1
MNE and indigenous community conflict: prior community
involvement and community autonomy as predictors of the
nature of conflict.
Malin Nystrom
Student ID: 11087609
24 June 2016
Master Thesis: Draft
MSc Business Administration – International Management
Supervisor: Dr. Ilir Haxhi
Second reader: Dr. Niccolo Pisani
University of Amsterdam
2
Statement of Originality
This document is written by Malin Nystrom who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is
original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for
3
Table of contents
1. Abstract………...………….4
2. Introduction………...5
3. Literature Review………....….9
3.2 MNE-community conflict – causes and consequences………..….…..9
3.3 Conflict resolution, mitigation and prevention……….……..12
3.4 Community autonomy………..………….……….…….………14
4. Theoretical Framework……….……….16
4.2 Research gap and questions………...………..16
4.3 Hypotheses………...….………...17
4.3.1 Prior community involvement………18
4.3.2 Community autonomy………..……..19
4.3.3 Moderation………..20
4.4 Conceptual model……….…………22
5. Data and method………....………....23
5.2 Data collection and sample……….……….23
5.3 Variables………..……….24 5.3.1 Dependent variable……….…...……….24 5.3.2 Independent variable..……….25 5.3.3 Moderator……….…….………..25 5.3.4 Control variables……….……….…….………..26 5.4 Model summary………28 6. Results………...……….29 6.2 Descriptive statistics………29
6.3 Multicollinearity and correlation analysis………..30
6.4 Regression analyses……….…33 6.4.1 Level of violence………33 6.4.2 Length of conflict………...36 7. Discussion………..40 7.2 Findings………40 7.3 Contribution……….…44
7.4 Limitations and suggestions for future research……….…45
8. Conclusion………..48
4
1. Abstract
Community involvement is considered one of the main ways of reducing MNE-community
conflict by addressing some of its underlying causes. Paradoxically, trends show increasing
levels of MNE social performance while also pointing at an increase in the frequency and
intensity of conflicts. This is because increasing recognition of indigenous peoples’ rights
empowers communities to oppose projects which they feel infringe on their way of life, but
also because firms’ efforts to manage community relations often fall short, failing to
compensate for this effect. This study cross sectional study analyses 675 cases of
MNE-community conflict, examining the predictive power of the degree of prior MNE-community
involvement and the degree of community autonomy on conflict between extractive industry
firms and indigenous communities, with the purpose of gaining greater insight into the
dynamics that shape the nature of such conflict.
While the results support that prior community involvement reduces the level of violence,
they also show that higher degrees of community autonomy lead to higher levels of
violence, as well as having a positive moderating effect on the relationship between
community involvement and violence – cancelling out the effect as autonomy increases.
This contributes to the current literature by challenging the key argument that prior
community involvement is the best means of preventing MNE-community conflict, adding
to our understanding, not only whether, but also when and why this is the case. By doing so
the study also creates a stronger link between two strands of the literature, and creates
5
2. Introduction
The story of indigenous peoples and multinational enterprises (MNEs) is a complicated one
that requires a nuanced approach and adaptable thinking. The topic has been receiving
increasing attention in global political environments and the media, as well as in
management literature and corporate agendas, and the conflicts that have arisen come under
ever closer scrutiny.
With resources becoming increasingly sought after at the same time as supply is
diminishing, companies in extractive industries are drawn further and further into
unexploited areas. Such areas have often been left untouched, and so has the cultures and
the way of life of the people inhabiting them. (Calvano, 2008) When MNEs implement
projects in such areas, indigenous peoples and the environment far too often suffer severe
consequences. (Scassola, 2013; Hilson, 2001) While there is increasing international
pressure being put on governments to strengthen the rights of their indigenous peoples, there
is still a huge gap between the adoption of standards and their implementation, and
extraction activities are often approved and justified by goals of economic development.
(Acuna, 2015; Adamson and Pelosi 2014; UNDRIP, 2013)
From a business perspective, in particular for MNEs in the extractive resources industry,
this creates significant threats in terms of costs and reputation. (Laplante and Spears, 2008)
A large number of studies have examined the best solutions for MNEs to handle conflict,
whether that is through solving, mitigating or preventing them. (Castro and Nielsen, 2001;
Davis and Franks, 2011; Frynas 2010; Franks, 2009; Foster, 2012; Lertzman and
Vredenburg. 2005; Shankleman, 2010; Kemp, 2010; Kemp et al, 2001; Nish and Bice, 2011)
While there is not one universal strategy which can be applied in all cases, there is a
6 understanding and dialogue, is key to the firm’s success. With one of the main causes for
community opposition being the feeling of not being in control of one’s own destiny,
involving indigenous communities in the process, prior to the implementation of a new
project is believed to reduce the risk, and the intensity of MNE-community conflict.
(Laplante and Spears, 2008; Nish and Bice, 2011; Harvey and Bice, 2014; Murphy and
Arenas, Kemp et al. 2001) This is consistent with UNDRIP’s (2013) call for free, prior and
informed consent to all extractive activities on their territorial land. However, with most
previous articles using qualitative methods, there is a gap in the literature, testing the
statistical significance of this recommended solution. Therefore, it is appropriate to ask:
- RQ1; What is the relationship between prior community involvement and the nature
of the conflict in cases of conflict between extractive industry MNEs and indigenous communities?
At the same time, another strand of the literature, headed by Hodge (2014), draws attention
to an interesting paradox. While extractive industry firms continue to improve their social
and environmental performance, MNE-community conflict is also increasing. Supported by
sceptics to the reality of MNEs implementing community involvement strategies, he argues
that this phenomenon is due to the rise in recognition of indigenous community rights,
empowering them to oppose MNE agendas. (Hodge, 2014; Brereton, 2014; Castro and
Nielsen, 2001; Hamann, 2014; O’Faircheallaigh, 2013) This suggests that companies’
attempts to act more responsibly might not have the desired effect, which adds another layer
to the discussion on how to avoid MNE-community conflict. Currently, there is a gap in the
literature, failing to consider how the rise in community autonomy might affect the solutions
7 - RQ2: Does community autonomy have a moderating effect on the relationship
between community involvement prior to extractive industry projects and the nature of the conflict?
The purpose of the study is to gain a greater insight into the dynamics that shape the nature
of MNE-community conflict by examining prior community involvement and community
autonomy as predictors of the level of violence and the length of conflict. This will be done
through a cross-sectional study, examining 675 cases of MNE-community conflict through
a stepwise, multiple regression analysis.
From a theoretical standpoint, the study tests a key argument in the previous literature –
community involvement as a means of reducing conflict - while also bridging the gap
between two different strands of research on MNE-community conflicts. This way the
results add to the understanding of not only whether, but when and why community
involvement might be an effective choice. This also has practical implications by drawing
attention to potential issues relating to community involvement, creating awareness which
can lead to better community management practices in the future.
The study will start by reviewing the previous literature on the topic. The literature review
in turn sets the context for the theoretical framework which provides motivation for the
research questions which the study seeks to address, as well as the hypotheses which are
formed to explore them. The next chapter will discuss the methodology, with a description
of the data collection process, sample, and the different variables. This is followed by a
section presenting the results from the analyses used to determine the statistical relations hip
8 their ability to answer the set out research questions and how this ties in with the previous
literature. It will also discuss limitations and suggestions for future research, before
9
3. Literature review
3.2 MNE-community conflict: causes and consequences
With a rising demand for natural resources, multinational enterprises (MNEs) move in
closer on areas which have previously been left untouched - areas often inhabited by
indigenous peoples. In such situations, the interests of MNEs and indigenous communities
are likely to collide which can lead to conflict, unless managed properly. (Calvano, 2008;
Bebbington and Bury, 2009)
Most documented cases of MNE-community conflict, sometimes referred to as
‘socio-environmental conflicts’ are with companies who fall within the category of 'extractive industries'. (Acuña, 2015:85) These include, but do not exclusively refer to mining, oil and
gas, and forestry. These industries are often accused of having significant negative
environmental impacts, ranging from deforestation or erosion before and during the
projects, to the risk of environmental disasters such as oil leaks. But these types of projects
can also have negative socio-economic impacts in the areas where they are implemented,
including land dispossession or dispossession of identities. (Calvano, 2008; Scassola, 2013;
Hilson, 2001; Acuña, 2015)
But extractive industries do not only present a significant threat for the preservation and
rights of indigenous communities and their territorial lands. From a business perspective,
operating in areas of indigenous peoples also present a significant risk to MNEs. The nature
of extractive activities requires large capital and long-term investments which might not
have an immediate pay off, making the industry particularly vulnerable to delay caused by
conflict. (Hodge, 2014) MNEs who end up facing community opposition may run the risk
of suffering both financial and reputational costs. (Laplante and Spears, 2008; Hodge, 2014)
10 between the MNE and the affected indigenous communities before, during and after the
implementation of projects.
Not following good community practices can have devastating consequences for the firm.
(Franks, 2009; Davis and Franks, 2011) Yet, as pointed out by authors such as Kemp and
Owen (2013) and Laplante and Spears (2008), so far, company efforts have been
underwhelming in this respect, with most CSR strategies failing to capture the complex
nature of these type of conflicts and address the underlying causes to opp osition.
To understand the relationship between MNEs and indigenous communities, Santoyo
(2006) draws attention to the importance of the complex history which indigenous
communities around the world have had to withstand. Since the colonial era, indigenous
communities have been forced to live as a part of a national system which they did not
choose for themselves. This has a led to relationships, with both government and companies,
characterised by conflict and opposition, where the community is usually at a disadvantage
due to the uneven power distribution which exists between communities, governments and
MNEs. (Calvano, 2008) Furthermore, historically speaking, it is the indigenous communities
situated in the most resource rich lands that are affected the most profoundly by such conflicts,
this phenomenon is known as the resource curse and means that countries with a higher level
of resources, tend to have a lower level of economic development, and it is the indigenous
populations that feel the brunt. (Parlee, 2015)
Efforts are being made on an international level to improve the situation for, and strengthen
the rights of indigenous peoples. In 2007, the general assembly adopted the United Nations
Declaration of the Rights of Indigenous Peoples (UNDRIP). UNDRIP calls, among other
things, for indigenous peoples free, prior and informed consent to all extractive activities
taking place on their territorial land. As of now, over 140 countries have signed the
11 Zealand, and Australia, all countries with large indigenous populations, were among the last
to join. When doing so, they argued for the non-binding nature of the agreement, giving
them significant interpretational freedom and weakening its impact. (Adamson and Pelosi,
2014; Ingimundarson and Magnusdottir, 2013; UNDRIP, 2013) Further, when it comes to
projects involving resource extraction, despite contradictions with UNDRIP, governments
often choose to justify them through an argument of economic development, something in
which indigenous communities typically have not had much say. (Banerjee, 2001; Acuna, 2015) Hodge (2014:28) defines development as the aim “to expand or realise the potentials of; bring gradually to a fuller, greater, or better.” This definition is problematic since what
is “better” for a country, people, or area will depend on who’s perspective you choose to adopt, and the ambiguity of the pledge leaves it open to reneges.
Banerjee (2001:39) describes indigenous communities as often having "very different
social, cultural, political and economic agendas from those of industry". Calvano (2008)
follows a similar level of reasoning to Banerjee (2001), arguing that there is a perception
gap between the different actors' ideologies, which underpins their unique point of view.
These types of conflicts of interest between different stakeholder groups makes creating and
maintaining a good relationship between stakeholders increasingly difficult. (Banerjee,
2001)
Extractive activities can cause major concerns for indigenous peoples, however, the
situation does not always need to be a negative one. Both Lertzman and Vredenburg (2005),
and Foster (2012), recognise the fact that foreign direct investment in indigenous regions
can provide opportunities for its people as well as presenting threats. However, even when
the community members recognise and are positive about the potential benefits that a
project could bring, there is a lot of uncertainty involved. While these types of projects tend
12 position. Whether or not that happens, the nature of the projects mean that they do not last
forever and for a community which has lived on the land for generations, there is an
understandable concern also for what will happen once the firm leaves. (Acuña, 2015)
The main issue arises when the community feels that a project affects their way of life in a
negative way and when they lack the power or resources to oppose it. (Acuna, 2015) Laplante and Spears (2008:76) argue that “communities are primarily concerned with questions of control over their own destinies”. When they feel that an MNE threatens this
sense of control, the indigenous community is likely to oppose their project.
3.3 Solving, mitigating and preventing MNE-community conflict
With the serious risks involved for both indigenous communities and MNEs, it is
understandable that one of the main focuses in MNE-community conflict literature is on the
subject of ways in which MNEs can either solve, mitigate or prevent MNE -community
conflicts. (Castro and Nielsen, 2001; Davis and Franks, 2011; Frynas 2010; Franks, 2009;
Foster, 2012; Lertzman and Vredenburg. 2005; Shankleman, 2010; Kemp, 2010; Kemp et
al, 2011; Nish and Bice, 2011) In most countries, the existing legal framework does not
sufficiently support the contrasting needs of MNEs, governments and indigenous peoples.
Therefore, it is important that MNEs acknowledge their responsibility towards their
stakeholders. (Adamson and Pelosi, 2014; Foster, 2012)
Lertzman and Vredenburg (2005) and Foster (2012) believe in the potential for positive
outcomes, promoting an ethical approach to resource extraction, building on cooperation
and understanding between community, management and environmental science. However,
as discussed, for the general population, change has not yet come far enough and MNEs
13 MNE-community relations. (Lertzman and Vredenburg, 2005; Foster, 2012; Kemp and
Owen, 2013; Laplante and Spears, 2008) While, from a bigger perspective, it makes
considerable business sense to engage in good community management practices, the direct
benefit of effective community management might not be as easily identifiable. The firm’s
desire to get a new project up and running risks interfering with good community practices
since this might cause the firm to skip important steps that are necessary for understanding
the full context and building good relationships with the affected community. (Brereton,
2014) As such, “Inequities, the short-term perspective of management and decision-making,
the underlying commercial pressures and incentives that prioritize company over
community culture and quality of life. All serve as resistance to dialogue.” (Hodge,
2014:32)
There is no definitive, universal strategy which is suitable in all cases, however, while
suggesting slightly different approaches, there is a general consensus that the one of the key
factors for better MNE-community relations is knowledge and understanding. This relates
to the values of both parties, as well as to the underlying causes, and the consequences that
conflict can have - financial and reputational for the firm, and for the way of life of
indigenous peoples. This, in turn, can only reasonably be achieved when the MNE and the
community engage in meaningful dialogue in an attempt to decrease the perception gap.
(Calvano, 2008; Benito and Gripsrud, 1992; Hodge, 2014; Nish and Bice, 2011; Harvey and
Bice, 2014) In fact, the majority of studies where maintaining a good relationship between
the firm and the community is considered a key success factor, argues for discussions
between the MNE and the indigenous community as an absolute minimum before
implementing a new project or entering a new area. (Castro and Nielsen, 2001; Davis and
Franks, 2011; Nish and Bice, 2011; Frynas 2010; Franks, 2009; Murphy and Arenas, 2010;
14 being one of the main causes for MNE-community conflict, the more involved the affected
community can be in the process, the better. This assumption corresponds with UNDRIP’s
call for indigenous peoples free, prior and informed consent, known as FPIC. Adam and
Pelosi (2014) provides a broad definition of FPIC as “the principle that a community has the right to give or withhold its consent to proposed projects that may affect the lands they
customarily own, occupy or otherwise use.” (Adamson and Pelosi, 2014:11) Without
cooperation between the firm and the community, it is difficult to build trust. (Murphy and
Arenas, 2010; Harvey and Bice, 2014) An important takeaway from UNDRIP’s
recommendation being that efforts are likely to be effective when implemented prior to the
project, rather than responding to opposition as it begins to gather. At that stage the firm
already risks delays and unwanted media attention. Engaging prior to the project will also
allow communication channels to raise any potential concerns, this way the MNE can
receive early warning signals, acting on issues before the conflict escalates. (Kemp et al.
2011) Starting an open and two-way dialogue with indigenous populations from the earliest
stages of a project seems logical, but it is often overlooked and subsequently plays a big part
in many of the conflicts.
3.4 Community autonomy
From a stakeholder perspective, MNEs have a responsibility towards the local communities
in which they wish to hold, or already undertake extractive activities. (Freeman et al. 2010;
Banerjee, 2001) When there is a lack of government intervention, the responsibility to
enforce the type of good community management practices discussed above, lies solely o n
the company. (Hilson, 2001) However, some researchers suggest that the nature of the
potential conflict does not only depend on whether or not the MNE chooses to include the
15 of community involvement, Brereton (2014) and Hamann (2014) also critiques the practical
reality of bridging the existing gap between MNE and community interests, questioning whether companies’ commitment to community dialogue and involvement can be relied upon. “Given the financial and other constraints under which mining companies operate,
we need to be realistic about what can be achieved through the mechanisms of voluntary
action and self-regulation”. (Brereton, 2014:38) Similarly, Castro and Nielsen (2001)
recognise that co-management efforts are often more complicated than first anticipated.
In this context, Hodge (2014) draws attention to an interesting paradox. While there is
evidence that the social and environmental performance of MNEs in extractive industries is
increasing, so is the number and intensity of MNE-community conflicts in indigenous community regions. Brereton (2014:37) summarises the explanation for this as “while companies are changing, the world around them is changing even more quickly”.
O'Faircheallaigh (2013) argues that an increasing recognition of indigenous rights gives
increasing leverage to oppose MNE decisions to both indigenous and non-indigenous
people who are affected by extractive industries operating in their region. What this really
means is that, as indigenous communities become increasingly empowered by higher
recognition from governments, pushed for by global human rights organisations and media
attention, they are increasingly able to oppose planned projects in their territories.
As discussed above, communities are often at a disadvantage when extractive industry
projects are planned and implemented due to the uneven power distribution between
government, MNEs and indigenous communities (Calvano, 2008). However, the findings
by Hodge (2014), indicate that making the distribution of power more even between the
different stakeholders might not be enough to avoid MNE-community conflict as long as
16 Clearly, the dynamics of MNE-community conflict are more complicated than a simple
cause and effect. Calvano (2008:793) summarises this nicely by stating, that for this new
accountability of MNEs to effectively manage their relationships with indigenous
communities “managers must first understand the causes of conflict with local communities,
and communities must understand what courses of action are available to challenge activities they deem harmful to their interests”.
4. Theoretical Framework
4.2 Research gap and questions
Most of the current literature on MNE-community conflict focus on one of two sides of the
debate, either on the underlying reasons of conflict, or on different ways for MNEs to solve,
mitigate or avoid conflicts with indigenous communities. One of the key success factors
which is put forward in the literature, lies in involving the indigenous community in the
process prior to the implementation of a new project. The reason for this is that community
involvement addresses what is considered as one of the main underlying causes for
MNE-community conflict – namely, the feeling of not being in control of one’s own life and
destiny. This is in line with the United Nations Declaration of the Rights of Indigenous
Peoples (UNDRIP) and the call for indigenous peoples’ free, prior and informed consent to
all extractive activities on their territorial land. (UNDRIP, 2013) While there is no definitive
strategy that will fit every situation, there is a clear consensus that MNE -community
relations will benefit from mutual understanding and increased knowledge-sharing between
stakeholders, as this will create trust and decrease the existing persistence gap. (e.g. Harvey
and Bice, 2014; Kemp et al. 2011) However, most studies to date are of a qualitative nature,
17 testing the effect of this proposed solution on the nature of conflict situations . Therefore,
an appropriate research question to ask is:
- RQ1; What is the relationship between prior community involvement and the nature
of the conflict in cases of conflict between extractive industry MNEs and indigenous communities?
A different strand of literature draws attention to an increase in the recognition of
indigenous community rights which empowers communities to oppose MNE projects that
encroach on their territorial lands. (O’Faircheallaigh, 2013) This is believed to be one of
the reasons for an apparent paradox where, despite rising social and environmental
performance in extractive industries, MNE-community conflict is continuously increasing
in both frequency and intensity. (Hodge, 2014) This adds another layer to the discussion on
how to avoid MNE-community conflict as it puts into question whether companies’ attempts
to act more responsibly is actually having the desired effect. Currently, the literature does
not address how the rise in community autonomy might affect the solutions that are being
put forward. Therefore, the second research question asks:
- RQ2: Does community autonomy have a moderating effect on the relationship
between community involvement prior to extractive industry projects and the level of nature of the conflict?
4.3 Hypotheses
To answer these research questions, six hypotheses are developed. These hypotheses make
18
4.3.1 Prior community involvement
The previous literature puts forward community involvement as the best way to prevent or
mitigate MNE-community conflicts. (Acuna, 2015, Laplante and Spears, 2008; Nish and
Bice, 2011; Harvey and Bice, 2014; Murphy and Arenas, Kemp et al. 2011; Calvano, 2008;
Benito and Gripsrud, 1992; Hodge, 2014) Involving the community prior to the project’s
implementation does not only correspond with UNDRIP’s call for free, prior and informed
consent, but also allows for communication between the parties, increasing both knowledge
and understanding, thereby reducing the perception gap and allowing issues to be brought
up before conflicts arise. (UNDRIP, 2013; Kemp et al. 2011; Banerjee, 2001) According to
Harvey and Bice (2011), one of the main motivations for involving communities in the
process, is that it increases trust. Further, it reduces the feeling of not being in control of
not being in control of one’s own destiny, which is considered one of the main causes for
conflict. (Acuna, 2015; Laplante and Spears, 2008)
In this way it is argued that community involvement acts as a way of increasing the sense
of control that community members feel over their own destiny, in turn, reducing the need
to oppose MNE actions.
For example, Kemp et al. (2011:94) argues that cooperating with the community and giving them channels to voice their concerns can “serve as ‘early warning systems’ for both companies and communities; prevent the risk of conflict escalation; and help identify
systematic issues rather than dealing with community grievances on an ad hoc basis”. Not
only does this display how involving the community can increase understanding, but they
also go on to argue how this provides a way for the community to gain a sense of control,
either through co-development of knowledge, or merely by building trust, as “The
introduction of such regulating forces may also be seen in the context of lo cal communities
19 economies, rather than have these determined by powerful corporate entities”. (Kemp et al. 2011:94)
The first two hypotheses (H1 and H2) address the relationship between prior community
involvement and the nature of the conflict in cases of conflict between extractive industry
MNEs and indigenous communities, hypothesising that:
- H1: Higher degrees of community involvement prior to the project leads to lower
levels of violence.
- H2: Higher degrees of community involvement prior to the project leads to shorter
conflicts.
4.3.2 Community autonomy
Hypothesis 1 (H1) tests the assumption, based on previous literature, that the best way to
avoid conflict is through involving the community in the process. (e.g. Murphy and Arenas,
2010; Kemp et al. 2011; Calvano, 2008) That is to say that the degree of community
involvement changes the nature of the conflict. Another branch in the literature on
MNE-community conflict discusses the effect that MNE-community empowerment has on the frequency
and intensity of conflict. (Hodge, 2014; O’Faircheallaigh, 2013; Nelles and Alacantara,
2014) Hodge (2014) draws attention to an apparent paradox – while the social and
environmental performance of extractive industry firms is showing an upwards trend, the
number and intensity of MNE-community conflict is increasing. This paradox is explained
by an increase in the recognition of indigenous peoples’ rights. While the power distance that exists between MNEs, governments and indigenous communities have often meant
communities have very little say in extractive activities on their territorial lands, w ith
20 when their interests collide. (Hodge, 2014; O’Faircheallaigh, 2013; Banerjee, 2001; Acuna,
2015)
On the basis of this, it is possible to argue that increasing community autonomy could in
fact lead to an escalation of conflicts, as communities can amount stronger opposition, either
through increased violence or by resisting MNE powers for longer periods, thereby dragging
out the conflict.
While involvement could be argued to lead to increased power by gaining sense of control
of the process, O’Faircheallaigh (2013) explains that autonomy does not guarantee the
community to be given more control by the firm, however, it does increase their power in a
different sense, giving them leverage to stand up against MNEs who they feel infringe on
their rights and threaten their way of life.
Therefore, it is hypothesised that:
- H3: Higher degrees of community autonomy leads to higher levels of violence.
- H4: Higher degrees of community autonomy leads to longer conflicts.
4.3.3 Moderation
While there is a consensus in the literature that involvement is necessary to create good
MNE-community relations, some raise concerns that it might be more complex than many
authors make out. Brereton (2014) and Hamann (2014) are two of the authors to question
whether firms invest enough in achieving effective involvement. Murphy and Arenas (2010)
argue that co-management will ultimately lead to less conflict, but according to Castro and
21 and companies risk falling short. (Kemp et al. 2001; Kemp and Owen, 2013; Laplante and
Spears, 2008)
Assuming that hypothesis 1 (H1) and hypothesis 2 (H2) are supported, we can argue that
while higher levels of involvement change the nature of the conflict by lowering the level
of violence and the length of the conflict. With the challenges surrounding effective
community involvement, this effect risks being weakened when levels of autonomy are
high, since high levels of autonomy would lead to higher levels of violence, creating two
competing effects.
The example of this can be seen in Hodge’s (2014) paradox. Despite the fact the companies’
social and environmental performance is increasing, the frequency and intensity of
MNE-community conflict continues to increase. This paradox is explained by both Hodge (2014)
and O’Faircheallaigh (2013) by communities gaining increasing recognition. Therefore, it is hypothesised that:
- H5: The degree of community autonomy has a positive moderating effect on the
relationship between the degree of community involvement prior to the project and the level of violence.
- H6: The degree of community autonomy has a positive moderating effect on the
relationship between the degree of community involvement prior to the project and the length of the conflict.
22
4.4 Conceptual model
The above stated hypotheses result in the following conceptual model (Figure 1):
23
5. Data and method
This study takes a quantitative approach, using a cross-sectional research design to
determine the relationship between the degree of community involvement prior to an MNE’s project affecting an indigenous community, and the nature of the conflict, measured by its level of violence and the length of the conflict, as well as the moderating effect of
community autonomy on that relationship. This is done by examining the results of both
correlation analysis and multiple regression analysis. However, before running the analyses
there will be a description of the data collection process and the sample, as well as the
different variables, their measurement and a model summary.
5.2 Data collection and sample
The data was collected through a number of sources. Mostly via databases on MNE
activities affecting indigenous communities, but also with support from other sources such
as MNE public data and non-profit reports. The total sample consists of 675 cases of
community conflict. The number of cases was only restricted based on what was reasonable
for the timeline of study, and covers cases from all types of extractive industries as well as
a cases involving indigenous communities and MNEs from all over the world. T his is
believed to create greater generalisability across the population of conflict between
indigenous communities and extractive industry companies. Cases which did not include
sufficient data on the variables used in the study were excluded from the initial sample
which was gathered, resulting in the final sample of 675 cases. For the purpose of this study,
indigenous peoples are defined broadly, covering communities which are officially
24 nations, aboriginals etc., as suggested by The UN Declaration of the Rights of Indigenous
Peoples. (UNDRIP, 2013)
The next section takes a closer look at the data in the sample and gives a description of the
dependent, independent and moderator variables and their measures, including examination
through some descriptive statistics such as frequency, distribution, standard deviation and
mean.
5.3 Variables
5.3.1 Dependent variable
The dependent variable is the nature of the conflict. This will be represented by two
variables - the length of the conflict, and the level of violence.
The length of conflict indicates how long it took to resolve the conflict, which is considered
a sign of its severity as longer conflicts tend to be more devastating for both the community
and the MNE. (Kemp et al. 2011) The length of the conflict is measured from the first
recorded instant of disagreement, until the conflict has ended, i.e. when there is an absence
of instances of disagreement, usually because the conflict has been resolved in some way.
For conflicts which are still ongoing, the length is measured up until the time of this study,
namely June 2016. The length of the conflict is then coded in three categories, short
conflicts of less than three years, medium length conflicts between three and eight years,
and long conflicts which go on for longer than 8 years.
The level of violence in the conflict is used to reflect on the nature of the conflict as it
demonstrates the intensity of the conflict. (Calvano, 2008) Here, different types of violence
25 but do not exclusively refer to – no- to low level of violence includes peaceful protests and
negotiations; low level of violence includes court actions; medium level of violence
includes intimidation; high level of violence includes physical damage; and finally, extreme
level of violence includes kidnapping or deaths.
5.3.2 Independent variable
The independent variable used in the study is the degree of community involvement prior
to the project. As discussed in previous sections, community involvement is considered a
key success factor for good community relationships which prevent or mitigate MNE
-community conflict. (e.g. Castro and Nielsen, 2001; Nish and Bice, 2011; Frynas 2010;
Franks, 2009; Murphy and Arenas, 2010; Kemp et al. 2001) This variable’s measure
contains four different levels - no involvement, discussions with community, agreement
with the community, and involvement of the community in some degree of co-development
of the project – scoring from 0 to 3 depending on the degree of involvement. For the highest
degree of involvement- involvement of the community in co-development – the two lower
degrees of involvement, both discussions and agreement with the community is considered
a prerequisite.
5.3.3 Moderator
The moderator is the degree of community autonomy for the affected community. This
variable measures the relationship that the indigenous community has to the government,
arguing that increasing recognition of indigenous communities, associated with higher
levels of community autonomy makes communities more empowered, whereas lower levels
26 (O'Faircheallaigh 2013; Hodge, 2014; Adamson and Pelosi, 2014). The variable ranges
across five degrees of community autonomy, from no autonomy at all, to cultural
recognition, limited recognition of rules, partial recognition of rules or right of consultation,
through to almost full- to full recognition of community rules with autonomy or right of
consent.
5.3.4 Control variables
To further determine the significance of the results in the analysis, control variables were
used at the country, firm and community level. While it is not possible to account for all
potential influencers, the control variables were chosen as they represent factors that may
further impact governance mechanisms in conflict situations.
At a country level, institutional strength will be used as a control variable as countries with
weak institutions are expected to generally provide less support to their indi genous
communities. (Calvano, 2008) For example, better regulatory frameworks would serve as
better protection for the communities that reside within the country’s borders. (Bebbington, 2009) The institutional strength was determined by calculating the average of a country’s
Worldwide Governance Indicators, provided by the World Bank. These include Voice and
Accountability, Political Stability and Absence of Violence, Government Effectiveness,
Regulatory Quality, Rule of Law, and Control of Corruption. (The World Bank, 2016)
The control variable at the firm level is the MNE’s experience with indigenous communities, measured in number of years at the start of the conflict. This variable is used
since MNEs might learn from previous experiences, making them likely to handle conflict
situations better when encountering resistance again. Benito and Gripsrud (1992) argues that
27 considered an important aspect of good community relations management, MNE experience
is likely to have some influence.
Finally, at the community level of analysis, cultural isolation is used as a control variable.
Communities which are culturally isolated from broader ethnic groups are likely to
experience wider perception gaps, limiting understanding between the parties and thereby
making conflict situations more likely to arise. (Tang and Tang, 2001; Banerjee, 2011)
Cultural isolation is coded over three levels, where one means the community is highly
isolated, for example having their own language and not being connected to any broader
ethnic group, the second type is communities which are connected to a broader ethnic group
but are isolated within that group, and finally, the third encompasses communitie s which
can be considered to be embedded within a broader, mainstream ethnic group, thereby not
28
5.4 Model summary
Table 1shows a summary of the model specifications used for analysis. Model 1-4, focuses
on the first dependent variable – the level of violence, whereas model 5-8 examines the
second dependent variable – the length of conflict. The analysis is completed through a
stepwise approach, starting by entering the control variables, followed by the in dependent
variable – the degree of community involvement. Next, the direct effect of the second
predictor variable – community autonomy, is tested, before finally running a moderation
analysis, incorporating all variables with community autonomy as the mod erator. The
procedure is repeated for the two dependent variables.
.
Table 1. Summary of Linear Regression Analyses
CVs DVs IV M
Institutional Strength MNE Experience Cultural Isolation Level of Violence Length of Conflict Involvement Autonomy
Model 1 x x x x Model 2 x x x x x Model 3 x x x x x Model 4 x x x x x x Model 5 x x x x Model 6 x x x x x Model 7 x x x x x Model 8 x x x x x x
29
6. Results
To determine the results, we will investigate the different hypotheses in two main steps –
correlation analysis and multiple regression analysis. The correlation analysis will give us
initial indications on whether the different variables are related to each other and the
strength of those relationships. This is a pre-condition for continuing with the analysis. The
multiple regression will test the direction of the relationships indicated in the first step of
the analysis, as well as testing whether the degree of community autonomy has acts as a
moderator. This section will also examine some descriptive statistics and ensure that the
results are not negatively affected by multicollinearity.
6.2 Descriptive statistics
Examining some descriptive statistics will give us a better picture of the data in the sample.
First we will look at the dependent variables. The dependent variable - length of conflict -
has a mean of 2.24. With options ranging from 1 to 3 we can see somewhat of a tilt towards
the higher end, with 301 cases being longer than 8 years. The distribution between the
different levels of violence is relatively even, with no- to low levels of violence occurring
180 times in the sample of 675 cases. The remaining levels have a frequency between 114
and 131. This is also seen through a mean of 2.79.
For the moderator variable, the different degrees of community autonomy that are observed
in the sample each have a frequency between 89 and 171, with the middle level – limited
recognition of rules – being the most frequent observation. With a mean of 3.18, the
30 Next, the descriptive statistics show that out of the 675 cases of MNE-community conflict there was a total of 412 cases where there was no prior community involvement for the
project, with a steep, continuous drop between the different levels and a no more than 32
out of the 675 cases scoring at the highest level, indicating that the community was involved
in some form of co-development of the project. This is reflected by a mean of .57. This is
not too surprising. Research is showing that companies in the extractive industries are
increasing their social and environmental performance, putting more emphasis on
community relationship management, but they still have a long way to go. (Hodge, 2014)
However, it should be noted that the sample only includes cases where conflict has occurred,
therefore not accounting for the degree of prior involvement which was implemented in
cases where MNEs operated successfully in a region with indigenous people without
spurring MNE-community conflict.
All variables display a standard deviation below 1.5, except MNE experience which has a
standard deviation of 24.184. This is due to the nature of the variable which is measured in
number of years, rather than being split into categories.
Means and standard deviations can be seen in table 3.
6.3 Multicollinearity and correlation analysis
When two or more predictor variables in a model are correlated, there is a risk of
multicollinearity. This can significantly impact the results of the analysis by decreasing the
reliability of the results. Field (2009) suggests two approaches for detecting
multicollinearity. The first step is checking the correlation between the variables – if the
correlation between two variables is higher than 0.8, this could indicate multicollinearity.
31 using the Variance Inflation Factor (VIF). Field (2009) argues that for VIF values under 10,
with a tolerance level above 0.2, multicollinearity is not likely to occur. As presented in
table 2,the variables in the models are within these thresholds and therefore do not display
any signs of multicollinearity.
The next step looks at the relationship between the different variables is through a
correlation analysis. A correlation analysis will determine the strength of the relationship
between all the variables - the degree of autonomy, the total level of community
involvement, the level of violence, the length of the conflict, the institutional strength, the
degree of cultural community isolation, and MNE experience, using Pearson correlation
coefficients. Preliminary analyses were performed to ensure that no violation of the
assumptions of normality, linearity and homoscedasticity occurred.
First of all, the two dependent variables - the level of violence and the length of conflict -
are significantly correlated at r = .231, p < .01, showing that high levels of violence are
associated with longer conflicts. The results show a weak, negative correlation between the
degree of involvement and the level of violence, at r = -.098, p < .05, with high levels of
involvement being associated with low levels of violence. Next, the correlation analysis
reveals a small, positive correlation between community autonomy and the level of
violence, at r = .109, p < .01, with high levels of autonomy being associated with high levels
Table 2. Multicollinearity Statistics
Tolerance VIF Level of Involvement .949 1.054 Degree of Autonomy .875 1.143 Institutional Strength .829 1.207 MNE Experience .941 1.063 Cultural Isolation .932 1.073
32 of violence. There is no significant correlation between the length of conflict and either of
the predictor variables – the level of involvement and the degree of autonomy.
Finally, there is a small, positive correlation between community autonomy and the degree
of involvement, at r = .128, p < .01, which means that there is an association between high
levels of autonomy and high levels of involvement in the sample.
Looking at the control variables the highest observed correlation occurs between the length
of conflict and institutional strength, with a significant negative correlation of r = -.347, p
< .01, followed by the correlation between the level of violence and institutional strength,
with a significant negative correlation of r = -.284, p < .01. This means that when the
institutional strength is high, both the level of violence in the conflict and the length of the
conflict is low. Meanwhile, for countries with low institutional strength, both length and the
level of violence in the conflict is greater. Institutional strength also has a relatively high,
significant positive correlation with the degree of community autonomy, at r = .311 p < .01,
indicating that communities in countries with strong institutions have higher a utonomy. The
rest of the control variables – MNE experience and cultural isolation display weak
correlation within the -2 < p > 2 threshold. (Hayes, 2013)
While all of the relationships between the dependent variable, independent variable and
moderator can be considered as weak, or even non-significant in the case of the dependent
variable length of conflict, the size of sample means that they still reach statistical
significance, which allows us to continue with the analysis. (Pallant, 2010) The information
is summarised in a correlations matrix below (Table 3), including means and standard
33 Correlation of course, does not necessarily mean causation. While it gives us an indication
of the strength of the relationships, it does not inform us about their nature or direction. To
test this, we need to use regression analysis.
6.4 Regression analyses
As stated above, to determine whether the hypotheses are supported or not we need to use
a regression analysis. The description of the results will address one hypothesis at the time,
following the order described in the model specification from the method section (Table 1),
followed by a look at the effect of the different predictor variables in comparison with the
control variables.
6.4.1 Level of violence
Addressing our first hypothesis (H1), Model 2 tests the relationship between the degree of
community involvement and the level of violence, hypothesizing that higher degrees of
community involvement prior to the project leads to lower levels of viol ence. The results
show that on its own as a predictor, the degree of involvement does not have a significant Table 3. Means, Standard Deviations and Correlations
Variable Mean Std. Deviation 1 2 3 4 5 6 7
1. Level of Violence 2.79 1.448 -2. Length of Conflict 2.24 .776 .231** -3. Level of Involvement .57 .841 -.098* -.059 -4. Degree of Autonomy 3.18 1.312 .109** -.019 .128** -5. Institutional Strenght 3.576 1.388 .284** .347** .188** .311** -6. MNE experience 15.99 24.184 -.155** -.168** .117** .084* .191** -7. Cultural Isolation 1.99 .673 -.166** -.094* -.034 -.103** .157** .141**
-*Correlation is significant at the 0.05 level (2-tailed) ** Correlation is significant at the 0.01 level (2-tailed)
34 relationship with the level of violence, as b = -.080, p = .217. As such, hypothesis 1 (H1)
does not appear to be supported.
For Model 3, which tests the relationship between the degree of community autonomy and
the level of violence in the conflict for hypothesis 3 (H3) - higher degrees of community
autonomy leads to higher levels of violence - the findings show that autonomy is a
significant predictor of violence, as b = .229, p < .001. From this we can establish that for
every 1 unit increase in the degree of community autonomy, we get a .229 unit increase in
violence. Therefore, hypothesis 3 (H3) is supported. Model 3 also displays an adjusted R2
of .136, meaning that the model, including control variables, accounts for 13.2% of the
variance in the level of violence. As such, as a predictor of the level of violence, the degree
of community autonomy only contributes to 3.8% of the variance (R2 chan ge = 0.038).
Turning to the third, and final hypothesis for the degree of violence (H5), testing the
potential moderating effect of the degree of autonomy on the relationship between the
degree of involvement and the level of violence, the variables for au tonomy and
involvement were mean-centered to compute an interaction term before running Model 4.
The interaction shows a positive moderation effect, as b = .218. This effect is significant at
p < .001 and accounts for 6.4% (RQ2change= .064) of the variance in the level of violence.
The overall model explains 16.1% (adjusted R2 = .161). This supports hypothesis 5 (H5) -
the degree of community autonomy has a positive moderating effect on the relationship
between the degree of community involvement prior to the project and the level of violence.
Interestingly, when introducing the interaction effect, the relationship between the
independent variable, community involvement, and the dependent variable, the level of
35 To get a better picture of how the interaction works, we can examine the conditional effect
of the degree of involvement on the level of violence at different values of the degree of
autonomy (Table 5). This shows that for low levels of community autonomy b = -.361, p <
.001. This means that for low levels of community autonomy, every 1 unit of invol vement
gives us a .361 decrease in violence. For average levels of community autonomy, we get b
= -.076, p = .227. Thus, at average levels of community autonomy, the relationship between
the degree of prior community involvement and the level of violence is not significant –
which corresponds with the results achieved in Model 2. Finally, for high levels of
community autonomy, the relationship becomes significant again at b = .210, p < .05.
However, these results indicate that the nature of the relationship changes so that at high
levels of community autonomy, high degrees of involvement leads to higher levels of
violence. Therefore, H1 gets conditional support. This is explained by the fact that the
positive moderation makes so that the higher the degree of autonomy the more positive the
effect of the degree of involvement on the level of violence. In other words, because the
relationship between involvement and violence is negative, and the relationship between
autonomy and violence is positive, and the degree of autonomy acts as a moderator,
community autonomy cancels out the negative effect of community involvement.
When viewing the results, it should be kept in mind that the effects cannot be compared to
each other based on their b-value, as, for example, 1 unit of community involvement cannot
be considered equal to 1 unit of community autonomy since these represent different
measures. Therefore, table 4 also reports on the variables’ beta values which are
standardised coefficients. This shows that the effect of all control variables – institutional
strength, MNE experience, and cultural isolation – reach statistical significance in all
36 highest beta value in all models except Model 4, where both the effect of community
involvement (beta = -.446) and the interaction effect (beta = .455) is higher.
6.4.2 Length of conflict
Moving on to the second dependent variable – the length of conflict – the first hypothesis
(H2) states that higher degrees of community involvement prior to the project leads to
shorter conflicts. The results from Model 6 show that b = .012, p = .726, indicating that the
relationship between the degree of community involvement and the length of the conflict is
not significant, therefore, we reject hypothesis 2 (H2). This is further strengthened by the
fact that, while the overall model accounts for 12.7% of the variance in the length of conflict
(adjusted R2 = .127), the degree of community involvement contributes with 0% (R2 change
= .000).
Model 7 tests hypothesis 4 (H4) – that higher degrees of community autonomy leads to
longer conflicts. Here, b = .059, p < .01, which means that the relationship is significant
and for every 1 unit increase in the degree of community autonomy, we get a . 059 unit
increase in the length of conflict. The degree of community involvement accounts for 0.9%
of the variance in the length of the conflict (R2 change = .009), out of 13.5% which is
explained by the model in total (adjusted R2 = .135). Consequently, hypothesis 4 (H4) is
supported.
The final hypothesis (H6) states that the degree of community autonomy has a positive
moderating effect on the relationship between the degree of community involvement prior
to the project and the length of the conflict. This relationship is tested in Model 8. The
results show that neither the interaction (b = .018, p = .590), nor the relationship between
37 overall model accounts for 13.3% of the variance in the length of conflict (adjusted R2 =
.133), the interaction accounts for only 0.09% (R2 change = .009), which is not enough to
reach significance. Therefore, hypothesis 6 (H6) is rejected – community autonomy does
not have a moderating effect on the relationship between the degree of community
involvement and the length of the conflict.
Further, comparing the beta values for the models relating to the dependent variable length
of conflict, we can see that both the effect of institutional strength and MNE experience
reach statistical significance, with institutional strength being the stronge st predictor in all
models (Model 5-8), with beta values ranging from -.322 to -.358.
38 Tab le 4. Li ne ar R egr es sion A nal ys is Re su lts De pe nd en t Vi ol en ce Le ng th Con trol s H1 H3 H5 Con trol s H2 H4 H6 M ode l 1 M ode l 2 M ode l 3 M ode l 4 M ode l 5 M ode l 6 M ode l 7 M ode l 8 B be ta sig B be ta sig B be ta sig B be ta sig B be ta sig B be ta sig B be ta sig B be ta sig Con trol Ins tit ut ion al St re ng th -.260 -.249 .000 -.251 -.241 .000 -.331 -.317 .000 -.329 -.315 .000 -.180 -.322 .000 -.182 -.325 .000 -.198 -.355 .000 -.200 -.358 .000 M NE E xpe rie nc e -.005 -.091 .015 -.005 -.087 .021 -.006 -.006 .007 -.005 -.090 .013 -.003 -.102 .006 -.003 -.103 .005 -.003 -.106 .004 -.003 -.106 .004 Cul tur al Isol ati on -.245 -.114 .002 -.253 -.075 .002 -.173 -.173 .030 -.161 -.075 .042 -.033 -.028 .433 -.032 -.028 .455 -.014 -.013 .736 -.012 -.011 .777 In de pe nd en t Com mun ity Invol ve me nt -.080 -.046 .217 -.768 -.446 .000 .012 .013 .726 -.049 -.053 .590 M od er at or Com mun ity Aut on om y .229 .207 .000 .059 .099 .010 In te rac tion Invol ve me nt x A ut on om y .218 .445 .000 .018 .069 .509 Con sta nt 4. 297 4. 322 3. 689 4. 029 2. 998 2. 994 2. 841 2,867 R .322 .325 .376 .410 .363 .363 .375 .375 R2 .104 .106 .142 .168 .132 .132 .140 .141 Adj us ted R2 .100 .101 .136 .161 .128 .127 .135 .133 R2 ch an ge .104 .002 .038 .064 .132 .000 .009 .009
39
Table 5. Conditional Effect at Different Levels of the Variable – DV: Level of Violence
Community Autonomy B sig.
Low -.361 .000
Average -.076 .227
40
7. Discussion
This study examines how prior community involvement and community autonomy affect
the nature of conflicts between MNEs in the extractive industries and indigenous
communities in the areas they operate in. This section will provide a discussion of the
findings from the analyses above, in relation to the theories discussed in the literature
review, answering the research questions proposed described earlier, as well as the study’s
theoretical and practical contributions, limitations, and suggestions for further research.
7.2 Findings
Starting with the first research question (RQ1) - what is the relationship between prior
community involvement and the nature of the conflict in cases of conflict between
extractive industry MNEs and indigenous communities? The literature considers the
involvement of indigenous communities prior to the implementation of a project as a good
way of preventing or reducing MNE-community conflict because it addresses one of the
main underlying reason for such conflicts and a driver for resorting to violence, namely, the feeling of lack of control over one’s own life and destiny. (Acuna, 2015; Laplante and Spears, 2008) Involvement is also argued to allow for increased knowledge and
understanding between the different parties, making the perception gap that exists
between them smaller, facilitating better relationships. (Nish and Bice, 2011; Banerjee,
2001) Further, dialogue can provide early warning signs of conflict, allowing preventive
measures, rather than reacting to issues as they arise, whereby significant damage might
already have been caused. (Kemp et al. 2011) However, when testing the relationship
between the degree of prior community involvement and the level of violence and the
41 prior to the project leads to lower levels of violence in the conflict, and that (H2) – higher
degrees of community involvement prior to the project leads to shorter conflicts – neither
is supported as the effects fail to reach statistical significance. Clearly, here the data fails
to provide statistical evidence that, as suggested in previous studies, prior community
involvement works as a method for limiting MNE-community conflict.
Looking at the descriptive statistics, the most notable result is the uneven distribution for
the degree of community involvement. Out of the 675 cases of MNE -community conflict
there was a total of 412 cases where there was no prior community involvement for the
project, with a steep drop between the different levels and only 32 out of the 675 cases
scored at the highest level, indicating that the community was involved in some form of
co-management. This could be a sign that, as suggested by the likes of Kemp and Owen
(2013), and Laplante and Spears (2008), MNEs still have a long way to go in terms of
community management practices. Although, here it is also worth mentioning that the
sample only includes cases where some sort of conflict has occurred. This means that the
results do not account for the potential cases where MNEs have operated projects
successfully without attracting conflict with the community, which could negatively affect
the results.
However, to fully answer the first research question (RQ1), we must also explore the
second research question (RQ2) - Does community autonomy have a moderating effect on
the relationship between community involvement prior to extractive industry projects and
the level of violence in the conflict?
First, hypothesis 3 and 4 looks at the direct effect of community autonomy on the intensity
of the conflict. H3 argues that higher degrees of community autonomy lead to higher levels
of violence, while H4 argues that higher degrees of community autonomy lead to longer
42 O'Faircheallaigh (2013), states that increasing recognition of indigenous communities
empowers them to stand up again MNEs imposing on their territorial lands, thereby
intensifying conflicts. While the results show support for both hypotheses, community
autonomy is shown to be stronger predictor of the level of violence than the length of the
conflict.
Hypothesis 5 and 6 address the potentially moderating effect of community autonomy on
the relationship between the degree of prior community involvement and the nature of the
conflict. H6, hypothesising that community autonomy has a positive moderating effect on
the relationship between community involvement and the length of conflict, is rejected. This
confirms the previous result that there is no significant relationship between community
involvement and the length of conflict. H5 however, hypothesising that community
autonomy has a positive moderating effect on the relationship between community
involvement and the level of violence, is supported – both that community autonomy acts
as a moderator, and that the moderation effect is positive. This answers the second research
question (RQ2).
To fully understand this relationship, we looked at the conditional effect of the degree of
involvement on the level of violence at different values of the degree of community
autonomy. First of all, we can see that only at average levels of community autonomy, the
relationship between community involvement and the level of violence is non-significant.
In fact, at low levels of community autonomy, H1 is supported so that higher degrees of
prior community involvement lead to lower levels of violence. In other words, for
communities with no or limited autonomy, there is evidence that involving the community
prior to the project will make the conflict less intense, thereby confirming the method
proposed in previous literature. As the degree of autonomy rises, this relationship is cancel