Advances in Production Engineering & Management Volume 14 | Number 2 | June 2019 | pp 225-238 https://doi.Org/10.14743/apem2019.2.324 ISSN 1854-625G Journal home: apem-journal.org Original scientific paper A new framework for complexity analysis in international development projects - Results from a Delphi study Gajic, S.a*, Palčič, I.b aUniversity of Novi Sad, Faculty of Technical Sciences, Novi Sad, Serbia bUniversity of Maribor, Faculty of Mechanical Engineering, Maribor, Slovenia A B S T R A C T A R T I C L E I N F O The main objective of this paper is to develop a framework for characterising project complexity in International Development (ID) projects. Contemporary challenges in ID projects have led to their growth in their complexity, which in recent years has driven researches in recent years to publish numerous papers that deal with this topic, demonstrating its importance in current project management research. Nevertheless, existing literature lacks in generally accepted framework that considers specifics of project complexity in ID projects. Thus, new framework was developed, based on a two-round Delphi survey, building upon existing TOE (technology-organisation-environment) framework with new empirical insights given from the experts in the field of ID projects. The main contribution of the paper is the validation of existing TOE complexity factors, in the context of International Development projects. Additionally, eight new complexity factors were proposed by the experts, and it was concluded that Environmental complexity had the biggest impact on International Development projects. From a managerial perspective, proposed complexity framework can be used for making a complexity footprint, which could indicate the critical areas of the project where complexity could be expected. In addition, the model represents a novel theoretical lens for assessing complexity in ID projects. Keywords: International development projects (ID); Project management; Complexity; Project complexity; Technology-organisation-environment (TOE); Delphi study *Corresponding author: gajic.sladjana@uns.ac.rs (Gajic, S.) Article history: Received 20 December 2018 Revised 15 May 2019 Accepted 20 May 2019 2019 CPE, University of Maribor. All rights reserved. 1. Introduction Over the last ten years, interest in project management has significantly grown. It has been reported that about 24 % of the world GDP ($19 trillion) is spent on projects every year [1]. Project failure seems to be the rule in all types of projects [2], and this seems to be particularly true for International Development (ID) projects that have the ultimate objective to reduce poverty or to improve governance and build institutional capacity [3-5]. Contemporary challenges (dynamic and uncertain environment, increasing number of stakeholders) that influence these projects are closely related to the complexities of these projects [6-10]. It has been acknowledged that there is a correlation between project performance and complexity, and still, there is a huge knowledge gap about how complexity relates to project management practice and no widely accepted framework of project complexity in ID projects [11-14]. Numerous papers have been published in the field of project complexity, with intention to explain the relationship between complexity theory and project management [8, 9, 15, 16]. Management of ID projects requires a novel framework for dealing with project complexity due to their specifics in comparison to conventional projects. The framework in this paper includes 225 Gajic, Palcic some of the elements of TOE (technology-organisation-environment) framework, proposed by Bosch-Rekveldt, as well as the complexity model proposed by Vidal and Marle [9, 17]. This paper investigates complexity factors that contribute to overall complexity of ID projects from the perspective of the experts. ID projects have specific context that increases their overall complexity. These projects cover almost every sector of activity, since they take place at multiple locations and in a different time zone. These are public-sector projects that show cultural complexity, unique context and institutional challenges. ID projects are specific due to their intangible, unique goals, unique ways of organizing, tool-intensity and the large number of stakeholders. Delphi method has been used in this study to verify the significance of existing complexity factors proposed in the literature and to further update the list of complexity factors in the context of ID projects. Novel, modified complexity framework was developed based on the insights of the Delphi study. We found that experts take as most important the following complexity factors in ID projects: Clarity of goals, Variety of stakeholders' perspectives, Dependencies between tasks, Interface between different disciplines and Dependencies on other stakeholders. Most of the factors with the highest grade are in the group of Environmental complexity, which makes this type of complexity the most important for this type of projects. This conclusion is a unique contribution, in comparison with the results given from the similar studies done on different type of projects (see for example [17]). The paper is organized as follows. Section 2 reviews recent works on complexity in the field of ID projects and presents the theoretical background of the proposed model. Section 3 introduces Delphi research, including data collection an analysis processes related to the Delphi questionnaires done by practitioners and experts. Section 4 proposes a refined measuring model for complexity in ID projects and shows the significance of this research. The final section presents conclusions, potential implications and limitations of the proposed model. 2. Literature review 2.1 International development projects International development (ID) projects are the projects that deal mainly with poverty reduction, and cover sectors of agriculture, transportation, water, energy, health, population, education, reform and governance, etc. [18]. These are public projects, funded by donors from developed countries and are implemented in under-developed countries, which bring numerous political and cultural challenges [4]. ID projects have certain similarities with conventional projects: they deliver goods and services, they are limited, temporary and unique endeavours that go through project life cycle; these projects are constrained by the "iron triangle"- time, cost, and use project management standards, tools and techniques for the implementation [3, 19]. Peculiarities of ID projects are often interconnected with their not-for-profit, social, technical, and political nature; they are funded by external donors and have intangible and even conflicting objectives difficult to measure. In addition, they often have more stakeholders in comparison to conventional projects - at least three most important stakeholders: funding agency, implementing agency and the beneficiaries [3], that often have conflicting expectations. ID projects have specific context in which they occur, with numerous political, cultural, legal, social, technical, economic and environmental challenges. Important characteristics of ID projects are optimism bias, planning fallacy, strategic misrepresentation, and they are prone to media scrutiny, intolerance of failure, rigid procedures etc. [2]. Ika et al. [4] claim that due to their evident socio-political complexity, ID projects could "fit at the far right end of the spectrum on a continuum from private sector projects, through public sector projects, to international projects". 226 Advances in Production Engineering & Management 14(2) 2019 A new framework for complexity analysis in international development projects - Results from a Delphi study 2.2 Project complexity Projects can be observed as complex, self-organising systems, with their specifications (requirements and constraints) that undergo design process that is highly social, consisting of hundreds of designers, customers, and other participants [24]. Complexity has been recognized as one of the most important streams in project management research [20]. One of the first problems in understanding project complexity is lack of consensus regarding the definition of complexity in the project context [4, 6, 7, 13-16]. A review of recent papers had confirmed that definitions of complexity continue to be ambiguous [17, 18]. One of the most cited definitions is Bac-carini's one defines complexity as "consisting of many varied interrelated parts and can be operationalised in terms of differentiation and interdependency" [21]. Followed by Baccarini's work, Williams explained that complexity consists of "structural complexity" - the number of varied components, "interdependency" - degree of dependence between these components and "uncertainty in goals and means" [22]. Geraldi et al. clarified complexity into structural, dynamical, uncertainty, pace and socio-political complexity. Sommer and Loch [23] define complexity as having "two dimensions": system size (the number of influence variables) and the number of interactions among influence variables. The term "complex" stems from the Latin words cum (together, linked) and plexus (braided, plaited). The Oxford dictionary defines complexity as "consisting of parts" and "intricate, not easily analysed or disentangled." [15]. When defining complexity, it is important to make a distinction between two terms [8]: "complex" and "complicated". Complex systems contain multiple parts with several connections and interactions between the parts and behaviour that is a result of these emergent properties. Complicated systems emerge as the result of complex systems, without the right tools for analysis and management. Vidal claims that there are two main scientific approaches to complexity [24]: 1) Descriptive complexity - this approach considers complexity as an intrinsic property of a project system, 2) Perceived complexity - complexity as subjective matter. Bakhshi et al. concluded that there are three main schools of thought within the construct of complex projects: the Project Management Institute (PMI) perspective, the "System-of-systems" (SoS) approach, and the complexity theories perspective [25]. Cicmil and colleagues distinguish two different terms when discussing complexity [5]: • Complexity in projects (how complexity can be manifested in projects), • Complexity of projects (what factors make projects complex or difficult to manage). The first stream is mainly theory-driven, and leans on complexity theories [5, 14, 19]. The second stream is practitioner-driven and aims to identify factors of complex projects and the strategies on how organisations can respond to complexity [20-23]. This paper focuses on the second stream. Lack of consensus in defining project complexity leads to lack of understanding the concept. In this paper, we will accept Vidal's definition of project complexity as "the property of a project which makes it difficult to understand, foresee, and keep under control its overall behaviour, even when given reasonably complete information about the project system" [24]. 2.3. Complexity factors in project management - Gathering elements from the literature Numerous attempts have been made to measure and model project complexity (Appendix A -Complexity measuring in the literature), and most of them attempted to measure complexity quantitatively by focusing on the most important complexity factors [8, 9, 14, 17, 21, 22, 26-28]. In this paper, the classification of complexity factors proposed by Bosch-Rekveldt has been adopted - the TOE framework (Table 1), that includes Technological, Organizational and Environmental complexity factors [17]. It builds upon Baccarini's and Williams' existing complexity frameworks [21, 29]. Advances in Production Engineering & Management 14(2) 2019 227 Gajic, Palcic _Table 1 Included elements of TOE framework_ Complexity type_Element name_ Organisational complexity Duration, Compatibility of different project management methods and tools, Size in CAPEX, Size of the project team, Number of locations, Resource and skills, Experience with parties involved, Interfaces between different disciplines in ID projects, Number of different nationalities in ID projects, Number of different languages, _Cooperation of JV partners, Trust in project team (JV partner), Organisational risks Technological complexity Number of goals, Goal alignment, Clarity of goals, Scope largeness, Uncertainties in scope, Quality requirements, Number of tasks, Variety of tasks, Dependencies between tasks, Uncertainty in technical methods, Conflicting norms and standards, _Newness of technology, Experience with technology_ Environmental complexity Number of stakeholders (internal and external), Variety of stakeholders' perspectives, Dependencies on other stakeholders, Political influence, Organisational internal support, required local content, Experience in the country of implemen-_tation, Stability of project environment, Risks from environment_ De Bruijn already categorized complexity factors in three groups: technical complexity, social complexity and organizational complexity. This categorization has been furtherly developed by Jaafari, and also by Xia and Lee, but they have investigated the significance of TOE factors only in large engineering projects [17]. Technical view included technical content of the project, Organizational view included people and organizational aspects of the project, and Environmental view was mainly focused on the influences in the project environment on the complexity. 3. Methodology 3.1 Delphi method As already mentioned, there is no clear consensus of the researchers on the complexity measures. Several studies have already been done in the area of project complexity, based on statistical calculations or surveys. ID projects are characterised with dynamic environments, numerous stakeholders, customized projects, and exposure to external conditions that often make traditional, research methods unrealistic for this type of research. The main benefit from conducting a qualitative study is validation of a local expression and ability to understand certain phenomena from the inside out [30]. Delphi method is designed to obtain reliable consensus about the topic from a panel of experts by conducting series of questionnaires combined with controlled opinion feedback, and with results of each round being fed into the next round [31]. In the field of management, a modified approach of Delphi method has been used to shape a group consensus about the relative importance of proposed issues [32]. Lindstone and Turoff [33] proposed the following definition: "Delphi may be characterized as a method for structuring a group communication process so that the process is effective in allowing a group of individuals, as a whole, to deal with a complex problem. To accomplish this "structured communication" there is a need to provide the following: some feedback of individual contributions of information and knowledge; some assessment of the group judgment or view; some opportunity for individuals to revise views; and some degree of anonymity for the individual responses." Delphi method is more objective in its outcomes than individual statements, even though the judgments of experts are based on subjective opinions. One of the main advantages of the approach is the fact that direct confrontation of the experts is avoided [32]. Delphi method was widely used in the field of industrial engineering and project management [34-36]. There are several quantitative methodologies that could be used for investigating the relative importance of issues (e.g., Emerging Issues Analysis, Environmental Scanning, Issues Management, Analytical Hierarchical Process). Most of them are future oriented and outline individual opinions. The major advantage of Delphi in comparison to these methodologies is that it is the most prominent of consensus methodologies [37]. Additionally, Delphi approach was selected in this paper to reconcile different opinions between practitioners and experts about the importance of different complexity factors in ID projects. The validity of results was assured by heterogeneity of the panellists and anonymous response format. 228 Advances in Production Engineering & Management 14(2) 2019 A new framework for complexity analysis in international development projects - Results from a Delphi study Delphi study does not need to include a representative sample of any population. It consists of qualified experts who have a deeper understanding of the selected research issues, which makes the selection of the participants one of the most critical requirements [38]. Existence of bias was reduced in the study by implementation of a well-structured, academically rigorous process, and by selecting qualified experts for participation in the study in accordance with predefined guidelines - preparing a Knowledge Resource Nomination Worksheet. In this paper two-round "ranking-type" Delphi was used to develop group consensus about the relative importance of complexity factors in ID projects (Figure 1 Delphi study algorithm). Purpose of the research was to develop a ranked list of most important complexity factors for ID projects. Two panels of participants were selected: the first group were academics, and the second group were practitioners - experienced project coordinators in ID projects. Three-step strategy was adopted as a research program in the paper. Firstly, list of complexity factors that contribute ID projects was identified, based on the literature review (see Appendix A). Secondly, the identified factors were quantitatively tested, in order to verify if selected factors were truly relevant to the experts. Additional factors were proposed by the panellists. Thirdly, the selected factors were again ranked in the second round of the Delphi, and recommendations weremade. Biases are reduced by strategically constructed questionnaires, controlled feedback, detailed analysis of the group response and by two rounds of the research. Iteration is essential factor of any Delphi study. In this paper, iteration involved redistribution of the Delphi survey accompanied with controlled feedback, given to panellists with simple statistical summaries of the responses from the first round. This step, together with preserved anonymity of the participants, eliminates the dominance bias and minimizes the effects of the von Restorff effect [39]. Recent studies provide some theoretical discussion related to the complexity factors in the field of ID and their relative importance [4]. This paper contributes in obtaining a more comprehensive view from the perspective of the two major stakeholders in international development: practitioners and academics. OVERALL RESEARCH PROCESS Carrying out state of the art on project Building up a first version of a project complexity (factors and description) complexity framework (PCF) Writing the questionnaire for Selecting varied experts as the Delphi study the Delphi panelists DELPHI STUDY PERIMETER Carrying out the first round of Analyzing the results of the the Delphi study first round of the Delphi study Summarizing the results of the round of the Delphi study Sending the results of the round to the panelists Carrying out the next round of the Delphy study Analysing the results of the next round of the Delphy study NO YES Drawing conclusions on the Delphi study (deeper analysis correlation,...) Refining the PCF using the results of the Delphi study Applying the refined PCF to some complexity related issues Fig. 1 Delphi study two round algorithm [24] Advances in Production Engineering & Management 14(2) 2019 229 Gajic, Palcic 3.2 Participants selection In this paper, experts were divided into two panels: academics and practitioners. Heterogenous group of participants allows for a somewhat different perspectives about the selected topic, as well as the comparison of the perspectives of the different stakeholder groups. A panel usually consists of 15 to 30 participants from the same discipline, or five to 10 per category from different professional groupings. Following the recommendations from Delphi literature, there was 11 participants in the first phase of the study, and 7 participants in the second phase of the study. The following steps were included in selection of the experts for the Delphi study (Table 2) [38]: Step 1: Preparing a Knowledge Resource Nomination Worksheet (KRNW) - identification of relevant disciplines or skills: academics and practitioners; identifying relevant academic and practitioner literature, Step 2: Populating KRNW with names, Step 3: Nominating additional experts by existing contacts, Step 4: Inviting experts for each panel until the target size of the panel is reached. The academics were selected based on the literature review of academic and practitioner journals. We have identified experts and asked them to nominate others for inclusion on the list. They were provided with a brief description of the Delphi study and explanation that we have identified them as the experts on complexity in the field of project management and invited to participate in the study. The practitioners were selected from the base of Erasmus plus project coordinators. Seven academics and four practitioners agreed to participate in the Delphi study. Web was used as the mean for reaching focus organizations. Related literature focused on two most prominent SCI journals in the area of project management (International Journal of Project Management and Project Management Journal) were reviewed in order to identify articles concerning ID projects and complexity. Delphi questionnaire was administrated using e-mail and Survey Monkey software. Table 2 Knowledge Resource Nomination Worksheet Disciplines or Skills Organizations Related Literature Academics European Commission Academics Journal List International Journal of Project management Practitioners - Erasmus plus coordinators' list Project Management Journal 4. Results and discussion 4.1 Data collection and analysis method The Delphi questionnaires were administered using e-mail and the Web. One of the benefits of using these "rapid" media is increasing speed of the turnaround time between questionnaires, which is important factor in the Delphi method [32]. Administration of the questionnaires included the proposed procedure for "ranking-type" Delphi studies by Schmidt [40], that includes the following steps: • Brainstorming for important factors and validation of the proposed factors, • Narrowing down the original list to the most important ones, • Ranking the list of important factors. 4.2 Brainstorming for important complexity factors In this phase, panellists were asked to rank complexity factors on the five-level Likert scale. These factors belonged to the three main groups of factors: technological, organizational and environmental group. In addition, participants were asked to list additional relevant complexity factors in ID projects with a brief explanation for each factor. Duplicates were removed, new complexity factors were classified, and the terminology of the proposed factors was unified. After this, consolidated lists were sent to participants. 230 Advances in Production Engineering & Management 14(2) 2019 A new framework for complexity analysis in international development projects - Results from a Delphi study Validation of categorized list of factors - in this phase, panellists were given a list with all the consolidated factors obtained from the first questionnaire, grouped into categories, with brief explanation of each factor, based on the information from the first questionnaire. Furthermore, an exact copy of the responses from the first phase was sent to participants. Panellists were asked to verify that their answers were correctly interpreted and placed them in an appropriate category. According to Schmidt [40], "without this step, there is no basis to claim that a valid, consolidated list has been produced." New elements were proposed by the experts: (1) Overlap of the project phases, (2) Interdependence of different stakeholders, (3) Diversity of stakeholder expectations, (4) Lack of clarity or consensuses on project benefits among project stakeholders, (5) Variation (1st type of uncertainty), (6) Foreseen uncertainty (2nd type of uncertainty), (7) Unforeseen uncertainty (3rd type of uncertainty), and (8) Chaos (4th type of uncertainty). TOE factors were rated on a five-level Likert scale by the panellists. Consensus measurement has a pivotal role in Delphi research which could be defined as a gathering around median responses with minimal divergence [41]. Two criteria were selected for consensus measurement: (a) mean > 3 and (b) Interquartile range IQR < 1. Six elements had the mean less than three, and were excluded from the next phase: (1) Quality requirements, (2) Duration of ID projects, (3) Compatibility of different project management methods and tools, (4) Size in CAPEX, (5) Number of different languages, and (6) Cooperation of Joint Venture partners. Consensus between two groups of panellists was not reached (IQR>1) for the following elements, and they were rated again in the second phase: (1) Number of goals, (2) Goal alignment, (3) Clarity of goals, (4) Number of tasks, (5) Dependencies between tasks, (6) Conflicting norms and standards, (7) Newness of technology (world-wide), (8) Experience with technology, (9) Resource and skills availability, (10) Interfaces between different disciplines, (11) Trust within the project team (Joint Venture partner), and (12) Experience in the country of implementation. 4.3 Narrowing down the original list to the most important complexity factors In the second phase, the list of factors was narrowed. Four participants did not proceed with the study; seven panellists remained in the second phase of the study (four academics and three practitioners). The main goal of this phase was to understand the rating of importance of the factors, based on the different perspectives of various stakeholder groups. All the new factors proposed by the participants had the mean >3, which makes them all significant to complexity of ID projects based on the opinion of the panellists. In addition, IQR was >1 - consensus was not reached on the following statements: • Number of different nationalities in ID projects influences project complexity; • Political influence in ID projects influences project complexity; • Stability of the project environment (in terms of exchange rates, material pricing etc.) in ID project influences project complexity; • Interdependence among different stakeholders' influences project complexity; • Chaos (4th type of uncertainty) influences project complexity. T-tests measure the difference between two groups of panellists. In the first round, they showed difference on seven answers, and in the second round there is no difference in the answers. Project coordinators thought that the number of goals, compatibility of different project management methods and tools, as well as the size in CAPEX and required local content in ID projects had more significant effect on project complexity than academics did. Academics find that variety of stakeholder's perspectives, dependencies on other stakeholders and the political influence had a more significant effect on project complexity than project coordinators did. Advances in Production Engineering & Management 14(2) 2019 231 Gajic, Palcic Table 3 Wilcoxon test Number of goals Goal alignment Clarity of goals Number of tasks Dependencies between tasks Conflicting norms and standards Newness of technology (worldwide) Experience with technology Resource and skills availability Interfaces between different disciplines Number of different nationalities Z -1.300b -1.13b -1.134c -1.633c -1.000c -.816b -0.333c -1.000c -0.577c -1.604c -0.736c Asymp. Sig. (2-tailed) 0.194 0.257 0.257 0.102 0.317 0.414 0.739 0.317 0.564 0.109 0.461 aWilcoxon signed ranks test bBased on positive ranks cBased on negative ranks Wilcoxon matched-pairs signed-ranks test measures changes in consensus between first and second round of Delphi study. Wilcoxon test has the purpose of measuring stability of the data and helping researchers determine if there was a difference between the data of two Delphi rounds. All the concepts reached stability (the significance level was set at .05) and thus the Delphi was terminated with two rounds. For 11 questions that were repeated in both rounds, there was no significant statistical difference (Table 3). 4.4 Ranking the list of important factors The descending order of the top ten weighted measures were found to be Clarity of goals (4.57), Variety of stakeholders' perspectives (4.45), Dependencies between tasks (4.43), Interfaces between different disciplines (4,29), Dependencies on other stakeholders (4.27), Risks from environment in ID project (4.27), Lack of clarity or consensuses on project benefits among project stakeholders (4.14), Unforeseen Uncertainty (4.14), Political influence (4.14), Number of stakeholders (internal and external) (4.09); for all the means see Appendix B. Most of the factors with the mean higher than three are in the group of Environmental complexity. Based on the research, it is concluded that Environmental complexity has the most significant effect on the composite complexity, in comparison to technical and organizational complexity contribution. It might be concluded that experts think that environmental complexity is the most important of three types of complexity, which is the main contribution of the paper. In addition, it was found that uncertainty significantly influences complexity, based on the new factors proposed by the experts. Four uncertainty types were proposed, based on the classification of Meyer et al. - variation, foreseen uncertainty, unforeseen uncertainty and chaos. Particularities of different types of uncertainty require different managerial approach: "Projects in which variation and foreseen uncertainty dominate allow more planning, whereas projects with high levels of unforeseen uncertainty and chaos require a greater emphasis on learning." Novel, modified complexity framework was developed based on the insights of the Delphi study (Fig. 2). 4.5 Implications of the study results In the increasingly complex and unpredictable environment of ID projects, understanding complexity is becoming very important for planning, managing and executing strategies. Organizations that are delivering ID projects need to understand and adapt to these changes and include constant feedback from all the stakeholders in all the project phases. One of the uses of the given complexity framework is creating awareness amongst the different stakeholders about the complexity on the project. Additionally, the framework could be used to access complexity on the project in different project phases and react in accordance to it. Project managers of ID projects are usually not enough equipped to adequately handle complex projects since they base on their managerial style on traditional project management tools and techniques. Understanding the potential complexities can be a first step to making better strategies to manage ID projects. 232 Advances in Production Engineering & Management 14(2) 2019 A new framework for complexity analysis in international development projects - Results from a Delphi study Ji « tj tj c s Ü -C III o» o f 3 C c & c ¡o o o S" ai c o u "re u to c -£ o S o ^ > c C ? § dJ o c CL 1 o 2 ^ O- Q- £ .'S c o» -g ! 2 s? ai ¡ 1 s ^ eu S O ■> c O - re <1) C il Z v » Q. Q. 1 et Û to £ T> 12 a> c S a a> Q. a> O S 2 I " 'E re CD Ô "O cu 5 CT ai oc •s si £ UJ — >N .— ■ti > = c X3