See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/324311895

Why and How Do Project Management Offices Change? A Structural Analysis

Approach

Article  in  International Journal of Project Management · April 2018

DOI: 10.1016/j.ijproman.2018.04.001

CITATIONS READS

6 363

3 authors:

Christophe Bredillet Stephane Tywoniak

Université du Québec à Trois-Rivières University of Ottawa
116 PUBLICATIONS   1,163 CITATIONS    64 PUBLICATIONS   394 CITATIONS   

SEE PROFILE SEE PROFILE

Mahshid Tootoonchy
Queensland University of Technology
7 PUBLICATIONS   29 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

On complexity and theorizing complexity in organizational and project studies View project

Institutional Stability and Change View project

All content following this page was uploaded by Christophe Bredillet on 08 April 2018.

The user has requested enhancement of the downloaded file.

Manuscript Details
Manuscript number JPMA_2017_733

Title Why and How Do Project Management Offices Change? A Structural Analysis
Approach

Article type Full Length Article

Abstract
The growing popularity of Project Management Offices (PMOs) as organizational structures is grounded in the
assumption they support more efficient and effective project management for better strategy implementation. However,
research emphasizes they fail to deliver expected value: their unstable nature precludes the delivery of long-term
benefits. This is compounded by the absence of a theory of PMO change and adaptation. Recent research, taking a
co-evolution lens rooted in evolutionary theory, suggests that PMOs should be studied in relation to the broader
organizational context, in order to better capture the dynamic interplay and fit between them. In this study, taking a
routine perspective as micro-foundation and unit of analysis, we focus on the co-evolution between PMO and Project
Portfolio Management (PfM) as organizational capability for six case studies. A structural analysis of the relational
routines' system between PMO, PfM and the Organizational context allow us to unveil dynamics at stake, i.e. why and
how changes occur, as well as eigen behaviors and the changing states of various routines elements (influential,
mediating, dependent or not-influential). This study makes five contributions. We show that: 1) PMO and PfM can be
conceptualized as collections of routines, 2) PMO and PfM co-evolve over time to adapt to organizational context
influence, 3) the co-evolution of a routines' system, abstracted as a non-trivial machine, exhibits an eigen behavior, 4)
applying a structural analysis approach allows to simulate the dynamics of a routines' system and to unveil the role of
key routine elements and 5) eigen values of routines' systems allow to characterize their eigen behavior.

Keywords PMO; PfM; co-evolution; routine dynamics; structural analysis; MICMAC; non-
trivial machine; eigen behavior

Taxonomy Theory of Research into Project Management, Project Management, The Project
Office

Corresponding Author Christophe Bredillet

Corresponding Author's Université du Québec à Trois-Rivières

Institution

Order of Authors Christophe Bredillet, Mahshid Tootoonchy, Stephane Tywoniak

Why and How Do Project Management Offices Change?
A Structural Analysis Approach

Highlights

 PMO and PfM can be conceptualized as collections of routines

 PMO and PfM co-evolve over time to adapt to organizational context influence
 the co-evolution of a routines' system, abstracted as a non-trivial machine, exhibits
an eigen behavior,
 applying a structural analysis approach allows to simulate the dynamics of a
routines' system and to unveil the role of key routine elements
 eigen values of routines' systems allow to characterize their eigen behavior.
Why and How Do Project Management Offices Change?
A Structural Analysis Approach

1
 1. Introduction

Project-based organizations have received continued attention (Keegan & Turner, 2002;
Lindkvist, 2004; Sydow, Lindkvist, & DeFillippi, 2004; Söderlund, 2008; Turkulainen, Kuala,
Artto, & Levitt, 2013; Kwaak, Sadatsafavi, Walewski, & Williams, 2015; Prado & Sapsed,
2016; Miterev, Mancini, & Turner, 2017). The simultaneous management of projects in
project-based organizations require extensive portfolio management to co-organize project-
related activities (Jerbrant & Gustavsson, 2013) and prioritize sometimes competing and
conflicting projects to achieve strategic objectives (Archer & Ghasemzadeh, 1999). Research
suggests the Project Management Office (PMO), as an organizing form within an organization
(Kerzner, 2003), has a positive influence on the success of Project Portfolio Management (PfM)
(Turner, 2014) and should improve the efficiency of project management (Stanleigh, 2006). If
PfM is the building block of strategy implementation (Dietrich & Lehtonen, 2005; Shenhar,
Dvir, Levy, & Maltz, 2001), then the PMO facilitates strategy implementation through
portfolios of projects. PMOs "accompany organizational changes leading toward strategic
objectives" (Aubry & Hobbs, 2011, p. 60).

However, despite the growing popularity of PMOs as a means of improving the projects’
performance (doValle, da Silviera, & Soares, 2008), three-fourths of PMOs shut down in the
first three years of their establishment and fail to produce convincing business value (Singh,
Keil, & Kasi, 2009) or to present satisfactory performance (Aubry & Hobbs, 201. 1Pellegrinelli
& Garagna (2009: p. 649) argue that the frequent reconfiguration of PMOs is related to the
value that PMOs are bringing. A PMO is established to answer a need; when the need is
addressed, the value of the PMO decreases, and the PMO is reconfigured to generate new
expected value. This dynamic nature of PMOs is well documented (Aubry, Müller, Hobbs, &
Blomquist, 2010; Dai & Wells, 2004; Hobbs & Aubry, 2006, 2007; Turner & Keegan, 2001).
However, the details of the transformation of PMOs, including their interactions with PfM, are
not clearly understood.

Based on these observations, and to better understand the evolutionary nature of PMOs, and
the underlying reasons leading to their evolution, researchers have explored the characteristics
of PMOs (Hobbs & Aubry, 2007) such as forms (Aubry, Hobbs, & Thuillier, 2007),
contribution to organizational performance (Aubry & Hobbs, 2011) and transformative nature
(Aubry, Müller, Hobbs, & Blomquist, 2010; Hurt & Thomas, 2009; Pellegrinelli & Garagna,
2009). These contributions reveal that PMOs present various structures and types (Hobbs &
Aubry, 2006), and that PMO forms are unstable and change frequently (Aubry, Hobbs, &

2
Müller, 2010; Aubry, Müller, et al., 2010). If the dynamic nature of PMOs is well documented,
as mentioned above, this paper argues that considering PMOs in isolation is not sufficient.
Investigating "the creation and reconfiguration of PMOs as an organizational innovation"
(Hobbs, Aubry, & Thuillier, 2008, p. 547), Hobbs et al. acknowledge that "The PMO and its
host organisation coevolve" (2008, p. 550). Their research emphasized the unstable nature of
organizational structures and the difficulty to uncover any patterns of evolution, and
highlighted five organizational tensions contributing to make sense of the new PMOs structure
("economic, political, customer relation, standardization versus flexibility and controlling the
project machine") (Hobbs, Aubry, & Thuilier, 2008, p. 551).

This paper builds on this foundation and aims to contribute to the emerging literature on
Organizational Project Management (Cattani, 2011; Drouin, Müller & Sankaran, 2013;
Sankaran, Müller & Drouin, 2017) by offering a new way to investigate the processes of PMO
evolution and change. We argue that our understanding can be furthered by adopting an
innovative research perspective in the context of organizational project management research,
i.e. using "organizational routines as a unit of analysis" (Pentland & Feldman, 2005) for
exploring change. Investigating the co-evolution of PMO and PfM as organizational capability
through the lens of routines enables to investigate the dynamics (Feldman et al, 2016) at stake
between multiple levels of organizing, linking micro and macro perspectives to reveal the
critical role of the interrelations between organizational routines, capabilities and structures
(Salvato & Rerup, 2011; Salvato & Rerup, 2017). In our view, routines provide the explicative
micro-foundations (Felin & Foss, 2009) of the dynamics under consideration as "an analysis
of micro-foundations considers both initial conditions and evolutionary processes" (Felin et
al., 2012, p. 1353).

It is well recognised that organizations face uncertainty (since Knight and Keynes), and that
increasing complexity in "an age of turbulences" (Bulton et al., 2015; Tsoukas, 2017) leads to
live in instability and dynamic pluralistic contexts (Denis et al., 2007; Smith & Lewis, 2011).
The purpose of our research is to understand why and how PMO and PfM, in relation to the
broader organizational context (Aubry, 2015), co-evolve over time through theoretical insights
drawn from a routine lens, process theories of organizational change (Van de Ven & Sun, 2011)
and empirical case studies. To our knowledge, there is no research to date that offers a dynamic
view of the co-evolution between PMOs and PfM. Applying a structural analysis approach to
the PMO, PfM and Organizational context routines' system allow to capture its dynamics of
evolution on longer time frames, i.e. its eigen behavior (Tsoukas, 2017).

3
 We contribute to extant research on PMOs and project organizing in three ways. First, our
findings account for the ongoing transformative nature of PMOs, supporting extant literature.
Second, with its grounding in a routine perspective, our conceptual framework enriches
existing research, showing that our conceptualization of both PMO, as an entity, and PfM, as
an organizational capability, can be operationalized using a routine perspective, and that PMO
and PfM co-evolve toward a dynamic equilibrium following an eigen behavior. Third, we
extend and contrast existing studies to highlight the benefits of shifting our attention away from
the design of PMO typologies towards the development of middle-range theories focused on
the dynamic interplay between PMO and PfM and the identification of patterns of adaptation
or co-evolution as a relevant ground for theory building.

In the following sections, the theoretical background and the conceptual framework are
discussed first. Second, the research strategy, including the case selection criteria, data
collection method and the structural analysis approach is explained. Third, six case studies are
introduced and the dynamic of co-transformation of the routines' systems elements are
analyzed. Then, the main findings are highlighted as well as the contributions made.
Limitations are discussed. Finally, some key implications vis-à-vis extant research are drawn,
and related future research underway is outlined.

2. Theoretical grounds and conceptual framework

2.1 Structural realism, organizational becoming and process-based view

An overview of the research assumptions of this study, including philosophical roots, onto-
epistemological perspective, process theories of change, as well as routines concepts is
provided in Figure 1, based on Saunders et al. 's research onion (2016, p. 124).

-----------

Figure 1: Overview of the research foundations

------------

This study is rooted in realist organizing and structural realist logic (Kilduff, Mehran &
Dunn, 2011). Indeed, it contributes to discover some fundamental structure of the phenomena,
and help to get closer to its true description and understanding. Structural realism seeks to
capture the relations of influence among elements of the routines system (see below Figure 2)
by applying mathematical methods.

4
 Considering the dynamic and transformative nature of the co-evolutionary system of
routines we investigate, we adopt an organizational becoming ontological stance (Tsoukas &
Chia, 2002). This stance is supported by a "process and temporality" view focusing on "change
and becoming" (Langley et al., 2013, p. 4), and grounded in a process-based (and recursivity
based) theory as epistemological foundation (Hernes & Bakken, 2003, p. 1516). This stance
has been reflected in a growing number of recent studies in project organizing (Packendorff,
Crevani, & Lindgren, 2014; Sergi, 2012; Vaagaasar & Andersen, 2007).

Our research relies on process theories of change (Van de Ven, 2013). In doing so, and
contemplating the suggested routine lens, we go beyond the attention on stability and repetition
emphasized by evolutionary theories (Feldman & Pentland, 2003; Knudsen, 2002; Nelson &
Winter, 1982; Pentland & Feldman, 2005; Pentland et al., 2012): we argue that systems of
routines experience Eigen behaviors: "patterns and stability arise out of recursive operations"
(Tsoukas, 2017, p. 141, Von Foerster, 1984).

2.2 A routine perspective

Going beyond the extant stream of research, we adopt a routine perspective (Salvato &
Rerup, 2011; Salvato & Rerup, 2017) as the unit of analysis and the micro foundation (Felin &
Foss, 2009; Felin, Foss, Heimeriks, & Madsen, 2012) to investigate the system of processual
relations leading to PMO / PfM co-evolution and adaptation.

However, the level of analysis is the PMO, PMO being in systemic processual interaction
with PfM and the organizational context, all seen as a collection of routines. Thus, we do not
study any "individual routine" per se, but rather the dynamic of the whole system of
"organizational routines" integrating PMO, PfM, and the organizational context (Pentland &
Feldman, 2005, p. 795).

Routines have been defined as the "repetitive, recognizable patterns of interdependent

actions, carried out by multiple actors" (Feldman & Pentland, 2003). This definition has been
widely adopted by other research (Becker, Lazaric, Nelson, & Winter, 2005; Dionysiou &
Tsoukas, 2013; Feldman & Pentland, 2008; Pentland, 2011; Pentland & Feldman, 2005, 2008).
Routines are characterized by their three constitutive elements: ostensive, performative and
artefact (Pentland & Feldman, 2005, Feldman et al., 2017)). The ostensive element or the
abstract level (Pentland et al., 2012) of a routine is its general intent, like a rule that governs a
specific behavior (Becker, 2004). The performative element of a routine or its concrete level
(Pentland et al., 2012) is made up of the specific actions performed by organizational members.

5
Finally, artefacts are the "physical manifestations of the organizational routines" like the
written rules, organizational records and procedures (Pentland & Feldman, 2005, p. 797).
Feldman & Pentland (2003) argue that in many cases the overarching pattern of a routine may
remain relatively stable while specific parts of the routine pattern may show considerable
change. Miner et al. (2008) decompose routines in their constituent parts as sub-routines: whilst
the overall pattern of a routine may remain stable, specific sub-routines may change. Also,
routine actors may choose from a range of sub-routines available for the implementation of a
given routine, giving rise to variations (Feldman, 2000). Finally, Felin et al. (2012) further
highlight the role of micro-level phenomena (individuals, social processes and structures)
which may provide sources of variation in routines. Together, these arguments explain the
duality of routines as sources of (quasi) stability and change.

PfM, as an organizational capability (Killen & Hunt, 2013), and PMO, as organizational
system can both seen as collection of routines. Indeed, according to Salvato and Rerup (2011,
p. 472), "[organizational capabilities] are collections of routines" that enable the organization
to perform its activities reliably. And, "evolutionary models envisage an organization as a
collection of routines, providing a robust micro-level unit of analysis to address adaptive
processes (Nelson and Winter, 1982)" (Salvato, 2003, p. 84).

Therefore, in the context of process theories of change (Van de Ven, 2013), selecting a
routine perspective is relevant.

2.3 A routine-based conceptual framework for the co-evolution of PMO and PfM

Building on Feldman and Pentland’s (2003) suggestion to consider the recursive relation
between the performative and ostensive aspects of organizational routines to unveil the internal
dynamics of routines, Dionysiou and Tsoukas (2013, p. 188) "suggest a process model to
account for the (re)creation of routines" based on the recursive interaction between the
performative, ostensive and artefact elements of routines. Specifically, the authors "examine
the mutual constitution of routines’ constituent parts (performative and ostensive) through
interaction, and […] develop endogenous explanations of routine (re)creation grounded on the
actions and understandings of mutually susceptible participants." (Dionysiou & Tsoukas,
2013, p. 181). Drawing on Feldman & Rafaeli (2002), they develop more precisely the nature
of the relational aspects of the interactions: "[Feldman & Rafaeli] note the significant role of
routines in specifying "connections" between participants, defined as "interactions between
people that enable them to transfer information" (2002, p. 312). Connections "enable people

6
who perform organizational tasks to develop shared understandings about what actions will
be taken in a specific routine and how these actions relate to a larger organizational picture
(Feldman & Rafaeli, 2002, p.310)" (Dionysiou & Tsoukas, 2012, p. 184).

Our conceptual framework represented in Figure 2 (for full development and justification,
see Bredillet, Tywoniak and Tootoonchy, 2017) relates organizational routine elements and
their mutual influence: PfM and PMO performative, ostensive and artefacts, and organizational
system. The relations are defined in table 1. Within this framework, the stability is in the
structure and the dynamics happen through changes in the routines. The dynamics unfold as
variations occur during the cycles of repetition of an action. Further to Figure 1, while the
process-based view is appropriate for project-related work (Packendorff, Crevani, & Lindgren,
2014; Sergi, 2012; Vaagaasar & Andersen, 2007), the recursivity-based theory (Hernes &
Bakken, 2003) clarifies the relationships between change and stability. The PMO in relation to
the organizational context and PfM can move from one stable condition to another.

-------------------

Figure 2: Conceptual framework: dynamic system of co-evolution between PMO,

PfM and Organizational Context

-------------------

In this figure "The conventional boxes and arrows of variance studies (representing
concepts and causal linkages respectively) return in new forms, wherein boxes tend to
represent states…" and the arrows indicate "the relation of precedence or the processual
relations or flows between states" (Langley et al., 2013, p. 8) of organizational routine
elements, enabling to build, a "sequence among events" (Van de Ven, 2013, p. 3).

-------------------

Table 1: Relations between PMO, PfM and Organizational Context

-------------------

3. Research strategy

Our research falls within Edmondson and McManus’s "intermediate archetype" (2007, p.
1160). Building on the conceptual framework, i.e. the dynamic system of co-evolution between
PMO, PfM and Organizational Context, we use a "comparative multiple case studies" design
(Bryman & Buchanan, 2009, p. 711) in order to test and refine, through structural analysis, and

7
further develop a provisional theory. Provisional theory building happens through the recursive
relation between the early theory, case data and the extant literature (Eisenhardt, 1989). Case
studies are “a research strategy which focuses on understanding the dynamics present within
single settings” (Eisenhardt, 1989, p. 534). Multiple case studies provide a strong infrastructure
for extending and/or building theories and the results engage in "moderatum generalizations"
(Williams, 2000, p. 215). Examining the same phenomenon in different samples enables the
cases to be compared and contrasted Indeed, "multiple cases are discrete experiments that serve
as replications, contrasts, and extensions to the emerging theory (Yin, 1994)" (Eisenhardt &
Graebner, 2007, p. 25).

3.1 Cases selection and presentation

Six cases from two project-based organizations (named below C1 and C2), located in Iran,
have been chosen as they provide a suitable organizational background to rich organizational
setting to investigate the dynamic of co-evolution of PMO and PfM. C1 and C2 are exemplar
of similar organizations in the Construction Industry, and one of the researchers had secured
excellent access to the two organizations, allowing to make sure that they would give best
information about the specific features and characteristics of the PMO/PfM routines
(Bleijenbergh, 2010).

The first organization “C1” is a large project-based organization with an established PMO
(PMO1), managing a significant number of large projects. “C1” is a private company
established in 1991 and it provides comprehensive services in engineering, procurement,
construction and management (EPCM) of infrastructure and facilities, pertaining to the oil, gas
and petrochemical industries worldwide. In PMO1, project managers report to portfolio
managers who report to the head of PMO.

The second organization, “C2”, was founded in 1975 as an international general contracting,
investor, developer and public limited company providing premier management, engineering,
procurement, construction, financing and investment services worldwide. There are five
industry groups (SPU) in this organization. Each industry group has one PMO (independent
structure and resources from the other PMOs – PMO2 to PMO6) to support their portfolios of
projects. For a given PMO, project managers report to the head of PMO.

In both organizations, most PMOs functions are recruitment, selection and assignment of
project/portfolio managers, preparing and revising PM/PfM procedures and guidelines,
projects’ quality control, managing the lessons learned database, developing the change

8
management database, unifying PM practices across the organizations, knowledge
management and PfM problem solving.

------------------

Table 2: Case studies description

------------------

3.2 Data collection methods and sources

The recommended data collection methods to capture organizational routine elements and
their relations are interviews, observation and document and archival data analysis (Feldman
and Pentland, 2008). Table 1 summarizes the methods used for this study.

------------------

Table 3: Routine elements and data collection methods

-------------------

1. Use of the framework / routines identification via interviews

In order to mitigate the limitations of interviews, different knowledgeable informants from

various hierarchical levels were invited to participate (Eisenhardt & Graebner, 2007). The
number of interviews in this qualitative research phase depends on theoretical saturation; once
the interviews are not revealing any new data, theoretical saturation has probably been reached
(Bryman & Bell, 2003). The conceptual framework was the basis for the semi-structured
interviews’ protocols (one for the portfolio managers, one for PMO members, and one for other
managers of the organization). The protocols were articulated to capture the recursive
interaction between the performative, ostensive and artefact elements of routines and the
continuous process of transformation of routines. The semi-structured interviews were
designed to capture critical incidents, actions taken and the consequences around three main
themes: 1) variation in performative elements and responses to incidents; 2) selection of
ostensive elements and artefacts; and finally, 3) retention of new elements. The target
participants were PMO staffs, the project teams and the organizational leaders. The selected
organizational routines were first checked to be within the scope of PfM and PMO. Then the
volunteer employees in PMOs and project teams were invited for the interviews. Investigated
routines were: cost / time / resource (CTR) management procedure, stakeholders’

9
recognition/management, lessons learned registration, quality control inspection, procurement
management plan, assigning the procurement expeditor, project communication management,
project control management, and inventory management.

2. After each interview, we reviewed the data collected, and compared with observations
and documents

Observation, a source of rich data, enabled to triangulate interviews and data analysis,
contributing to theoretical saturation. Documents are an important source of information to
capture the ostensive and artefacts elements of routines (Pentland, 2003a). All relevant
documents including minutes of meetings, annual/progress reports, procedures, different
databases and job descriptions (as available) were carefully analyzed.

3. Focus group meeting and structural analysis matrices

For the structural analysis, a first input (unweighted matrix of direct influences) came from
the system of organizational routine elements and the relations identification (rated "0", for no
relation, and "1" for an existing relation). The second input (weighted matrix of direct
influences) required the weighting of these relations by a few experts, as advised by Godet
(2010). We invited the interviews' participants to attend a focus group meeting. During this
meeting, participants were asked to share their ideas and discuss the "importance" or "strength"
of each relation. The participants were requested to rate each relation (“0”, “1”, “2” or “3”) one
by one and mention their reasons for their colleagues. The discussion continued until they
agreed on a number for the relation concerned.

3.3 Data analysis: structural approach

The purpose of structural analysis is to unveil the dynamic at stake within a system of
interrelated elements (the conceptual framework, in this study), the state of each elements in
this dynamic - most influential, mediating and dependent elements (Godet & Durance, 2011,
p. 63), and how the system may evolve. For this purpose, we used a structural analysis approach
called MICMAC (Matrix-based Multiplication Applied to a Classification) (Arcade et al.,
2003; Godet, 1982, 2010; Godet & Durance, 2011).

The MICMAC approach gives the opportunity to model a system as whole at a qualitative
level. It makes possible to understand the dynamic of the main elements and incorporate
qualitative information such as the nature or the perceived influence of relations. MICMAC
allows to represent the influence rank of the different elements and select the key drivers of the
system (Mirakyan & de Guio, 2015, p. 164).

10
 MICMAC principles

A system is composed of elements: Figure 3 represents a sample system with three elements,
“X”, “Y” and “Z”. The three elements of this system can be in relations. If any relation between
two elements exists, and depending on the "direction" of this relation or "influence", the value
of this relation would be a non-zero value. For example, if “X” has an influence on “Y” the
value of the “x-y” relation can be a non-zero value like “1”. The system of elements can be
defined in a matrix (“A”), in which non-zero values show the existence of a relation between
two elements. These relations between elements and the elements are the variables of the
system to be studied.

------------------

Figure 3: Sample system with three elements

------------------

Referring to what we mentioned above about the conceptual framework the arrows indicate
"the relation of precedence or the processual relations or flows between states" (Langley et
al., 2013, p. 8) of routine elements.

For the sample system, the assumptions of relations between the elements are as the
following lines:

“X” affects both “Y” and “Z”;

“Y” affects only “Z”;
“Z” affects “X”

The system matrix (3X3 matrix, aka. Matrix of Direct Influences) is:

------------------

Figure 4: Sample system matrix (A)

------------------

The structural analysis consists of three main steps (Arcade, Godet, Meunier, &
Roubelat, 2003):

Step 1. Defining the elements and their possible relations (here, the conceptual framework)

11
 Step 2. Defining the two system matrices. 1) the unweighted matrix of direct influences
(UW-MDI) where the relations are assessed “1” if any relation exists and “0” if not. 2) the
weighted matrix of direct influences (W-MDI). The "importance" or "strength" of each relation
of relations is assessed qualitatively where low importance is rated “1”, medium “2” and high
“3”.

Step 3. This phase consists in identifying the key elements, i.e. most influential, mediating,
dependent, essential to the system's development, first by using direct classification, i.e.
capturing the initial state of the system as originally defined by the routine elements and their
relation, then through indirect classification (MICMAC), i.e. capturing the end-sate of
evolution of the system resulting from the dynamic embedded in the original state of the
system. This indirect classification is obtained after increasing the power of the matrix until a
stable hierarchy of elements obtains. Comparing the hierarchy of elements in the various
classifications (direct and indirect) is a rich source of information. It enables one not only to
confirm the importance of certain elements but also to uncover certain elements which, because
of their indirect actions, play an important role (yet were not identifiable through direct
classification)1. In this study, we use the LIPSORMICMAC” software2.

Map of Influence/Dependence and interpretation

From the Step 3 direct and/or indirect classification, elements can be plotted on a two-
dimensional map, named Map of Influence/Dependence. The axes of this map are defined as
Influence and Dependence (see Figure 5Error! Reference source not found.). The Influence
score of an element is calculated by adding the rating of a line and the Dependence score is
calculated by adding the rating of a column (see Figure 4). Then the elements are ranked both
in term of relative influence and dependence according to their scores in comparison with other
elements. This map has four major quadrants (Godet & Durance, 2011).

------------------

Figure 5: Example of Map of Influence/Dependence

------------------

1 Source: Methods of prospective, Micmac, http://en.laprospective.fr/methods-of-prospective/softwares/59-

micmac.html, accessed 12 March 2017.
2 Available for download at http://en.laprospective.fr/methods-of-prospective.html, accessed 12 March 2017

12
 Quadrant 1. Non-influential Elements: These elements are neither influential nor dependent.
They have little impact on the dynamic of the system, so they can be removed from the analysis.

Quadrant 2. Resultant Elements: These elements are not influential but very dependent. The
development of these elements is highly affected by influential and mediating elements.

Quadrant 3. Mediating Elements: These elements are both highly influential and dependent;
so, they have an unstable nature and affect the dynamic of the system.

Quadrant 4. Influential Elements: The elements in this quadrant are highly influential and
independent. These elements describe the dynamics of transformation the system under study.

To illustrate using our simple system example (see Figure 5) the direct influences analysis
leads to see X as influential, Y as mediating and Z as resultant element, while after 3 iterations
reflecting the indirect influences the system reach its stable hierarchy showing X and Z as
mediating and highly dynamic elements and Y as non-influential element. The lines show the
dynamic "trajectory" of each element, i.e. their evolution and changing "states" over time.
Indeed, increasing the power of the matrix, i.e. iteration, can be apprehended as an evolution
of the processual system over time (Schreyögg & Sydow, 2011).

Coming back to our conceptual framework, any variation in the state of an organizational
routine element and/or in a processual relation between two routine elements can cascade
throughout the system and lead to direct and indirect variations over time, and to co-evolution
resulting in new conditions (Langley, 1999) for PfM and PMO. Overall, the dynamic system
of co-evolution between PMO, PfM and Organizational Context is formed of seven elements;
therefore, the system is modelled with a 7X7 matrix. The direct influence of each element on
itself is zero; 42 relations (arrows) can be possibly investigated because each element in the
system can affect the remaining six elements. However, only 19 relations had been identified
in the conceptual framework. Applying a MICMAC analysis to the system of organizational
routines will allow to shed light on why and how the system change over time, and on the role
of the elements and relations in this evolution.

4. Findings and Discussion

We present and discuss the findings of the analyses conducted for the six
PMO/PfM/Organizational context systems. First, we compare the use of unweighted and
weighted matrices as support for highlighting and making sense of the dynamic of the system.
Second, we examine the dynamic of evolution of the six systems, showing why and how the

13
state of elements and relations between them may lead to a specific dynamic of evolution, a
"trajectory" from an initial state to a relatively stable end state of the systems. Third, we further
deepen the investigation, suggesting that the systems follow an Eigen behavior3, a
"convergence towards an equilibrium" (Tsoukas, 2017, p. 141).

4.1 Comparing the use unweighted and weighted matrices

We use PMO1 as the basis for comparing the use of unweighted and weighted matrix as a
support for the interpretation of the PMO/PfM/Organizational context system dynamics.

4.1.1 Contrasting the matrices of direct influences

After the structural analysis steps described above in section 3.3, the conceptual framework,
the system of elements and their relations as defined in figure 2, is the foundation of the analysis
(step 1).

Two system matrices were drawn from this framework, 1) the unweighted matrix of direct
influences (UW-MDI) (table 4), and 2) the weighted matrix of direct influences (W-MDI)
(table 5). The weighting process is explained above in section 3.3 and was the result of focus
group meetings.

------------------

Table 4: The unweighted and weighted matrices of direct influence for PMO1

-------------------

Each matrix (UW-MDI and W-MDI) allows to identify the state of each elements using a
map of influence/dependence (step 3).

Figure 6 shows the maps resulting of the direct classification of each element for the
unweighted matrix of direct influence (UW-DI) and weighted matrix of direct influence (W-
DI), i.e. for the initial state of the system.

------------------

3 Eigen value stands for ‘self-value’ in German. "Eigen values emerge from the continuous sequence of
recursive operations; they represent the convergence towards an equilibrium point – a state of stability. […] If the
primary variable at hand is behaviour, we can speak of ‘eigen behaviour’. An eigen behaviour represents the
stability that arises from ongoing recursive operations." (Tsoukas, 2017, p. 140).

14
 Figure 6: Unweighted and weighted maps of direct influence for PMO1

------------------

We notice that the weighting process doesn't lead to change in state for the elements
"Organizational context" (O) and PMO-Artefacts (Pa) both being influential (quadrant 4) and
PMO-Ostensive (Po) (resultant, quadrant 2). We observe change in state for PfM-Artefacts
(Fa) (from non-influential (quadrant 1) for UW to influential (quadrant 1) for W matrix, for
PfM-Ostensive shifting from influential (quadrant 4) for UW to non-influential (quadrant 1)
for W matrix) and for both PfM-Performative and PMO-Performative, moving from mediating
(quadrant 3) for UW to resultant (quadrant 4) for W matrix.

4.1.2 Contrasting the matrices of indirect influences

The comparison for indirect classification after increasing the power of the matrix until a
stable hierarchy of elements is obtained (i.e. the end-state of the system’s dynamic) is shown
in figure 7.

------------------

Figure 7: Unweighted and weighted maps of indirect influence for PMO1

------------------

The results show that the weighting process does not lead to changes of end-states for PfM-
Artefacts and PfM-Ostensive who were both non-influential (quadrant 1), and for PMO-
Performative (mediating, quadrant 3). We notice a change of end-state for PMO-Artefacts
(non-influential (quadrant 1) for UW to influential (quadrant 4) for W matrix), for
Organizational context (mediating (quadrant 3) for UW to influential (quadrant 4) for W
matrix, for PfM-Performative (mediating (quadrant 3) for UW to resultant (quadrant 2) for W
matrix, and for PMO-Ostensive (resultant (quadrant 2) for UW to non-influential (quadrant 1)
for W matrix).

4.1.3 Key observations

Comparing and contrasting the use of unweighted and weighted matrices of direct and
indirect influences lead to some observations.

First, grasping the inner structure system enables to understand the dynamic and potential
of evolution it carries structurally. The unweighted matrix captures the inner structure

15
PMO/PfM/Organizational context system and its potential dynamic, i.e. the processual
relations and flows of information between states of routine elements. The unweighted map of
direct influence capture therefore the initial state, or initial condition, of the system and of the
state of each routine element reflecting the structural relationships between them, while the
unweighted map of indirect influence allows to unveil the end-state of the system and the end-
state of each routine element resulting from the inner structure. In a way, it helps to observe
the actualization of the potential dynamic of evolution of the system embedded in its inner
structure.

Second, weighting matters. The weighted matrix, while based on the system inner structure,
reflects the shared perception(s) of the actors about the strength and/or the importance of the
relations between the state of the routine elements and not only the presence of relations. Thus,
a weighted matrix embeds the quality of the relations between states of routines elements as
perceived by the organizational actors, providing richer information about the system.
Therefore, in the following part of the analysis we use weighted matrices only.

Third, considering both direct and indirect influences analyses allow to capture the change
of states of routines elements along the system evolution process. Indeed, the direct influence
analysis captures the initial state of the PMO/PfM/Organizational context system and the
respective states of routine elements as initial conditions for the dynamic of evolution (why
the system evolves), while the indirect influence analysis highlights the characteristics of the
relatively stable end-state of the system and the way each routine element state develops along
the system evolution (how the system evolves).

In the next section, we compare the evolution of the PMO/PfM/Organizational context

systems reflected in our six cases studies, based on weighted matrices of direct and indirect
influence.

4.2 . Analysis of the dynamic evolution of the systems

For each of the six case studies, we investigate the various aspect of the evolution of the
PMO/PfM/Organizational context system from an initial state as described by the weighted
matrix of direct influence to an end-state, where the system shows a relatively stable state, i.e.
till reaching a stable hierarchy of elements regarding their degree of influence/dependence.

Table 5 offer a summary of the analyses and the results and implications are discussed
below.

16
 In the discussion, we are focusing only on the common patterns emerging from the analyses.
The idiosyncratic variations between the cases are not considered, falling outside the scope of
this study. Furthermore, we use the word state – following (Langley et al., 2013, p. 8) cited
above – in order to describe both an active behavior played by a routine element, and its defined
condition at a given state of the PMO/PfM/Organizational context system. This dual
characterization is provided by the routine element location in the map of
influence/dependence.

------------------

Table 5: Summary of the analyses for the six systems

-------------------

First, the systems reach their end-state, a state of relative stability, after 4 to 6 iterations,
meaning that after this, there is no change in the state of the 7 routine elements and their states
(influential, mediating, resultant or non-influential). The difference between the number of
iterations means that systems requiring fewer iterations to reach stability were less subject to
emerging change behaviors, and that the cascading impact of changes in the state of elements
was dampened. For instance, reaching an end-state after 4 iterations means that the longest path
of indirect influence throughout the system is 4: the longest chain of elements without going
two times through the same elements is 4. For instance, for PMO4 system, an evolution of the
state of Organizational context (O) affects directly PMO-Artefacts (Pa) and flows indirectly to
affect PMO-Performative (Pp) in a second stage, then PfM-Performative (Fp) (third stage) and
finally PMO-Ostensive (Po) routine element in fourth stage. This exemplifies the longest chain
for PMO4 system (4 relations between "elements" – excluding non-influential "elements", by
definition).

Second, the comparison between the state of each routine element in the initial state and the
end-state of the PMO/PfM/Organizational context system provides insights about the way the
systems evolve.

The state shifts of the routines elements during the evolution of the system, from initial state
to end-state, is summarized in figure 8 (symbolized by the arrows). As shown in table 5,
examples of patterns of evolution are: Organizational context (O) remains stable in quadrant 4
(influential); PMO-Artefacts (Pa) evolves from quadrant 4 (influential) to quadrant 1 (non-
influential).

17
 The following patterns are identified. Organizational context (O) keeps its influential state
throughout the evolution of the system. PfM-Artefacts (Fa) and PfM-Ostensive (Fo) elements
remain non-influential regarding the evolution of the system. PMO-Performative (Pp) and
PfM-Performative (Fp) elements shift from a resultant state to a mediating state. PMO-
Artefacts (Pa) shifts from an influential state to a non-influential state and PMO-Ostensive (Po)
shifts from a non-influential state to a resultant state.

------------------

Figure 8: State shifts of the routines elements during the evolution the system

------------------

The evolution of the system and the shift in the states of routines elements calls for five
main comments.

First, the internal dynamic of the system in its initial state is based on the influential state of
both Organizational context and PMO-Artefacts, both providing the initial conditions for
further evolution, and influencing both PMO-Performative and PfM-Performative resultant
routine elements. The performative routines both at PMO and PfM level depends on the
information provided by the organizational context and the set of rules or guidelines provided
by the PMO (i.e. artefacts). For example, PMO1 prepared a form for controlling the resources
of projects and the time that the resources should be deployed in the projects’ execution time
(“Pp-Fp”). This form was called Top Execution Plan (“TEP”). The leader of a mega project
used TEP and mentioned in several meetings that using it, he could better control the resources.
Subsequently, the PMO1’s head realized that the projects using the TEP form faced fewer
issues and the clients were more satisfied with the projects’ performances (“Fp-Fo”). The TEP
procedure was officially developed (“Pa-Fa”) and all the project/portfolio managers were
requested to follow the instructions (“Fo-Fp”).

Second, when the system evolved to its end-state, the Organizational context remained
influential. But the PMO-Performative and PfM-Performative routines elements were taking a
mediating state in the dynamic of the system. Indeed, while being influenced by the
organizational context, these performative routines affected the emerging resultant PMO-
Ostensive routine element, i.e. they were captured at PMO level as schematic forms of routines
at the end-sate of evolution of the system. In the meantime, PMO-Artefacts became non-
influential. In its end-sate, the system, embedding the set of rules or guidelines provided by

18
PMO in its performance, relied on the mediating state of routine performative elements both at
PMO and PfM levels. In our example (TEP), although all the project managers were requested
to fill TEP based on the aforementioned procedures, every project manager had a different way
to fill out the form (“Fa-Fp”). Some of them use the projects’ WBS and others used Excel
worksheets to divide the activities and spread them over; these project managers enacted
routines in different ways. Three successful projects at the end of the year were selected to
identify the best practices. The TEP procedure was updated based on the best observed
practices. This method of filling the TEP became a norm among some other project managers
as well (“Fp-Fo”). The PMO1’s staff then ran several meetings to let all the project managers
know the best method to fill the TEP (arrow “Pp-Po”). After a few months, a new version of
the TEP procedure and form based on the organisation’s best practices was issued (“Pa-Fa”).

Figure 9 summarizes the key states of routine elements at the initial state and end-state of
the system.

------------------

Figure 9: Key states of routines elements at initial and end-state of the system

------------------

Third, studying the evolution of a system from an initial state to an end-state involves a
clarification of what we study. The initial state is defined by the matrix of direct influence,
which captures the relations between the various state of routine elements, i.e. the direct
relations between these elements. In the case of PMO4, O affects directly Pa. Multiplying the
matrix by itself (an iteration) means moving through a first stage of evolution, apprehending
how elements affect each other following a two-step path. In case of PMO4, O affects Pp (chain
O > Pa > Pp). Once the end-state is reached any further iteration won't lead to a change in the
state of the routines elements. Only a structural transformation of the relations (adding some,
removing other, or modifying their strength or importance for instance) may lead to a new
evolution of the system starting from new initial conditions i.e. new matrix of direct influence.
Aa a consequence, we make clear that we investigate the change behavior of both the
PMO/PfM/Organizational context system and of each routine element within a defined
structure of relations and initial state of routines elements until it reaches its full evolution
toward a relatively stable end-state: we do not study the structural transformation of the system,
but the evolution of a given system for a given structure.

19
 Fourth, a direct implication of the former is that a given system carries its own potential of
evolution in its structure, i.e. the matrix of direct influence. The reason why a system moves
according to a specific pathway is "written" in its structure of relations between state of routines
elements.

Fifth, an iteration, in this study, is a logical stage, and is not related to any particular time-
span. If the approach allows to understand how a system and the state of the routine elements
evolve from one iteration to another or from the initial state to the end-state, it doesn't say
anything about temporality of the change process.

In the next section, we further deepen the investigation, suggesting that a given
PMO/PfM/Organizational context system follows an Eigen behavior from its initial state to its
end-state, a [dynamic] "stability that arises from recursive ongoing operations" (Tsoukas,
2017, p. 141).

4.3 Eigen behavior

A mentioned above the PMO/PfM/Organizational context system, for a given structure of

relations between states of routine elements, evolves from an initial state to a relatively steady
end-state. The system embraces "stability and change, routine and novelty are interwoven
(Dionysiou and Tsoukas, 2013; Sonenshein, 2016; Tsoukas and Chia, 2002)" (Tsoukas, 2017,
p. 138). The PMO/PfM/Organizational context system can be seen as a "nontrivial machine"
(NTM) (von Foerster, 1984).

"A nontrivial machine keeps changing its rule of transformation: ‘a response once
observed for a given stimulus may not be the same for the same stimulus given later’
(von Foerster, 1984, p. 10). The critical difference from a trivial machine is that the
nontrivial machine has an internal state that keeps changing" (Tsoukas, 2017, p. 140).

While the image of nontrivial machines may suggest unpredictability, "For von Foerster,
patterns and stability arise out of recursive operations. When a NTM reprocesses what it has
already produced, circular behavior is created, which gives rise to stability." (Tsoukas, 2017,
p. 141). They experience an Eigen behavior. Furthermore, "seeing organizations as NTMs
enables us to see routines as stable and variant, enduring and changing (Deken et al., 2016;
Dionysiou and Tsoukas, 2013; Sonenshein, 2016)". Furthermore, the concept of Eigen behavior
is inseparable from conjunctive thinking and "taking complexity seriously" (Tsoukas, 2017, p.
143). Other process views explaining the dynamics towards some organizational routines’
stability – Organizational path dependence, Organizational imprinting, Commitment and sunk

20
cost, Structural inertia, Institutional persistence – do not offer such relations to inherent
organizing complexity (Schreyögg & Sydow, 2011, p. 331).

These developments echo with our study. In order to highlight the characteristics of the
Eigen behavior of the six PMO/PfM/Organizational context systems, we calculate, for each
weighted matrix of direct influence the characteristic polynomial, the Eigenvalue and the
Eigenvectors.

The characteristic polynomial of a square matrix (in our case, W-MDI) is an invariant under
matrix transformation (here the iterations). The Eigenvalues (λ) are the roots of the
characteristic polynomial, where Av = λv, A being the matrix of transformation (here, W-
MDI), and v being the Eigenvector. Eigenvalues are used for stability analyses.

If Eigenvalue is zero, the system remains position or amplitude constant.

If any Eigenvalue has a positive real part, the system tends to move away from a fixed point
(unstable system).

If any Eigenvalue has a negative real part, the system tends to move back to a steady state
(stable system).

If any Eigenvalue has an imaginary part, and a negative real part, the system oscillates
around a steady state (stable oscillating system).

The system moves to a steady state if Eigenvalues are negative real numbers or converge
towards a limit cycle behavior to oscillate around a steady state if Eigenvalues are complex
numbers with negative real parts (Ogata, 2010).

Where Eigenvalues indicate the scale of transformation, Eigenvectors specify the directions
of transformation. Table 6 summarizes the results of these calculations for the six systems4.

------------------

Table 6: Summary of the analyses for the six systems

-------------------

Considering our six cases, each system has Eigenvalues with both positive and negative real
parts as well as imaginary parts.

4 Calculations are made using Matlab 2016

21
 The positive real part of Eigenvalues is related to the initial dynamic behavior and move
away from the initial state, while the Eigenvalues’ negative real part or imaginary part (i.e. the
Eigenvalue is a complex number) is associated to a steady end-state (stable system). And when
the Eigenvalue has an imaginary part and a negative real part, the system oscillates around a
steady state.

Due to the dynamic AND stability associated to Eigenvalues, it can be inferred that the
systems have a dynamic behavior but reach a stable condition at some point, i.e. after a number
of iterations. The different range of Eigenvalues (positive, negative and real, complex) supports
the fact that systems move to steady state according to different trajectories, and in our case,
the dynamic path of each routine elements in the map of influence/dependence through the
various iterations.

Based on our findings, systems are dynamic during the first 2 or 3 iterations, then start to
reach a steady state after 3 to 6 iterations showing oscillation around a state of equilibrium.
Figure 10 illustrates this Eigen behavior.

------------------

Figure 10: PMO/PfM/Organizational context system Eigen behavior and routines

elements trajectories (PMO6 as example)

------------------

Besides exemplifying why and how an Eigen behavior occurs within a system of
interactions between routine elements, we argue that our findings contribute also to account for
complexity. Tsoukas relates organizations as NTMs and complexity of organizational behavior
with what he calls "conjunctive theorizing", i.e. "taking complexity seriously" (Tsoukas, 2017,
p. 143). This view involves seeking rich accounts (see also for instance Tsoukas & Hatch,
2001). In this study, investigating complex project organizing phenomena, we illustrate how a
conjunctive approach, interweaving a process-based approach, a routine perspective, and case
studies and structural analysis, may lead to illuminate what second-order project management
means (Saynisch, 2010a, 2010b). Furthermore, we open a new perspective on how to capture
project organizing complexity through the structural analysis of relations between routine
elements systems and characterization of these systems and their dynamic of evolution by
Eigenvalues. While not the primary purpose of this study, this contribution opens the way to

22
elegantly solve the unique conundrum of project organizing complexity (Geraldi, 2009; Geraldi
et al., 2011), moving beyond lists of factors to capture its relational and socio-material practice
essence.

5. Conclusion

In this paper, we draw on extant research on PMO studies, taking an innovative process
perspective, relevant for inquiring into the dynamic transformative and co-evolutionary nature
of PMO in relation to PfM and the organizational context. Starting from a becoming ontology
(Langley, Smallman, Tsoukas, & Van de Ven, 2013; Tsoukas & Chia, 2002) we argue why
and how PMO and PfM co-transform.

Our findings suggest the reasons of evolution (why) lie in the initial state of the routines
system as figured by the W-MDI (e.g. table 4). The development of the routine system and the
changing states of the routines elements from the initial state to the end-state uncovers how the
whole system evolves (how).

5.1 Main findings and contributions

5.1.1 Main findings

The following summarize the main findings with regards to the theorization of dynamic of
change, the main purpose of this paper.

First, whilst we acknowledge some differences between the paths of co-transformation for
the six cases, the results unveil important commonalities. In this paper we focus on
commonalities arising from different cases.

Second, during the transformation of the routines system, the routine elements states evolve
and may move from one quadrant to the other in the map of influence. The trajectory of the
elements shows that none of the elements move between three or all the four quadrants. After
four to six iterations all the systems reach a dynamic stability i.e. an eigen behavior rather than
a chaotic behavior.

Third, the changing state of routine elements at different stage of transformation of the
routines system lead them to play different roles. Two routines elements were non-influential
(Fa, Fo). The organizational context (O) remained influential throughout the transformation
from the initial to the end state of the routines system. However, the role of other elements
changed during the transformation process. The PMO artefact element (Pa) played an
influential role in the initial state but was non-influential in the end state. The changing role of

23
PMO and PfM performative elements (Pp, Fp) from resultant in the initial state (the
performative routines driven by the influence of the organizational context and the PMO
artefacts) to mediating role in the end-state, explaining why and how and why the
organizational context (O) influenced the PMO ostensive element (Po). Further, if Po was a
resultant element in end-sate, it was a non-influential element in the initial state of the routines
system.

The trajectory of evolution of the routine systems observed was consistent with an
organizational sense-making process (Weick, 1979, 1995, Dionysiou & Tsoukas, 2013): a
change in the organizational context triggered the introduction of a new routine artefact, which
then generated a series of experiments to discover effective performances, which were finally
captured in an ostensive definition – “How can I know what I think until I see what I say?”
(Weick, 1979).

Beyond these findings the research makes the following contributions.

5.1.2 Conceptualizing PMO and PfM as a collection of routines

The findings from our study led to interpretations confirming Salvato and Rerup’s (2011)
view. PMO, as an organizational structure, can be theorized as a collection of routines. In
addition, PfM as an organizational capability (Bredillet, Tywoniak, & Tootoonchy, 2017) can
be abstracted as a collection of routines. Conceptualizing both PMO and PfM as a collection
of routines helps us to understand the co-evolution of PfM in relation to PMO and the
organizational context.

5.1.3 The co-evolution of PMO and PfM

By analyzing the initial state of the elements of the PMO/PfM/Organizational context

routines system and its characteristics, we uncover why and how the system transforms. The
influent role of the organizational context, through the initial to the end state of the routines
system, highlight the driving force underlying PMO / PfM reconfiguration according to their
purpose (Pellegrinelli & Garagna, 2009). The initial state of the routines system (UW/W-MDI)
reveals why PMO and PfM co-evolve to address organizational needs. The transformation of
the routines system and changing states of its routine elements show how a given system may
evolve toward a dynamic equilibrium following an eigen behavior.

5.1.4 Eigen behavior

24
 Based on the weighted matrix of direct influence (W-MDI) representing the routines system,
we can calculate the eigen values of the systems. Depending on the eigen values, we can infer
that the PMO/PfM/organizational context routines system may be abstracted as ‘nontrivial
machine’ (Tsoukas, 2017), exhibiting an eigen behavior. Furthermore, the systems in all the
cases reach to dynamically stable state after four to six iterations, using structural analysis
(MICMAC). The eigen values analyses are leading to the same conclusions, providing however
additional information about the level of complexity and the dynamic trajectory (eigen
behavior) of each system (eigen value number).

5.1.5 The structural analysis approach as the research strategy

The structural analysis (MICMAC) approach allowed to consider the initial-state, the co-
transformation process and the end-state of the routines system. The most important advantage
of applying structural analysis was the ability to simulate the dynamics of the system in regard
to various assumptions made on the strength of the relations between routines elements. In
addition, the iterations unveiled the changing states and roles of the routine elements of a given
system, allowing to understand which elements are at stake in various stage of development.

5.1.6 Complexity characterization

Eigen values have been used in this research to determine the stability or eigen behavior of
the routines system. Thus, the system's complexity can be characterized by the meaning of
eigen values. If the imaginary part of the eigen values is non-zero, the trajectory of the system's
transformation may follow different paths but will finish spiraling around an end-state, i.e. an
oscillating behavior exhibiting dynamic stability around an end-state.

5.2 Limitations

Limitations apply to this research.

First, we stated in section 4.2, that "the idiosyncratic variations between cases are not
considered". We are fully aware that variations due to idiosyncratic variations as noise can be
highly significant. However, in the context of this study, noise does not affect the relational
aspects of the interactions (see § 2.3) or the overall structure of connections: therefore, it does
not play a significant role. Moreover,

"From a mathematical viewpoint, idiosyncratic variations represent high dimensional

information that Eigen analysis recovers with a large number of small eigenvalues,
whereas coordinated variations represent low-dimensional information that Eigen

25
 analysis recovers with a small number of large eigenvalues. Hence the (large)
eigenvalues […] selectively capture pattern, whereas the late (small) eigenvalues
selectively recover noise". (Gauch & Zobel, 1988, p. 3).

This analysis focuses on patterns (the focus of this paper), and not large numbers of small
eigenvalues enabling to capture the role of noise.

Second, the evolution of a given routines system between the initial and the end-state is
investigated for a given structure of the relations between routine elements. A change in the
structure (e.g. adding a relation between routine elements) would lead to a new investigation.

Third, the number of iterations, part of the structural analysis process, is not related to the
time span of adaptation to change of state. Duration of adaptation to a change of state has not
been taken into consideration. Iterations are here the mathematical stages of the behavior of the
system.

Fourth, an important limitation of applying structural analysis is the inter-subjective nature

of this method (Godet, 2010); structural analysis is not the whole reality but just a new lens to
look at the most important elements and relations of a modelled system as agreed by
informants. This inter-subjective nature is related to the sensitivity of the structural analysis
matrix. The results of the MICMAC method showed how the perceptions of a group about
organizational hierarchy could lead to different paths of transformation.

Fith, this study offers an explanatory and interpretative approach; therefore, we aim for
analytical generalization and not statistical generalization (Yin, 1994).

5.3 Concluding comments & further research

The findings of the study support, add and challenge some of the assumptions of past
research (Alvesson & Sandberg, 2011). The findings support former results, i.e. that PMOs
have a dynamic transformative nature (Bates, 1998; Dai & Wells, 2004; Hobbs & Aubry, 2006,
2007, 2010; Santosus, 2003; Turner & Keegan, 2001). The research adds some new findings
to the extant literature. Indeed, conceptualizing PMO and PfM as sets of routines sheds light
on the process of their dynamic co-transformation. Furthermore, this study offers a fresh way
to characterize the complexity and the dynamic of co-evolution of a PMO/PfM/organizational
context system, based on its eigen values and related eigen behavior. In addition, the findings
of our research challenge some assumptions of extant literature. Former research (Hobbs,
Aubry, & Thuillier, 2008) investigated the unstable nature of PMOs and the difficulty to
uncover any patterns of co-evolution. Our research sheds light on possible patterns of

26
behaviors, in relation to the changing states of the routines elements as uncovered by the
structural analysis process and as unveiled by the eigen values of routine systems matrixes, for
the co-evolution of a routines system.

Finally, two main directions are currently underway. One is investigating the dynamic co-
transformation process considering a broader routines system including different levels of
PMOs (e.g. strategic and operational PMOs) and broader range of industries. The second
direction is considering complexity characterization, using eigen value of the
PMO/PfM/Organizational context system and for other capabilities related to PMOs in order
to study if we can find similar patterns of co-evolution in other systems. The purpose is to
categorize the systems, and support the development of middle-range theories focused on the
dynamic interplay between PMO, PfM or other capabilities and the organizational context.

References:

Alvesson, M., & Sandberg, J. (2011). Generating research questions through problematization.
Academy of Management Review, 36(2), 247-271.
Arcade, J., Godet, M., Meunier, F., & Roubelat, F. (2003). Structural analysis with the
MICMAC method & actor's strategy with MACTOR method. AC/UNU Millennium
Project: Futures Research Methodology.
Archer, N. P., & Ghasemzadeh, F. (1999). An integrated framework for project portfolio
selection. International Journal of Project Management, 17(4), 207-216.
Aubry, M. (2015). Project Management Office Transformations: Direct and Moderating
Effects That Enhance Performance and Maturity. Project Management Journal, 46(5),
19-45.
Aubry, M., & Hobbs, B. (2011). A fresh look at the contribution of project management to
organizational performance. Project Management Journal, 42(1), 3-16.
Aubry, M., Hobbs, B., & Müller, R. (2010). Identifying forces driving PMO changes. Project
Management Journal, 41(4), 30-45.
Aubry, M., Hobbs, B., & Thuillier, D. (2007). A new framework for understanding
organisational project management through the PMO. International Journal of Project
Management, 25, 328–336.
Aubry, M., Müller, R., Hobbs, B., & Blomquist, T. (2010). Project management offices in
transition. International Journal of Project Management, 28(8), 766-778.
Becker, M. C., Lazaric, N., Nelson, R. R., & Winter, S. G. (2005). Applying organizational
routines in understanding organizational change. Industrial and Corporate Change, 14,
775–791.
Biesenthal, C., & Wilden, R. (2014). Multi-level project governance: Trends and opportunities.
International Journal of Project Management, 32(8), 1291-1308.

27
Boulton, J. G., Allen, P. M., & Bowman, C. (2015). Embracing Complexity: Strategic
Perspectives for an Age of Turbulence. Oxford: Oxford University Press.
Buchanan, D. A., & Bryman, A. (Eds.). (2009). The Sage Handbook of Organizational
Research Methods. London: Sage.
Bredillet, C. N., & Tywoniak, S., & Tootoonchy, M. (2017). Exploring the dynamics of project
management office and portfolio management co-evolution: A routine lens.
International Journal of Project Management.
Bryman, A., & Bell, E. (2003). Business Research Methods. Oxford: Oxford University Press.
Cattani, G. (2011). Project-Based Organizing and Strategic Management: Advances in
Strategic Management. Emerald Group Publishing Limited.
Crawford, J. K. (2002). The Strategic Project Office : A Guide to Improving Organizational
Performance. New York: Marcel Dekker.
Denis, J.-L., Langley, A., & Rouleau, L. (2007). Strategizing in pluralistic contexts: Rethinking theoretical
frames. Human Relations, 60(1), 179-215.
Desouza, K. C., & Evaristo, J. R. (2006). Project management offices: a case of knowledge-
based archetypes. International Journal of Information Management, 26(5), 414-423.
Dietrich, P., & Lehtonen, P. (2005). Successful management of strategic intentions through
multiple projects-Reflections from empirical study. International Journal of Project
Management, 23(5), 386-391.
Dionysiou, D. D., & Tsoukas, H. (2013). Understanding the (re)creation of routines from
within: a symbolic interactionist perspective. Academy of Management Review, 38(2),
181–205.
doValle, J. A., da Silviera, W., & Soares, C. A. P. (2008). Project management office (PMO)
– principles in practice. ACE International Transactions, PM 07, 1-9.
Drouin, N., Müller, R., & Sankaran, S. (Eds.). (2013). Novel approaches to organizational
project management research: Translational and transformational (Vol. 29).
Copenhagen Business School Press DK.
Eisenhardt, K. M. (1989). Building theories from case study research. Academy of Management
Review, 14(4), 532-550.
Eisenhardt, K. M., & Graebner, M. E. (2007). Theory building from cases: Opportunities and
challenges. the Academy of Management Journal, 50(1), 25-32.
Feldman, M. S. (2000). Organizational routines as a source of continuous change.
Organization Science, 11, 611–629.
Feldman, M. S., & Rafaeli, A. (2002). Organizational Routines as Sources of Connections
and Understandings. Journal of Management Studies, 39(3), 309-331.
Feldman, M. S., & Pentland, B. T. (2003). Reconceptualizing organizational routines as a
source of flexibility and change. Administrative science quarterly, 48, 94-118.
Feldman, M. S., & Pentland, B. T. (2008). Issues in Empirical Field Studies of Organizational
Routines. In M. C. Becker (Ed.), Handbook Of Organizational Routines: Edward Elgar
Publishing.

28
Feldman, M. S., Pentland, B. T., D'Adderio, L., & Lazaric, N. (2016). Beyond Routines as
Things: Introduction to the Special Issue on Routine Dynamics. [Article]. Organization
Science, 27(3), 505-513.
Felin, T., & Foss, N. J. (2009). Organizational routines and capabilities: Historical drift and a
course-correction toward microfoundations. Scandinavian journal of management,
25(2), 157-167.
Felin, T., Foss, N. J., Heimeriks, K. H., & Madsen, T. L. (2012). Microfoundations of routines
and capabilities: Individuals, processes, and structure. Journal of management studies,
49(8), 1351-1374.
Gauch, H. G., & Zobel, R. W. (1988). Predictive and postdictive success of statistical analyses
of yield trials. Theoretical and Applied Genetics, 76(1), 1-10.
Geraldi, J. G. (2009). What complexity assessments can tell us about projects: dialogue
between conception and perception. Technology Analysis & Strategic Management,
21(5), 665-678.
Geraldi, J., Maylor, H., & Williams, T. (2011). Now, let's make it really complex
(complicated): A systematic review of the complexities of projects. International
Journal of Operations & Production Management, 31(9), 966-990.
Godet, M. (2010). La prospective. Retrieved 25/10/2015, from
http://en.laprospective.fr/methods-of-prospective/softwares/59-micmac.html.
Godet, M., & Durance, P. (2011). Strategic Foresight for Corporate and Regional
Development. Paris: DUNOD–UNESCO–Fondation Prospective et Innovation.
Hernes, T., & Bakken, T. (2003). Implications of Self-Reference: Niklas Luhmann’s
Autopoiesis and Organization Theory. Organization Studies, 24(9), 1511-1535.
Hill, G. M. (2004). Evolving the project management office: a competency continuum.
Information Systems Management, 21(4).
Hobbs, B., & Aubry, M. (2006). Identifying the structure that underlies the extreme variety
found among PMOs. Paper presented at the PMI research conference, Montreal,
Canada.
Hobbs, B., & Aubry, M. (2007). A multiphase research program investigating project
management offices (PMOs): The results of phase 1. Project Management Journal,
38(1), 74-86.
Hobbs, B., Aubry, M., & Thuillier, D. (2008). The project management office as an
organisational innovation. International Journal of Project Management, 26(5), 547–
555.
Hurt, M., & Thomas, J. L. (2009). Building value through sustainable project management
offices. Project Management Journal, 40(1), 55-72.
Kerzner, H. (2003). Strategic planning for a project office. Project Management Journal, 34(2),
13-25.
Knudsen, T. (2002). Economic selection theory. Journal of evolutionary economics, 12, 443-
470.
Langley, A. (1999). Strategies for theorizing from process data. Academy of Management
Review, 24(4), 691–710.

29
Langley, A., Smallman, C., Tsoukas, H., & Van de Ven, A. H. (2013). Process studies of
change in organization and management: unveiling temporality, activity, and flow.
Academy of Management Journal, 56(1), 1-13.
Miner, A. S., Ciuchta, M., & Gong, Y. (2008). Organizational Routines and Organizational
Learning. In Organizational Routines and Organizational Learning. Cheltenham, UK:
Edward Elgar Publishing.
Mirakyan, A., & de Guio, R. (2015). Three Domain Modelling and Uncertainty Analysis:
Applications in Long Range Infrastructure Planning: Springer International Publishing.
Nelson, R. R., & Winter, S. G. (1982). An Evolutionary Theory of Economic Change.
Cambridge, MASS: Belknap Press.
Ogata, K. (2010). Modern Control Engineering (5th ed.). Boston, London: Pearson
Packendorff, J., Crevani, L., & Lindgren, M. (2014). Project leadership in becoming: A process
study of an organizational change project. Project Management Journal, 45(3), 5-20.
Pellegrinelli, A., & Garagna, L. (2009). Towards a conceptualisation of PMOs as agents and
subjects of change and renewal. International Journal of Project Management, 27(7),
649-656.
Pentland, B. T. (2011). The foundation is solid, if you know where to look: comment on Felin
and Foss. Journal of Institutional Economics, 7(2), 279-293.
Pentland, B. T., & Feldman, M. S. (2005). Organizational routines as unit of analysis. Industrial
and Corporate Change, 14, 793–815.
Pentland, B. T., & Feldman, M. S. (2008). Designing routines: On the folly of designing
artifacts, while hoping for patterns of action. Information and Organization, 18, 235–
250.
Pentland, B. T., Feldman, M. S., Becker, M. C., & Liu, P. (2012). Dynamics of Organizational
Routines: A Generative Model. Journal of Management Studies(8), 1484-1508.
Salvato, C. (2003). The Role of Micro‐Strategies in the Engineering of Firm Evolution. Journal
of management studies, 40(1), 83-108.
Salvato, C., & Rerup, C. (2011). Beyond collective entities: Multilevel research on
organizational routines and capabilities. Journal of Management, 37(2), 468-490.
Salvato, C., & Rerup, C. (2017). Routine Regulation: Balancing Conflicting Goals in
Organizational Routines. Administrative Science Quarterly, 0001839217707738.
Retrieved from http://dx.doi.org/10.1177/0001839217707738.
doi:10.1177/0001839217707738
Sankaran, S., Müller, R., & Drouin, N. (Eds.). (2017). Cambridge Handbook of
Organizational Project Management. Cambridge University Press.
Saynisch, M. (2010a). Beyond frontiers of traditional project management- An approach to
evolutionary, self-organizational principles and the complexity theory—results of the
research program. Project Management Journal, 41(2), 21-37.
Saynisch, M. (2010b). Mastering complexity and changes in projects, economy, and society
via Project Management Second Order (PM‐2). Project Management Journal, 41(5),
4-20.
Schreyögg, G., & Sydow, J. (2011). Organizational Path Dependence: A Process View.
Organization Studies, 32(3), 321-335.

30
Sergi, V. (2012). Bounded becoming: Insights from understanding projects in situation.
International Journal of Managing Projects in Business, 5(3), 345–363.
Shenhar, A. J., Dvir, D., Levy, O., & Maltz, A. (2001). Project success: a multidimensional
strategic concept. Long Range Planning, 34(6), 699-725.
Singh, R., Keil, M., & Kasi, V. (2009). Identifying and overcoming the challenges of
implementing a project management office. European Journal of Information Systems,
18, 409-427.
Smith, W. K., & Lewis, M. W. (2011). Toward a Theory of Paradox: A Dynamic equilibrium Model of
Organizing. Academy of Management Review, 36(2), 381-403.
Sonenshein, S. (2016). Routines and Creativity: From Dualism to Duality. Organization
Science, 27(3), 739-758.
Stanleigh, M. (2006). From crisis to control: New standards for project management. Ivey
Business Journal Online, 1-4.
Tsoukas, H. (2017). Don't simplify, complexify: from disjunctive to conjunctive theorizing in
organization and management studies. Journal of management studies, 52(2), 132-153.
Tsoukas, H., & Chia, R. (2002). On organizational becoming: Rethinking organizational
change. Organization Science, 13(5), 567-582.
Tsoukas, H., & Hatch, M. J. (2001). Complex thinking, complex practice: The case for a
narrative approach to organizational complexity. Human Relations, 54(8), 979-1013.
Turner, J. R. (2014). The Handbook of Project-Based Management: Leading Strategic Change
in Organizations.4th Edition. New York, NY: McGraw-Hill.
Vaagaasar, A. L., & Andersen, E. S. (2007). On task evolvement in renewal projects.
International Journal of Project Management, 25(4), 346–355.
Van de Ven, A. (2013). Process theories of change. In E. H. Kessler (Ed.), Encyclopedia of
management theory (Vol. 2, pp. 612-616). Thousand Oaks, California: SAGE
Publications, Inc.
Van de Ven, A. H., & Sun, K. (2011). Breakdowns in implementing models of organization
change. The Academy of Management Perspectives, 25(3), 58-74.
Von Foerster, H. (1984). Principles of self-organization – in a socio-managerial context. In
Ulrich, H. and Probst, G. J. B. (Eds), Self-Organization and Management of Social
Systems. Berlin: Springer-Verlag, 2–24.
Williams, M. (2000). Interpretivism and Generalisation. Sociology, 34(2), 209-224.

31
Figure 1: Overview of the research foundations
Research philosophy Structural realism
(Kilduff et al., 2011)

Ontological assumption Becoming Being

(Tsoukas & Chia, 2002)
(Langley et al., 2013)

Epistemological perspective Process based Recursivity based Equilibrium based

(Hernes & Bakken, 2003)

Theories of change Process theories of change Eigen behavior

(Van de Ven, 2013; Hernes, 2014) (Tsoukas, 2017)

Level of analysis PMO PfM / Org context

Unit of analysis Structure Capabilities Routines Actions

(Salvato & Rerup, 2011)

Dynamic elements Routine elements (performative, ostensive and artefact)

(Pentland & Feldman, 2005)

Micro level components Individuals Processes Structures

(Felin et al., 2012)

Research strategy Structural analysis Multiple Case Studies

(Godet, 2010)
Figure 2: Conceptual framework: dynamic system of co-evolution between PMO, PfM and Organizational Context

PfM
Fo-Fp
Rules
Fp-O applied
Context of the Fp-Fo
Performative Ostensive
joint activity
O-Fp
(Organization)
Fa-Fp Fa-Fo
Artefacts

Pp-Fp
Pa-Fa Fp-Po
Fp-Pp Fo-Po
Fo-Pp
O-Pa
Artefacts
Pa-Pp Pa-Po

Pp-O
Performative Ostensive
Pp-Po
O-Pp Rules
applied Po-Pp

PMO

The path of PMO and PfM co-evolution

The path of variation @ PfM level
The path of variation @ PMO level
Figure 3: Sample system with three elements

x-y

X Y
x-z

z-x y-z
Z
Figure 4: Sample system matrix A

Direction of relation 
influence > dependence Influence 
  X Y Z score

X 0 1 1 2 ‘Influence score’ of X = (X > Y) (= 1) + (X > Z( (= 1) = 2
Y 0 0 1 1
Z 1 0 0 1

Dependence score 1 1 2

‘Dependence score’ of Z = (X > Z) (= 1) + (Y > Z) (= 1) = 2
 Figure 5: Example of Map of Influence/Dependence

Influence 100%

90% 4 3

80% Influential Mediating

Elements Elements
70%

60%
Starting point x
Average Influence 50% Y
Stability / after 3 iterations Z

40%

30%
1 2
20%

Non-influential Resultant
10% Elements Elements

0%
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Average Dependence
Dependence
 Figure 6: Unweighted and weighted maps of direct influence for PMO1

Influence 100%
4 - influential 3 - mediating
90%

80%

70% O

Fp
60%
Fo

50% Fa
Average Influence
Pp
40%
Po

30% Pa

UW-DI
20%
1 – non-influential 2 - resultant W-DI
10%

0%
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Average Dependence
Dependence
 Figure 7: Unweighted and weighted maps of indirect influence for PMO1

Influence 100%
4 - influential 3 - mediating
90%

80%

70% O

Fp
60%
Fo

50% Fa
Average Influence
Pp
40%
Po

30% Pa

UW-IDI
20%
1 - non-influential 2 - resultant W-IDI
10%

0%
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Average Dependence
Dependence
 Figure 8: State shifts of the routines elements during the evolution the system

Pp
Po 2 - resultant
Fp

Fp
3 - mediating
Pp
O O
4 - influential
Pa
Po Pa
Fa Fa 1 – non influential
Fo Fo

Initial state End-state Role/state of

routine elements
 Figure 9: Key states of routines elements at initial and end-state of the system

Initial state End-state

O Pp
+ +
Pa Fp O Po
Pp
+
Fp

influential mediating resultant influential mediating resultant

 Figure 10: PMO/PfM/Organizational context system Eigen behavior and routines elements trajectories (PMO6 as example)

Influence 100%

4 - influential dynamic 3 - mediating

90%
transformation
during the first
80% 3 iterations

70%
Oscillation around a O
60% steady state after 3 Fp
iterations Fo
Average Influence 50% Fa
Pp
40% Po
Pa
30%

20% Oscillation

10% 1 – non-influential 2 - resultant

0%
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Average Dependence
Dependence
 Table 1: Relations between PMO, PfM and Organizational Context

-------------------
Relations From (element) To (element)
O-Fp Organizational-Context PfM-Performative
O-Pp Organizational-Context PMO-Performative
O-Pa Organizational-Context PMO-Artefacts
Fp-O PfM-Performative Organizational-Context
Fp-Pp PfM-Performative PMO-Performative
Fp-Po PfM-Performative PMO-Ostensive
Fo-Pp PfM-Ostensive PMO-Performative
Fo-Po PfM-Ostensive PMO-Ostensive
Fp-Fo PfM-Performative PfM-Ostensive
Fo-Fp PfM-Ostensive PfM-Performative
Fa-Fp PfM-Artefacts PfM-Performative
Fa-Fo PfM-Artefacts PfM-Ostensive
Pp-O PMO-Performative Organizational-Context
Pa-Fa PMO-Artefact PfM-Artefact
Pp-Po PMO-Performative PMO-Ostensive
Po-Pp PMO-Ostensive PMO-Performative
Pa-Pp PMO-Artefact PMO-Performative
Pa-Po PMO-Artefact PMO-Ostensive
Pp-Fp PMO-Performative PfM-Performative
 Table 2: Case studies description

Organizations C1 C2
PMOs PMO1 PMO2 PMO3 PMO4 PMO5 PMO6
Number of 25 6 11 14 14 11
projects /
portfolio
Number of 5 3 3 3 3 3
employees /
PMO
Number of 250+ / 11 43 / 4 72 / 6 45 / 8 37 / 5 30 / 4
employees /
Project
Management
& PfM
practices
Number of 3 3 3 3 3 3
meetings, meetings / meetings meetings meetings meetings meetings
interviews, 9 2 2 2 2 2
and focus interviews interviews interviews interviews interviews interviews
group / 1 focus / 1 focus / 1 focus / 1 focus / 1 focus / 1 focus
(structural group group group group group group
analysis)
Number of 2 2 3 2 2 2
organizational
routines
 Table 3: Routine elements and data collection methods

Definition Challenges How to collect data

Performative The actual Performances are Semi-structured interviews with
performances of distributed over the routine actors
routines time and space
Ostensive The schematic form Distributed and Semi-structured interviews with
of routines subjective the organizational managers
related to the routine
Artefact The physical Artefact can be Documents and archival data
manifestation of mistaken for the analysis (rules, procedures and
routines whole routine or organizational historical data)
the ostensive
element
Performative Creation of routines Different Selecting different enactments
and is based on the participants realize (performative) results in
ostensive reciprocal relation different starting unintentional variations of
interactions of performative and and ending points ostensive elements; changes in
ostensive for one routine norms (ostensive) causes
intentional variations of
performative elements
Artefact and The control of Most artefacts do Documents and archival data
performative behavior not specify (using the change management
interactions particular actions database) and semi-structured
interviews (the regarding
personnel)
Artefact and Alignment of The physical Following critical incident theory
ostensive formal documents artefact does not to understand what made a
interactions with the determine the flow manager change a procedure or
understanding of a job rule
about what routine
actors are doing
 Table 4: The unweighted and weighted matrices of direct influence for PMO1

UW-MDI for PMO1 W-MDI for PMO1

O Fp Fo Fa Pp Po Pa O Fp Fo Fa Pp Po Pa

O 0 1 0 0 1 0 1 O 0 3 0 0 3 0 3

Fp 1 0 1 0 1 1 0 Fp 1 0 1 0 1 0 0

Fo 0 1 0 0 1 1 0 Fo 0 1 0 0 1 1 0

Fa 0 1 1 0 0 0 0 Fa 0 2 3 0 0 0 0

Pp 1 1 0 0 0 1 0 Pp 2 1 0 0 0 1 0

Po 0 0 0 0 1 0 0 Po 0 0 0 0 1 0 0

Pa 0 0 0 1 1 1 0 Pa 0 0 0 3 2 3 0
 Table 5: Summary of the analyses for the six systems

Cases PMO1 PMO2 PMO3 PMO4 PMO5 PMO6

Nb of 5 5 4 4 6 6
iteration /
stability1
Evolution of the state of each element from (quadrant) / to (quadrant)
Quadrant: 1: non-influential; 2: resultant; 3: mediating; 4: influential.
O 4/4 4/4 4/3 3/3 3/3 4/3
Fp 2/2 2/2 2/3 3/3 4/4 3/3
Fo 1/1 1/1 1/1 1/1 1/1 4/4
Fa 1/1 1/1 1/1 1/1 1/1 1/1
Pp 2/3 2/2 2/3 3/3 2/3 3/3
Po 1/1 1/1 1/1 2/2 2/1 2/2
Pa 4/4 4/1 4/1 4/1 1/1 4/1
1Number of iterations for reaching stability: when no difference ≥ 2% in elements % of influence and
dependence between two consecutive iterations.
 Table 6: Summary of the analyses for the six systems

Characteristic Polynomial Eigenvalue Eigenvalue

(real numbers) (complex numbers)

PMO1 -λ^7+12*λ^5+22*λ^4+39*λ^3+120*λ^2+135*λ+36 4.52 0.61+/-1.81i

-0.38 -2.06+/-0.61i
-1.24

PMO2 -λ^7+7*λ^5+16*λ^4+42*λ^3+72*λ^2+74*λ+36 3.92 0.31+/-1.76i

-1.10 -0.72+/-0.89i
-2.00

PMO3 -λ^7+9*λ^5+16*λ^4+32*λ^3+66*λ^2+64*λ+24 4.03 0.44+/-1.62i

-1.10 -0.77+/-0.43i
-1.36

PMO4 -λ^7+26*λ^5+63*λ^4+56*λ^3+96*λ^2+120*λ+45 6.18 0.43+/-1.21i

-0.68
-0.88
-2.52
-2.97

PMO5 -λ^7+17*λ^5+29*λ^4+28*λ^3+70*λ^2+58*λ+30 4.96 0.47+/-1.29i

-2.16 -0.48+/-0.55i
-2.78

PMO6 -λ^7+15*λ^5+31*λ^4+50*λ^3+183*λ^2+308*λ+180 5.01 0.83+/-1.91i

-1.15+/-0.51i
-2.19+/-0.68i

View publication stats