion.

However, our research does not support a mediating role ofproject coordination on the effect of behavior control on projectperformance. One plausible explanation is that behavior controlincreases the conformance of the project team’s behavior to pre-defined procedures and rules, but it does not necessarily facilitatecoordination activities regarding project budget, schedule, andscope, especially in the context of IS outsourcing projects.Whereas behavior control reduces the variance in team behavior,it does not necessarily increase team members’ awareness ofproject goals. Taken together, our research suggests that outcomecontrol and behavior control do not affect project coordination inthe same way, and thus they play different roles in managing ISoutsourcing projects. We suggest that these different roles aremore pronounced in complex IS outsourcing projects because ofthe low contract specificity for the types of activities involved. Thisstudy advances our understanding of the role of outcome control,behavior control, and project coordination in project performanceand suggests more complex relationships among them than priorstudies have revealed.

Although the positive effect of project control on IS projectperformance seems to be generally accepted, much of theproject control literature has focused on IS insourcing pro-jects. As a result, the effect of project control on IS outsour-cing project performance is not well known (e.g.,60) Ourresearch suggests that outcome control has a direct positiveeffect on IS outsourcing project performance, whereas beha-vior control has no significant effect on project performance.Among the few studies examining outcome control and pro-ject performance, the results of our study are consistent withthe main findings which demonstrate a positive and signifi-cant relationship between outcome control and project per-formance. The direct effect between outcome control andproject performance is consistent with Gopal and Gosain31

while the indirect effect found in our study may explain theinsignificant result reported by Tiwana and Keil.32 Moreover,it is possible that our results differ from those of Tiwana andKeil32 given our focus on complex IS outsourcing projects thatinvolved activities where coordination plays an important rolein managing interdependencies across organizational

Table 6. Mediation analysis.

Variables

Model 1: Independent Variables to DependentVariable

b std. error

Model 2: IndependentVariables toMediatorb std. error

Model 3: Independent Variables and Mediator toDependent Variable

b std. error

Constant −0.443 0.577 2.147*** 0.440 −1.331* 0.583Project Size 0.001 0.004 −0.004 0.003 0.003 0.004Vendor Reputation 0.549*** 0.078 0.228*** 0.058 0.455*** 0.075Outcome Control 0.513*** 0.102 0.492*** 0.078 0.310** 0.107Behavior Control −0.130 0.077 −0.058 0.059 −0.106 0.074Project Coordination 0.413*** 0.092Adjusted R2 0.359 0.284 0.420F-statistic 27.32*** 19.65*** 28.18***Degrees of freedom 4, 184 4, 184 5, 183

N = 189 *p < 0.05 **p < 0.01 ***p < 0.001

JOURNAL OF COMPUTER INFORMATION SYSTEMS 7

boundaries. We offer an explanation that about half of thetotal effects of outcome control on project performance wascarried indirectly through project coordination.

In contrast, we did not find support for a direct effect ofbehavior control on project performance. Some prior studiesalso have not found a significant effect of behavior control onproject performance.31,32 One plausible explanation is thatwhen contract specificity is low, there exists ambiguity regard-ing what vendors need to adhere to.7 It could also be thatappropriate evaluation information may not be available, pro-ject management methodology may differ often causing thevendor to execute their own approach, or behaviors may notbe observable.7 Though we believe that these challenges couldbe mitigated through coordinating various project activities,our study shows that the indirect effect of behavior control onproject performance was not significant.

Taken together, our research suggests that outcome con-trol is effective for complex IS outsourcing projects.Importantly, when exercising outcome control, managersshould take care to manage project interdependencies asnearly half of the project performance is attributed to theeffectiveness of coordinating project activities. Withouteffective project coordination, the indirect effect of out-come control on project performance becomes weaker.Our results suggest that project managers should notblindly use project control methods; rather, they shoulduse appropriate resources to implement outcome controland manage project coordination to improve their projectperformance. Unlike IS insourcing projects where behaviorcontrol has been found to impact project performance, theresults from this study suggest that managers do not benefitfrom exerting efforts to exercise behavior control in con-temporary IS outsourcing projects.

Our study has limitations. First, the findings from ourstudy may only reasonably generalize to the IS outsourcingproject domain. The applicability of our results in other set-tings involving IS projects should be empirically assessed.The second limitation is that our research model did notinclude antecedents or moderators. We proposed and testedan uncomplicated research model given that our objective wasto better understand the relationship between project coordi-nation and control and their effects on project performance.Future studies are needed to extend our research model toinclude conditions such as behavior observability, outcomemeasurability, and knowledge of the transformation processthat facilitate or moderate the findings offered in this study.Finally, our research model does not include informal control(clan and self) though we recognize its role in project controlliterature. Clan control is a team-based35 and self-control anindividual-based25 phenomena, and our interest revolvedaround controls where motivation is induced more fromformal specifications as opposed to shared values and normsor self-regulated behaviors. Moreover, clan control is typicallyexamined in a context where a clan (i.e., a team) is collocatedallowing a project manager to influence their behavior.Nevertheless, more studies including informal controls andtheir relationships with project coordination and project per-formance are needed.

Conclusion

The purpose of this study was to enhance our understandingof the two key challenges faced by IS outsourcing projects:project coordination and control. Using project coordinationand control theories along with IS outsourcing and informa-tion processing literatures, we developed a model that speci-fies the mediating role of project coordination in therelationship between project control and project performance.The results suggest that the effect of outcome control onproject performance is partially mediated by project coordi-nation whereas behavior control had no effect.

Our findings provide empirical evidence about the rela-tionship between project coordination and control, and theirrelationship to project performance. The results also shedlight regarding the mixed findings reported in previous stu-dies related to project control. Finally, this study adds to thebody of evidence about the importance of project coordina-tion and control to explain project performance. The researchcommunity continues to examine project coordination andcontrol in the IS outsourcing context as firms continue to seekto improve the project success rate. We hope our researchserves as a stepping stone for furthering our understanding ofhow organizations can better utilize project coordination andcontrol mechanisms to manage the increasingly complex ISoutsourcing projects.

ORCID

Dong-Gil Ko http://orcid.org/0000-0002-1967-0333

References

1. Solli-Sæther H, Gottschalk P. Stages-of-growth in outsourcing,offshoring and backsourcing: back to the future? J Comput InfSyst. 2015;55(2):88–94. doi:10.1080/08874417.2015.11645760.

2. Earl M. The risks of outsourcing IT. MIT Sloan Manage Rev.1996;37(3):26–32. Spring

3. Sadowski M A bit of history: outsourced development and off-shore development 2013 [cited 2018 December 10]. https://cayugasoft.com/bit-history-outsourced-development-offshore-development/

4. Fish KE, Seydel J. Where it outsourcing is and where it is going:a study across functions and department sizes. J Comput Inf Syst.2006;46:96–103.

5. Deloitte. Outsourcing amid complexity 2013. https://www2.deloitte .com/content/dam/Deloitte/global/Documents/Governance-Risk-Compliance/dttl-grc-outsourcingamidcomplexity.pdf

6. Overby S. Outsourcing Contracts and Negotiations Getting MoreComplex. CIO. 2012 April 28. doi:10.1094/PDIS-11-11-0999-PDN

7. Srivastava S, Teo T. Contract performance in offshore systemsdevelopment: role of control mechanisms. J Manage Inf Syst.2012;29(1):115–58. doi:10.2753/MIS0742-1222290104.

8. Rai A, Maruping L, Venkatesh V. Offshore information systemsproject success: the role of social embeddedness and culturalcharacteristics. MIS Q. 2009;33(3):617–41. doi:10.2307/20650313.

9. Ditmore J Why IT outsourcing often fails. Information Week.2012:1339. doi:10.1094/PDIS-11-11-0999-PDN

10. McDougall D. The distributed criterion design. J Behav Educ.2006;15(4):236–46. doi:10.1007/s10864-006-9030-x.

11. Toolbox.com. ITtoolbox outsourcing survey 2004 [cited 2018December 10]. http://hosteddocs.ittoolbox.com/dell1_survey.pdf

8 KO ET AL.

12. Maruping LM, Venkatesh V, Agarwal R. A control theory per-spective on agile methodology use and changing userrequirements. Inf Syst Res. 2009;20(3):377–99. doi:10.1287/isre.1090.0238.

13. Narayanaswamy B, Gonde C, Tredger M, Hussain M, Vergani D,Davenport M. Serial circulating markers of inflammation in bili-ary atresia—evolution of the post-operative inflammatory process.Hepatology. 2007;46(1):180–87. doi:10.1002/hep.21701.

14. Kwan I, Schroter A, Damian D. Does socio-technical congruencehave an effect on software build success? A study of coordinationin a software project. IEEE Trans Software Eng. 2011;37(3):307–24. doi:10.1109/TSE.2011.29.

15. Bapna R, Barua A, Mani D, Mehra A. Researchcommentary-cooperation, coordination, and governance in multi-sourcing: an agenda for analytical and empirical research. Inf SystRes. 2010;21(4):785–95. doi:10.1287/isre.1100.0328.

16. Cha H, Pingry D, Thatcher M. Managing the knowledge supplychain: an organizational learning model of information technol-ogy offshore outsourcing. MIS Q. 2008;32(2):281–306.doi:10.2307/25148841.

17. Vlaar P, van Fenema P, Tiwari V. Cocreating understanding andvalue in distributed work: how members of onsite and offshorevendor teams give, make, demand, and break sense. MIS Q.2008;32(2):227–55. doi:10.2307/25148839.

18. Rico R, Sabcgez-Mnznares M, Gil F, Gibson C. Team implicitcoordination processes: a team knowledge-based approach. AcadManage Rev. 2008;33:163–84. doi:10.5465/amr.2008.27751276.

19. Mani D, Barua A, Whinston AB. Outsourcing contracts andequity prices. Inf Syst Res. 2013;24(4):1028–49. doi:10.1287/isre.2013.0478.

20. Mockus A, Fielding R, Herbsleb J. Two case studies of opensource software development: apache and Mozilla. ACM TransSoftware Eng Method. 2002;11(3):309–46. doi:10.1145/567793.567795.

21. Sabherwal R, Becerra-Fernandez I. An empirical study of theeffect of knowledge management processes at individual, group,and organizational levels. Decision Sciences. 2003;34(2):225–60.doi:10.1111/deci.2003.34.issue-2.

22. Gopal A, Espinosa JA, Gosain S, Darcy DP. Coordination andperformance in global software service delivery: the vendor‘sperspective. IEEE Trans Eng Manage. 2011;58(4):772–85.doi:10.1109/TEM.2010.2096821.

23. Srikanth K, Puranam P. The firm as a coordination system:evidence from software services offshoring. Organ Sci. 2014;25(4):1253–71. doi:10.1287/orsc.2013.0886.

24. Jaworski B. Toward a theory of marketing control: environmentalcontext, control types, and consequences. J Mark. 1988;52(3):23–39. doi:10.1177/002224298805200303.

25. Kirsch LJ. The management of complex tasks in organizations:controlling the systems development process. Organ Sci. 1996;7(1):1–21. doi:10.1287/orsc.7.1.1.

26. Henderson JC, Lee S. Managing I/S design teams: a control the-ories perspective. Manage Sci. 1992;38(6):757–77. doi:10.1287/mnsc.38.6.757.

27. Kirsch LJ, Sambamurthy V, Ko D-G, Purvis RL. controllinginformation systems development projects: the view from theclient. Manage Sci. 2002;48(4):484–98. doi:10.1287/mnsc.48.4.484.204.

28. Nidumolu SR, Subramani MR. The matrix of control: combiningprocess and structure approaches to managing softwaredevelopment. J Manage Inf Syst. 2003;20(3):159–96. doi:10.1080/07421222.2003.11045774.

29. Eisenhardt KM. Control: organizational and economicapproaches. Manage Sci. 1985;31(2):134–49. doi:10.1287/mnsc.31.2.134.

30. Ouchi W. A conceptual framework for the design of organiza-tional control mechanisms. Manage Sci. 1979;25(9):833–48.doi:10.1287/mnsc.25.9.833.

31. Gopal A, Gosain S. Research note-The role of organizationalcontrols and boundary spanning in software development

outsourcing: implications for project performance. Inf Syst Res.2010;21(4):960–82. doi:10.1287/isre.1080.0205.

32. Tiwana A, Keil M. Control in internal and outsourced softwareprojects. J Manage Inf Syst. 2009;26(3):9–44. doi:10.2753/MIS0742-1222260301.

33. Krishnan MS, Kellner MI. Measuring process consistency: impli-cations for reducing software defects. IEEE Trans Software Eng.1999;25(6):800–15. doi:10.1109/32.824401.

34. Katz D, Kahn R. The Social Psychology of Organizations. NewYork: John Wiley & Sons; 1966.

35. Kirsch LJ, Ko D-G, Haney M. Investigating the antecedents ofteam-based clan control: adding social capital as a predictor.Organ Sci. 2010;21(2):469–89. doi:10.1287/orsc.1090.0458.

36. Gittell J. Organizing work to support relational co-ordination.Int J Hum Resour Manag. 2000;11(3):517–39. doi:10.1080/095851900339747.

37. White S, Siu-Yun Lui S. Distinguishing costs of cooperation andcontrol in alliances. Strategic Manage J. 2005;26(10):913–32.doi:10.1002/(ISSN)1097-0266.

38. Kirsch LJ. Portfolios of control modes and IS projectmanagement. Inf Syst Res. 1997;8(3):215–39. doi:10.1287/isre.8.3.215.

39. Okhuysen GA, Bechky BA. 10 coordination in organizations: anintegrative perspective. Acad Manag Ann. 2009;3(1):463–502.doi:10.5465/19416520903047533.

40. Galbraith JR. Organization Design. Reading (MA): Addison-Wesley; 1977.

41. Gulati R, Singh H. The architecture of cooperation: managingcoordination costs and appropriation concerns in strategicalliances. Adm Sci Q. 1998;43(4):781–814. doi:10.2307/2393616.

42. Eisenhardt KM, Furr N, Bingham C. Crossroads -Microfoundations of performance: balancing efficiency and flex-ibility in dynamic environments. Organ Sci. 2010;21(6):1263–73.doi:10.1287/orsc.1100.0564.

43. Crowston K. A coordination theory approach to organizationalprocess design. Organ Sci. 1997;8(2):157–75. doi:10.1287/orsc.8.2.157.

44. Cray D. Control and coordination in multinational corporations.J Int Bus Stud. 1984;15(2):85–98. doi:10.1057/palgrave.jibs.8490483.

45. Dekker H. Control of inter-organizational relationships: evidenceon appropriation concerns and coordination requirements. Acc,Organ Soc. 2004;29(1):27–49. doi:10.1016/S0361-3682(02)00056-9.

46. Klimoski RJ, Mohammed S. Team mental model: construct orMetaphor. J Manage. 1994;20(2):403–37. doi:10.1177/014920639402000206.

47. Mathieu J, Heffner TS, Goodwin GF, Salas E, Cannon-Bowers JA.The influence of shared mental models on team process andperformance. J Appl Psychol. 2000;85:273–83.

48. Helgeson J, Voss K, Terpening W. Determinants of mail-surveyresponse: survey design factors and respondent factors. PsycholMarketing. 2002;19(3):303–28. doi:10.1002/(ISSN)1520-6793.

49. Podsakoff PM, MacKenzie SB, Lee J-Y, Podsakoff NP. Commonmethod biases in behavioral research: a critical review of theliterature and recommended remedies. J Appl Psychol. 2003;88(5):879–903. doi:10.1037/0021-9010.88.5.879.

50. Hoegl M, Weinkauf K, Gemuenden H. Interteam coordination,project commitment, and teamwork in multiteam R&D projects:A longitudinal study. Organ Sci. 2004;15(1):38–55. doi:10.1287/orsc.1030.0053.

51. Liberatore M, Luo W. The analytics movement: implications foroperations research. Interfaces. 2010;40(4):313–24. doi:10.1287/inte.1100.0502.

52. Susarla A, Subramanyam R, Karhade P. Contractual provisions tomitigate holdup: evidence from information technologyoutsourcing. Inf Syst Res. 2010;21(1):37–55. doi:10.1287/isre.1080.0204.

53. Moore GC, Benbasat I. Development of an instrument to measurethe perceptions of adopting an information technology

JOURNAL OF COMPUTER INFORMATION SYSTEMS 9

innovation. Inf Syst Res. 1991;2(3):192–222. doi:10.1287/isre.2.3.192.

54. Bagozzi RP. Performance and satisfaction in an industrial salesforce: an examination of their antecedents and simultaneity.J Mark. 1980;44(2):65–77. doi:10.1177/002224298004400208.

55. Nunnally JC. Psychometric Theory. New York: McGraw Hill; 1967.56. Neter J, Wasserman W, Kutner M. Applied linear statistical mod-

els: Regression, analysis of variance, and experimental designs.Homewood (IL): Irwin; 1990.

57. Hayes AF, Preacher KJ. Statistical mediation analysis witha multicategorical independent variable. Br J Math Stat Psychol.2014;67:451–70. doi:10.1111/bmsp.12028.

58. Preacher KJ, Hayes AF. Asymptotic and resampling strategies forassessing and comparing indirect effects in multiple mediatormodels. J Educ Behav Stat. 2008;40:879–91.

59. Shrout P, Bolger N.Mediation in experimental and nonexperimentalstudies: new procedures and recommendations. Psychol Methods.2002 December;74(4):422–45. doi:10.1037/1082-989X.7.4.422.

60. Gregory RW, Beck R, Keil M. Control balancing in informationsystems development offshoring projects. MIS Q. 2013;37(4):1211–32. doi:10.25300/MISQ.

61. Faraj S, Sproull L. Coordinating Expertise in SoftwareDevelopment Teams. Manage Sci. 2000;46(12):1554–68.doi:10.1287/mnsc.46.12.1554.12072.

10 KO ET AL.

Appendix(Measures)

Project Performance:8,22,61

Please indicate the extent to which you agree or disagree with each of thefollowing statements about the outcomes of the activities performed by thisvendor:

Project Coordination:50,51

Please respond to each of the following statements about coordinationwith this vendor during the project. Extent to which your team and thevendor team were well coordinated with regard to

Behavior Control:25,27,38

Please indicate the extent to which you agree or disagree with each of thefollowing statements about monitoring this vendor’s activities:

Outcome Control:25,27

Please indicate the extent to which you agree or disagree with each of thefollowing statements about monitoring this vendor’s activities:

Vendor Reputation:53

Please indicate the extent to which you agree or disagree with each of thefollowing statements prior to the start of this project:

Project Size:31

In total, approximately how many people were directly involved in theproject?

Internal IS staff: _______External consultants/vendor staff: _______

* Dropped items.

a. project scopeb. project schedulec. project budget

a. We expected this vendor to follow a written sequence of steps to achievetheir goals.

b. We assessed the extent to which this vendor followed existing proceduresduring the project.

c. We enforced this vendor to adhere to written rules and procedures.d. *We placed significant weight on project-related behaviors when

assessing this vendor’s performance.e. *We held this vendor accountable for how they behaved during the

project.

a. The activities performed by this vendor were completed within schedule.b. The activities performed by this vendor were completed within budget.c. The activities performed by this vendor satisfied contract requirements.d. The activities performed by this vendor were of high quality.a. Based on the activities performed by this vendor, we would choose the

same vendor for future projects.b. Based on the activities performed by this vendor, we would recommend

this vendor to other project managers.

a. This vendor had a good reputation.b. This vendor had a proven track record.

a. *We used pre-established targets as benchmarks to evaluate this vendor’sperformance.

b. We placed significant weight on this vendor’s project completion withinschedule.

c. *We placed significant weight on this vendor’s project completion withinbudget.

d. We placed significant weight on this vendor’s project completion to thesatisfaction of the users.

e. We evaluated performance by the extent to which this vendoraccomplished project goals.

f. The rewards for this vendor were linked to results.

JOURNAL OF COMPUTER INFORMATION SYSTEMS 11