Power Differences and Dynamics in Multiparty Collaborative Systems: A Systematic Literature Review

Abstract

This paper presents the results of a systematic literature review on power distribution and power dynamics in multiparty systems. Multiparty systems are underorganized social structures in which power dynamics unfold and impact collaboration effectiveness. We use a theory-driven approach to integrate the empirical literature that explored power differences and dynamics in multiparty systems and we have a two-fold contribution to literature. First, we explore the way power is conceptualized in multiparty systems. Second, we investigate which predictions and propositions of the Social Distance Theory of Power and the Approach Inhibition Model of Power can be used to integrate research on power distribution and dynamics in multiparty systems. We extend the predominantly experimental empirical support of these two theories with insights from the multiparty systems literature. With respect to the way in which power is conceptualized in the multiparty systems literature, our study shows a shift from a possession over resources to a relational perspective on power in the last decades. Moreover, based on the insights of the two psychological theories of power, the study reflects upon the benefits and drawbacks of high versus low power for collaboration effectiveness among stakeholders, pointing towards ways in which facilitators can work with power differences in multiparty systems. Finally, the study points toward directions for future research concerning power dynamics in multiparty systems.

1. Introduction

Power is an inherent phenomenon in social systems, and it shapes the social structure and the nature of interpersonal and intergroup relations [1,2]. Multiparty collaborative systems (MPCSs) are social systems composed of representatives from different organizations (further on referred to as stakeholders or parties) that come together to address complex societal issues that cannot be addressed by organizations alone [3]. The topics range from environmental issues [4,5], regional development and complex construction projects [6,7], or education [8,9]. As power dynamics emerge in multiparty negotiation and collaboration situations and taking into consideration that it has important consequences for the outcomes of these complex social systems [6,8,10,11], it is paramount to understand how power and power shifts alter the dynamics of such MPCSs.

In social systems, power is defined as an asymmetric control over valued resources [12] and it reflects the ability of powerful parties to exert influence over the least powerful ones in an attempt to achieve individual or collective goals [13]. This asymmetry in the control over resources generates a dependence of the powerless on the powerful. The literature to date often refers to power distribution and power dynamics in MPCSs, yet there is no integrative review on how power was conceptualized and explored. As such, we set out to review the literature on power dynamics in MPCSs using a framework that combines two psychological theories of power, namely the Social Distance Theory of Power (SDT) [12] and the Approach/Inhibition Model of Power (AIM) [14]. Both theories describe cognitive, emotional, and behavioral consequences of power differences in social systems and they have received substantial empirical support so far, especially from experimental research. Our paper aims to extend the literature on SDT and AIM by looking at the empirical evidence from the MPCSs literature. The research questions we put forward are as follows: (a) “How is power conceptualized in the context of MPCSs?” and (b) “What is the role of power and power differences for the affective, cognitive and behavioral dynamics of MPCSs?”

The present paper has two important contributions to the literature. On one hand, we provide a systematic and theory-driven overview of the literature on power differences in MPCSs, aiming to identify the main perspectives on power adopted in this stream of research and to highlight the dynamics in multiparty interactions driven by power differences. On the other hand, we investigate the extent to which the empirical evidence from the MPCSs literature fits the predictions and theoretical propositions stemming from AIM [14] and SDT [12]. In short, our systematic literature review integrates the literature on power differences and dynamics in MPCSs by taking the two psychological theories “into the wild”.

2. Theoretical Background

2.1. Power in MPCSs

MPCSs are complex socio-technical systems composed of multiple and diverse stakeholders that are set up to tackle complex problems (often with a significant impact on society) that cannot be solved by a single organization or group. Examples of problems addressed by MPCSs include environmental issues [4,5], regional development or complex construction projects [6,7], or complex decisions related to education systems [8,9]. The issues discussed in the context of MPCSs take on different conceptualizations across the literature, such as wicked problems [6], metaproblems [5,15], or societal concerns [8]. The common ground is that these issues lack a clear problem definition or a clear stopping point for the search of a solution, and stakeholders have to define their task and collectively engage in addressing it in order to achieve their individual and collective goals. Such MPCSs collaborate effectively to the extent to which they manage to identify, define, and ultimately achieve a joint goal [16]. Typically, MPCSs include stakeholders with diverse perspectives and objectives, yet they have the potential to identify and pursue collective, integrative goals [6,10]. Social interdependence theory (SIT) states that stakeholders may achieve their goals on their own (independence) or in interaction with others, driven either by a positive or a negative interdependence [17]. In the case of MPCSs, positive interdependence means that stakeholders realize that they can only achieve their individual goals if all stakeholders are supported to achieve theirs, while negative interdependence means that stakeholders perceive that in order to achieve one’s goal, they have to prevent others from achieving theirs [6,16,17]. Positive interdependence is, therefore, the desirable systemic state in MPCSs, yet stakeholders may sometimes engage in “the winner takes it all” dynamics.

Power viewed as asymmetric access to valued resources certainly shapes the perception of social interdependence. An equal distribution of power may foster perceptions of positive interdependence, while in MPCSs with asymmetric distribution of power, the powerful stakeholders may be tempted to abuse their power and prevent other stakeholders from achieving their goals [4,6,15,18,19]. Power can be acquired either by possessing socially valued assets [20,21] or it can emerge during social interaction [22]. The possession view on power defines it as something that stakeholders have, such as knowledge [21], formal authority, critical resources, or discursive legitimacy [20]. Stakeholders that have control over these resources are thought to be powerful, which enables them to influence others and control the formation of the group. In contrast, the emergent view on power [22] states that it is built during the social interaction created in an attempt to solve an issue of collective interest. Once stakeholders achieve power, they can further control resources. The manner power is conceptualized by those involved in the collaboration situation typically combines these two perspectives as an interplay of “Having power to” (relational power) and “Having power over” (the possession view on power).

Two theoretical perspectives are particularly relevant for addressing the dynamics of power in MPCSs, namely the Social Distance Theory of Power (SDT) [12] and The Approach/Inhibition Model of Power (AIM) [14].

According to SDT [12], power reflects asymmetries in the control over resources, which results in the dependency of those with lower levels of power on those with higher levels of power. As a result, social distance is experienced differently [12], with powerful stakeholders experiencing higher levels of social distance than powerless ones due to higher levels of self-sufficiency. Social distance reflects a sense of psychological separation from other actors (groups or individuals) [23], which will ultimately result in different approaches during social interaction [24]. For instance, in the context of MPCSs, the social distance experienced by powerful parties may be reflected in a lack of engagement with collective tasks and goals.

AIM [14] states that power results in different consequences for powerful versus powerless stakeholders based on the differential activation of the approach or inhibition systems. The activation of the approach system is typical for the powerful agents and results in a disinhibited relational approach. This is reflected by taking initiative and engaging in interaction, guided by the desire to accomplish their goals. The powerless agents experience an activation of the inhibition system and, as such, they tend to approach the situation more cautiously, focusing on threats and responding to the possibility of being punished [14]. In the context of MPCSs, AIM [14] would predict that powerful actors tend to be more engaged with other parties in the system than powerless actors. However, this engagement with the stakeholders of the MPCSs is motivated by self-interest and focused on self-gain rather than the systemic good.

Although SDT [24] and AIM [14] received substantial empirical support in psychology and related fields (especially from experimental research), these theories are not very influential in multiparty research. However, they are critical for understanding the dynamics of MPCSs for two reasons. First, as oftentimes MPCSs are composed of individuals representing groups and organizations that are engaged in interaction, these two psychological theories (especially AIM, [14]) provide relevant insights into the manner in which individual representatives experience power dynamics. Second, theoretical insights (especially derived from SDT [24]) can serve to understand systemic dynamics in MPCSs. Both SDT and AIM assert that power asymmetries shape the emotional, cognitive, and interaction dynamics in social systems [12,14]. The following section builds on the insights from SDT and AIM and further summarizes the expected effects of power for individuals engaged in MPCSs, as well as its systemic impact. In this endeavor, we will focus on issues that are congruent as well as on the conflicting perspectives of the two theories.

2.2. Implications of Power for Individuals Engaged in MPCS

With respect to emotional dynamics, SDT claims that powerful stakeholders maintain a high social distance, show lower levels of motivation to engage relationally with others, and indirectly experience less social learning opportunities resulting in empathetic inaccuracy [12]. AIM [14] proposes the idea that powerful stakeholders pay attention to others through the lens of self-interest and stereotypes motivated by self-gain, which reduces the accuracy of their perception of the collective and the integrative potential of the situation. Sun and the collaborators [25] do show that power is negatively related to perspective-taking, further leading to less helping behaviors.

In addition, SDT [12] states that those with higher levels of power tend to display socially disengaging emotions, such as pride, anger, or contempt, while those with lower levels of power display socially engaging emotions such as gratitude, guilt, and embarrassment. This comes at odds with the conceptualization proposed by AIM [14,26] claiming that those with high levels of power are characterized by positive moods and emotions, while those with low levels of power tend to be more withdrawn and to mainly experience negative emotions during interactions. Recent studies support the idea of power related to positive emotions and negative emotions [27,28]. Both theories claim that individuals with high levels of power tend to show dominance, pride, assertiveness, and anger when others make mistakes [12,14], whereas those with low levels of power show guilt or embarrassment [12,14]. Thus, despite an apparent contradiction with respect to emotional dynamics, the two perspectives are rather complementary in explaining the implications of power dynamics on emotions.

At a cognitive level, both theoretical perspectives [12,14] claim that powerful agents engage in stereotyping and view low power stakeholders in an instrumental way, looking for ways in which they may prove useful towards goal attainment [29,30]. However, the two perspectives propose different explanatory mechanisms [12,14,31]. SDT [12] refers to the role of construals in order to explain cognitive differences driven by the asymmetric levels of power. Stakeholders develop complex mental representations (i.e., construals) of others, events, and timelines that vary in abstractness [32]. Due to the perceived social distance that places individuals further from each other, powerful stakeholders will display a more abstract construal compared to the powerless [12]. The high levels of abstraction of construal will then be congruent with stereotyping powerless stakeholders such that their representations include minimal, basic characteristics that are relevant for the powerful stakeholder’s gain [12,30]. In contrast, Keltner and collaborators (AIM; [14]) propose that powerful stakeholders do not process in-depth information due to the cognitive overload generated by focusing on goal-relevant information and tasks. Both theories propose that motivation is yet another mechanism that leads to stereotyping, with powerful individuals not having the motivation to tend to the powerless [12,14].

SDT [12] discusses issues related to social comparisons, which are not included in AIM [14]. The social distance experienced by powerful stakeholders may influence social comparison in two ways. On one hand, it may reduce social comparison [33], as powerless stakeholders are perceived to be too distant to actually matter. On the other hand, when social comparison occurs, the comparisons made by the powerful stakeholders are bound to focus on the others’ dissimilarities [33]. As such, it is easy for the powerless to be perceived as irrelevant or as dissimilar but useful, which could be another reason for the perception of their instrumentality [12].

SDT [12] states that powerful stakeholders present a more abstract way of thinking, with recent evidence adding this element to AIM as well [26]. As abstract representations are resistant to change, powerful stakeholders are expected to show higher levels of confidence and less willingness to change [12]. In contrast, in line with both AIM and SDT, powerless stakeholders are more responsive to social influence due to their dependence on others [12,14]. For example, powerless individuals are not a source of influence as they are not involved in voice behaviors [34]. Tost and collaborators [35] further support the claim that powerless individuals are more easily influenced by the advice from others compared to powerful ones.

With respect to the behavioral predictions of the two theories, due to their high construal level, powerful stakeholders are able to view a certain situation from a higher perspective. In turn, this enables them to choose the goals that are most appropriate for themselves [12]. Although AIM [14] does not explicitly discuss goal selection, it advances the idea that powerful stakeholders are motivated by the rewards, with goal attainment being such a reward in the context of MPCSs. Powerful stakeholders seem to be able to identify behaviors that are consistent with the chosen goals and act in a consistent manner [26]. Moreover, regarding goal-directed behaviors, SDT [12] states that powerful stakeholders are oriented towards goal attainment without being concerned about the feasibility of their plans. As such, powerful stakeholders pursue their goals no matter how distal or proximal goal attainment is [36].

AIM [14] offers a cognitive explanation for how MPCSs with asymmetric levels of power manage to integrate, define, and attain collective goals, namely that the higher-order perspective elaborated by the powerful stakeholders can be refined and translated into feasible solutions by the in-depth information processing of less powerful actors. Similarly, SDT claims that due to their high levels of dependence, individuals with lower levels of power tend to show interest in others’ views, ask important questions, and engage in reality checks; as such, they increase the level of decision comprehensiveness within the system [12]. Thus, by often engaging in minority dissent and task conflict, the actors with low levels of power may enable the emergence of a richer perspective on the situation [8].

One last behavioral element on which SDT and AIM hold different claims pertains to self-control. On one hand, SDT [12] states that high power stakeholders display greater self-control as their high construal enables them to focus on long-term goals and avoid actions that may disrupt goal attainment. Studies show that high levels of power lead to greater self-control, which lowers abusive supervision [37]. In contrast, AIM [14] states that powerful stakeholders tend to lack self-control due to their activated approach system (i.e., orientation towards rewards), with higher chances of displaying socially inappropriate behaviors. It is possible that, despite acting in more socially inappropriate ways, high-power individuals act in this manner in areas that cannot negatively impact goal attainment.

2.3. Systemic Insights Derived from SDT and AIM

With respect to the systemic consequences of power asymmetry, where each stakeholder brings about their own goals, coregulation processes play a major role. When faced with symmetric power distribution, stakeholders become involved in mutual goal prioritization (i.e., supporting each other’s goals) [31], a process that may enable collaborative effectiveness. Power asymmetry, on the other hand, may trigger goal sacrifice (i.e., the powerless stakeholders sacrifice their own goals) or goal contagion (i.e., the powerless actors accept the goals of powerful actors as their own) [31], processes that may hinder the collaborative process. Those that have social power are the ones that influence the goals of the powerless in an asymmetrical manner, which helps maintain their initial position [38].

Moreover, powerful stakeholders may also attempt to control structural elements of the collaboration, such as parties invited to engage in the collaboration process, the choice of setting (location, date, and hour of the meetings), and the content of the meetings [39]. However, the exclusion of powerless stakeholders is observed as a possible impediment in goal achievement and a method of delegitimizing the less powerful [15]. When powerless parties are excluded from the collaboration, the final decision will not include their interests and, as such, it will be less comprehensive [3]. As a consequence, when the powerless return to their community or constituency, their efforts will be discredited, and they will be held accountable for their failure [15].

In addition, the parties involved in collaboration also exist outside the issue at the core of the MPCSs, as they are part of an institutional field, a conglomerate of organizations in the system [40], or they are simply standalone organizations. A party’s power can be conceptualized with respect to the MPCS or with respect to the entire institutional field [15,41]. Power derived from the institutional field represents a form of systemic power [42]. When the issues discussed during collaboration align with the issues of the entire institutional field, those that are powerful within the system will also be powerful in the collaboration process as well. However, when the issues are not aligned, the power dynamics may shift [15,41] and institutionally powerless actors may become powerful stakeholders in MPCSs. Such a shifting perspective on power adds to the complexity of the MPCSs as the expectations of institutional power may conflict with systemic power, a complexity that is not directly addressed neither by AIM [14] nor by SDT [12].

One cognitive element that the SDT [12] does not cover, yet AIM [14] does, refers to the type of causal inferences made regarding the outcomes. Keltner and collaborators [14] state that those with higher levels of power tend to see themselves as responsible for the outcome (i.e., make internal attributions for goal attainment), while those with lower levels of power tend to see others as responsible for the outcomes (i.e., make external attributions). Studies outside MPCSs did not show a direct effect of internal causal attributions made by the powerful, yet they did show an interaction effect between power levels and the success on these attributions, with the powerful assuming responsibility only in the case of success [43].

Thus, current theoretical approaches regarding power in multiparty collaboration are not oriented towards integrating multiple consequences of power, for both low-power and high-power stakeholders. By integrating perspectives from both AIM [14] and SDT [12], this study represents an effort to establish how power is viewed in MPCS, as well as to analyze the effects of its dynamics. Specifically, our aim is to explore the following: (a) “How is power conceptualized in the context of MPCS?” and (b) “What is the role of power and power differences for affective, cognitive and behavioral dynamics of MPCS?”

3. Materials and Methods

In order to answer the research questions, the initial search was conducted in April 2021. We searched for relevant articles in four databases, namely ProQuest, Scopus, Web of Science, and Google Scholar, in this order. Because our interest was towards power dynamics in MPCS, we used the (“Multiparty collaboration” AND Power) AND (Emotions OR “Emotional Regulation” OR “Emotional Climate” OR Aspirations OR Attributions OR “Decision Making” OR “Decision Effectiveness” OR “Decision Comprehensiveness”) search string. The keywords proposed for the search string mirror the goals of the present literature review, with an accent on power dynamics in multiparty systems at emotional, cognitive, and behavioral levels. We used the same keywords to perform the search within all mentioned databases. Because the search in Web of Science resulted in more than 10,000 results, in this particular case, the string was modified to include only “Multiparty” AND “Power” and we selected only the journals from fields relevant to the subject (such as psychology, management, business, economy, public administration, and communication). As this is one of the first efforts in integrating research on power dynamics in MPCSs, our search covered a large time window and we did not impose any year constraints on the results.

At the initial search, we received 71 entries from ProQuest, 1 from Scopus, 837 from Web of Science, and 392 from Google Scholar, with a total of 1301 entries. The initial screening phase involved making a decision based on the title of these publications. In order to be included in the analysis, the papers needed to make reference to power or MPCSs. After the title-based screening, 68 papers were selected for entering abstract-based screening. Based on the abstract reading, 18 papers were further selected. The eligibility criteria applied for all further steps included the following: (a) discussing power in a multiparty context, (b) discussing power in relationship with collaboration outcomes, and (c) the study is empirical or built on empirical examples. The final step involved screening based on the entire text, with a total of 8 papers were saved.

While reading the selected papers, based on their reference list, 2 papers were further selected as potentially relevant for the present review and they were next assessed as eligible. After this step, we searched the Google Scholar profiles of each author present in the selection. Sixteen other papers were selected based on title and abstract. Six of them were kept for further analysis.

As several months have passed since the onset of the study, we searched the databases once more to identify papers published within the past 12 months (during 2021). The search in ProQuest rendered five results, Scopus and Web of Science rendered zero results, and Google Scholar rendered 32 results. Based on abstract screening, 3 papers were selected for full-text reading. None of the three papers met the criteria for being included in the review.

We identified two additional papers referred in the papers selected based on the search of the literature on MPCSs. We applied the same inclusion criteria (a) discussing power in a multiparty context, (b) discussing power in relationship with collaboration outcomes, and (c) the study is empirical or built on empirical examples. In Table 1, we present the papers included in the review and a summary of the findings regarding power.

Table 1. Summary of papers included in the study.

Paper	Design	Findings on Power Dynamics
Larson (2003) [4]	qualitative	Low-power stakeholders are more oriented towards collaboration compared to high-power stakeholders. Low-power stakeholders add to the collaboration by anticipating problems and preventing them, adapting strategies to context-specific situations, and managing conflict resolution.
Turcotte et al. (2008) [5]	qualitative	Power is an episodic phenomenon and it plays an important role in formalizing knowledge. Stakeholders become involved in interaction with a clear motivation such as gaining a competitive advantage, which leads to empowerment. The effects of the collaboration go beyond the collaborative system, impacting the institutional field.
Curșeu and Schruijer (2020) [6]	quantitative	Powerful stakeholders become involved in interaction and exclude powerless stakeholders, with negative effects on the systemic level outcomes (such as goal achievement, collaborativeness, and power).
Fleștea et al. (2017) [8]	quantitative	Power differences lead to lower levels of psychological safety and higher levels of relational conflict. Controlled cognition on the part of low-power stakeholders leads to cognitive dissent and task conflict.
Trif et al. (2020) [9]	quantitative	Powerful stakeholders make internal causal attribution for the outcomes of the collaboration. Causal attributions explain the relationship between power and perceived climate and between power and future collaborative intentions.
Olekalns et al. (2007) [18]	quantitative	High power stakeholders have to prove that they have the same goals to be included in coalitions (that they will not use low power stakeholders to achieve their own goals). For low-power stakeholders, both relative and absolute trust are important in securing resources. When they show low levels of calculus-based trust and high levels of knowledge-based trust, they will fare better. For medium power, only absolute trust is a significant predictor.
Tello-Rozas et al. (2015) [19]	qualitative	In the process of goal setting, high-power stakeholders take initiative and show more goal-directed behavior. There are three actions in ensuring engagement in MPCSs—mobilizing, organizing, and acting. In the mobilizing stage, the powerful parties use their authority to mobilize others and empower them by allocating resources (persuasion). In the organizing stage, power is gradually shared with others in an effort to empower via delegation and creating opportunities. In the acting stage, power is useful in ensuring proposals and partnerships for collaboration.
Hardy and Philips (1998) [20]	qualitative	When power is distributed, collaboration is facilitated. When power is imbalanced, compliance is the prevalent strategy. The powerless are dependent on the powerful in order to make themselves heard.
Marshall and Rollinson (2004) [21]	qualitative	Stakeholders that perceive themselves as powerful (i.e., in terms of expertise) show confidence in themselves and do not pay attention to others. Authority as a source of power enables higher levels of control.
Purdy (2012) [39]	qualitative	Powerful stakeholders (enabled by formal authority, resources, or discursive legitimacy) present goal-directed behaviors by selecting the participants, the process (time, location), and the content of the discussion.
Acey (2016) [44]	qualitative	Due to power asymmetry, powerful stakeholders define both the problem and the solution.
Gray and Hay (1986) [45]	qualitative	Due to their motivation to maintain their status, high-power stakeholders will not be easily persuaded, while low-power stakeholders are open to change as it may change their position within the system. Goal-directed behaviors displayed by powerful stakeholders include limiting participation to parties that would facilitate consensus. Lower levels of power are related to the decision to not participate. The lack of representation leads to the impossibility of influencing the institutional field.
Hardy (1998) [46]	qualitative	In very organized domains, high-power stakeholders are privileged in terms of access to information and resources. In under-organized domains, less powerful parts are able to assert influence. One statement in an interview shows the perceptions of refugees (as a low-power stakeholder) in a stereotypical manner.
Hardy and Philips (1998) [47]	empirical example	Despite the initial perception of being trustworthy, high-power stakeholders may only use low-power stakeholders in an instrumental manner in order to secure their own win (powerful stakeholders develop a façade of trust).
Antonova (2007) [48]	qualitative	Powerful stakeholders take on the goal-setting process and are proactive towards accomplishing their goals. When power dynamics are misunderstood, some parties assume more powerful positions that enable them to take control over the situation.
Antonova (2014) [49]	qualitative	Power dynamics appear in the problem setting stage and direction-setting stage. Powerful stakeholders set goals in accordance with their needs, without taking into account low-power stakeholders. High-power stakeholders present privileges regarding choosing strategies. One strategy used is limiting the other stakeholders’ access. In the direction-setting stage, powerful stakeholders take control over the agenda and the solutions deemed acceptable.
García-López and Arizpe (2010) [50]	qualitative	In top-down approaches, powerful stakeholders avoid mutuality by excluding parties perceived as lacking power. By performing this, powerful stakeholders are in control of problem definition processes. A bottom-up, instead of a top-down approach to the situation, would enable low-power parties to become more involved.
Vangen and Huxham (2003) [51]	qualitative	Perceived power disparity may be an impediment in building trust due to the risk of assuming credit for the results.

Table 1. Summary of papers included in the study.

Paper	Design	Findings on Power Dynamics
Larson (2003) [4]	qualitative	Low-power stakeholders are more oriented towards collaboration compared to high-power stakeholders. Low-power stakeholders add to the collaboration by anticipating problems and preventing them, adapting strategies to context-specific situations, and managing conflict resolution.
Turcotte et al. (2008) [5]	qualitative	Power is an episodic phenomenon and it plays an important role in formalizing knowledge. Stakeholders become involved in interaction with a clear motivation such as gaining a competitive advantage, which leads to empowerment. The effects of the collaboration go beyond the collaborative system, impacting the institutional field.
Curșeu and Schruijer (2020) [6]	quantitative	Powerful stakeholders become involved in interaction and exclude powerless stakeholders, with negative effects on the systemic level outcomes (such as goal achievement, collaborativeness, and power).
Fleștea et al. (2017) [8]	quantitative	Power differences lead to lower levels of psychological safety and higher levels of relational conflict. Controlled cognition on the part of low-power stakeholders leads to cognitive dissent and task conflict.
Trif et al. (2020) [9]	quantitative	Powerful stakeholders make internal causal attribution for the outcomes of the collaboration. Causal attributions explain the relationship between power and perceived climate and between power and future collaborative intentions.
Olekalns et al. (2007) [18]	quantitative	High power stakeholders have to prove that they have the same goals to be included in coalitions (that they will not use low power stakeholders to achieve their own goals). For low-power stakeholders, both relative and absolute trust are important in securing resources. When they show low levels of calculus-based trust and high levels of knowledge-based trust, they will fare better. For medium power, only absolute trust is a significant predictor.
Tello-Rozas et al. (2015) [19]	qualitative	In the process of goal setting, high-power stakeholders take initiative and show more goal-directed behavior. There are three actions in ensuring engagement in MPCSs—mobilizing, organizing, and acting. In the mobilizing stage, the powerful parties use their authority to mobilize others and empower them by allocating resources (persuasion). In the organizing stage, power is gradually shared with others in an effort to empower via delegation and creating opportunities. In the acting stage, power is useful in ensuring proposals and partnerships for collaboration.
Hardy and Philips (1998) [20]	qualitative	When power is distributed, collaboration is facilitated. When power is imbalanced, compliance is the prevalent strategy. The powerless are dependent on the powerful in order to make themselves heard.
Marshall and Rollinson (2004) [21]	qualitative	Stakeholders that perceive themselves as powerful (i.e., in terms of expertise) show confidence in themselves and do not pay attention to others. Authority as a source of power enables higher levels of control.
Purdy (2012) [39]	qualitative	Powerful stakeholders (enabled by formal authority, resources, or discursive legitimacy) present goal-directed behaviors by selecting the participants, the process (time, location), and the content of the discussion.
Acey (2016) [44]	qualitative	Due to power asymmetry, powerful stakeholders define both the problem and the solution.
Gray and Hay (1986) [45]	qualitative	Due to their motivation to maintain their status, high-power stakeholders will not be easily persuaded, while low-power stakeholders are open to change as it may change their position within the system. Goal-directed behaviors displayed by powerful stakeholders include limiting participation to parties that would facilitate consensus. Lower levels of power are related to the decision to not participate. The lack of representation leads to the impossibility of influencing the institutional field.
Hardy (1998) [46]	qualitative	In very organized domains, high-power stakeholders are privileged in terms of access to information and resources. In under-organized domains, less powerful parts are able to assert influence. One statement in an interview shows the perceptions of refugees (as a low-power stakeholder) in a stereotypical manner.
Hardy and Philips (1998) [47]	empirical example	Despite the initial perception of being trustworthy, high-power stakeholders may only use low-power stakeholders in an instrumental manner in order to secure their own win (powerful stakeholders develop a façade of trust).
Antonova (2007) [48]	qualitative	Powerful stakeholders take on the goal-setting process and are proactive towards accomplishing their goals. When power dynamics are misunderstood, some parties assume more powerful positions that enable them to take control over the situation.
Antonova (2014) [49]	qualitative	Power dynamics appear in the problem setting stage and direction-setting stage. Powerful stakeholders set goals in accordance with their needs, without taking into account low-power stakeholders. High-power stakeholders present privileges regarding choosing strategies. One strategy used is limiting the other stakeholders’ access. In the direction-setting stage, powerful stakeholders take control over the agenda and the solutions deemed acceptable.
García-López and Arizpe (2010) [50]	qualitative	In top-down approaches, powerful stakeholders avoid mutuality by excluding parties perceived as lacking power. By performing this, powerful stakeholders are in control of problem definition processes. A bottom-up, instead of a top-down approach to the situation, would enable low-power parties to become more involved.
Vangen and Huxham (2003) [51]	qualitative	Perceived power disparity may be an impediment in building trust due to the risk of assuming credit for the results.

As such, this review now includes 18 papers: 16 from the initial search and 2 from the final search.

In the next step, the findings reported within the 18 papers were analyzed and integrated within the core theoretical frameworks selected for this study: SDT and AIM.

4. Results

4.1. Conceptualization of MPCSs

The way in which collaboration among multiple parties or stakeholders was labeled in the literature differs across studies and various terms are used, such as: multiparty collaboration [6,9], multi-team systems [8], multilateral negotiations [4], multistakeholder organization/governance [5,44], three-party negotiation [18], large-scale social-driven collaborations [19], interorganizational domain collaboration/relations [4,20,21,45,46,47], multistakeholderism [48,49], or participatory processes [50]. Common to all these approaches is the diversity of stakeholders that engage with a complex issue or challenge, which involves trying to come to a joint problem definition, to generate alternatives and make a decision, to negotiate and reach consensus regarding the issue at hand. Such MPCSs display complex relational dynamics in which power plays a critical role as it impacts both the individual stakeholders, by shaping the way they think, feel and act, as well as the system as a whole.

4.2. Perspectives on Power and Power Differences

Despite power being a pervasive element in MPCSs [4,6,18], researchers tend to diverge in their definitions of power and how power is acquired in social systems. Still, one point of substantial consensus in the literature is that power represents the ability to influence others to act according to one’s will [13].

Power may be acquired either by possessing socially valued resources, formal authority, or legitimacy [4,20,21,44,45,46], or based on a relational perspective focused on attempts to use social influence tactics. The latter further covers the emergent view on power [6,8,9], the practice view [19], or forms of episodic power [5]. Each perspective on power will be briefly presented in the following part.

The possession perspective on power states that power is something stakeholders possess before entering the collaboration situation, which will further influence the outcomes in their favor [52]. There are three main sources of power that stakeholders may control when entering a collaboration process: formal authority (i.e., status conferred by an organization), control over valued resources (e.g., financial or human resources), and legitimacy (i.e., ability to speak for an issue) [20,23,39,46]. Although knowledge itself has been considered a source of power [53], it is argued that knowledge without expertise or authority does not lead to effective social influence [21]. According to Larson [4], those that do not possess power (from a symbolic, social, or economic perspective) are seen as disadvantaged, with limited courses of action.

In MPCSs, the objective power a stakeholder possesses (i.e., authority, resources, legitimacy) may not bear as much importance because, by creating coalitions, stakeholders are able to multiply their power [18]. As such, the relational view on power moves towards highlighting the idea that power is not something that stakeholders have, but it rather develops through the dynamics taking place in MPCSs, when stakeholders engage in persuasion, display authority, and instill coercion during the process of identity formation [22]. Building on Turner’s [22] perspective on power, interpersonal perception [54], and the social relations model [55], Trif and collaborators [9] proposed a conceptualization of power integrating the idea that, during interactions unfolding in MPCSs, each stakeholder is both the target and the perceiver of power differences. Namely, they argued that as the interactions among stakeholders unfold, the perceptions on the self-attributed power, the power that is attributed to others, as well as the power that is attributed to an actor by the other stakeholders in the system change as well and they might alter differently the subsequent dynamics of the MPCSs.

The practice view on power [19] bridges the two main perspectives on power (i.e., possession perspective and relational perspective). What begins with the possession of power by certain stakeholders (i.e., power imbalance) is then redistributed among the parties as needed [56]. Thus, although collaboration starts with a certain distribution of power among parties—with some possessing more power than others, initial power distribution is often not decisive, as coalitions can shift the initial power configuration in the system. Shared power also involves initial levels of power as a possession that is then shared among stakeholders through empowerment [48,49]. Turcotte and collaborators [5] and Lawrence [57] take a similar perspective and show that even if stakeholders may not have initial levels of power, they will gain power during interactions in what are described as episodes of power.

Lawrence [57] also proposed the idea of systemic power, directly tackled by Curșeu and Schruijer [6], showing that power does not only describe stakeholders involved in collaboration but it also describes the entire system in which they interact. In particular, in MPCSs that excluded relevant stakeholders, the perception of systemic power decreased in time [6].

The documented effect of power differences on the manner stakeholders engage in interaction differs depending on the theoretical view on power that is employed. For instance, when power is conceptualized from the possession view, according to the most influential theories of power, the lack of power leads to lower levels of influence and the lack of involvement in the psychological formation of the group [22]. In this case, low-power stakeholders display a more compliant approach to the situation [4,22], as they cannot instill any change. When power is conceptualized from a relational point of view, the lack of power leads to less control over resources, yet it does not limit the possibility of asserting social influence [22]. In this latter case, less powerful stakeholders engage in interaction in order to gain influence, which further leads to power [8,9]. These differences impact the emotional, cognitive, and behavioral levels differently.

4.3. The Individual-Level Effects of Power in MPCSs

We will further discuss the dynamics of power in MPCSs according to AIM [14] and SDT [12], beginning with the core assumption related to engagement. The results show that powerful stakeholders are indeed motivated by selfish reasons to engage in interaction [9], thus supporting the proposition derived from AIM [14]. When power is conceptualized from the relational perspective, the involvement of powerful stakeholders within the MPCSs is most evident as they need to secure coalitions with other stakeholders [18], secure their position (e.g., by gaining competitive advantage) [5], and ensure that their own goals will be represented in the final solution [6]. Another reason supporting the active involvement of the powerful parties could be their effort to build false trust that could further enable them to control the powerless [47]. Interestingly, in the context of MPSs, low levels of power lead to less cooperative intention for future interactions, and thus lower levels of future engagement [9]. When low-power stakeholders see others as responsible for the outcomes in the collaboration setting, they display a defensive reaction and orient towards gaining power and reducing collaboration [9]. As in this study power is analyzed from a relational perspective, it seems that low-power stakeholders may be motivated by gaining more power when entering MPSs.

The proposed theories diverge in their arguments concerning the emotional experience tied to power dynamics. One theory proposes differences regarding the valence of emotions [14], while the other explains emotional reactions based on their social engagement level [12,31]. Although no paper directly discusses power levels and emotions in MPCSs, based on the finding that asymmetric power distribution is positively related to relational conflict [8], we may expect a rise in the level of negative emotional reactions experienced by those that are targeted by the conflict. Moreover, power disparity acts as an impediment in the way of building trust, as low-power stakeholders fear that they may be used as means to achieve other goals then their own [9,51]. Trust reflects the others’ perceptions regarding positive behavioral intentions or benevolence [58], yet the beginning of the collaborative situation is often characterized by distrust [16]. If parties do not manage to tolerate distrust and engage in open interaction, they will not discover their interdependencies. As such, less powerful parties will be anchored in their dependency on others, which could lead to negative emotional reactions. Lastly, as power imbalance arises and stakeholders ascribe more power to other parties, the intergroup climate is perceived as less auspicious. Powerless stakeholders become increasingly aware of the threats coming from the powerful [9], but also of their dependence on the powerful, which may ultimately trigger negative emotions. Thus, the literature on power in MPCSs did not directly focus so far on emotional and affective consequences of power, and the only results in this direction underline a positive effect of power disparity on negative emotions.

With respect to the cognitive aspects, empirical studies on MPCSs do not focus on exploring the effects of power on stereotyping, yet both theories consider this phenomenon [12,14]. However, there is one qualitative study [46] that points towards a tendency of the most powerful parties (i.e., government officials in this case) to engage in stereotyping the powerless (i.e., the refugees during the refugee crisis).

Another common element of SDT and AIM is related to low-power stakeholders being perceived in an instrumental manner [12,14]. Studies focused on trust and psychological safety acknowledge this effect in both quantitative [9,18] and qualitative methodologies [47,51]. As a strategy, the stakeholders with high levels of power tend to act nice and establish a façade of trust in order to be able to control the powerless without them realizing it [47]. As low-power stakeholders become aware of these tendencies, they feel less safe and do not take further risks by engaging in disclosure and open communication [8]. One possible concern is related to the powerful parties taking over the merits of the powerless [48]. In order to avoid instrumentality, the powerless parties create alliances, and they only include powerful agents if they can establish identity-based trust (perception of common goals) [18]. The SDT and AIM propose multiple mechanisms explaining instrumentality, yet they were not empirically tested so far. A mechanism that is common to both approaches concerns the fact that the powerful parties are not motivated to attend to those with low levels of power. Although this is applicable when perceived interdependence is low [18], when interdependence is high, powerful stakeholders have the motivation to attend to the powerless, as the former are at risk of being excluded from coalitions by the latter [18].

With respect to the way power shapes cognitive and behavioral dynamics, the findings show that powerful individuals have higher confidence levels and are less likely to change, with the opposite pattern occurring for the low power stakeholders [12,14]. In line with Turner’s view [22], persuasion is an important process in gaining power especially in the initial steps of the collaboration [19] and it further leads to control over resources. Ultimately, when acknowledging the positive interdependence of goals, powerful stakeholders will facilitate interaction and create collaborative opportunities. However, normally, those with high initial levels of power are more resistant to persuasion, as they will try to maintain their position by any means. In the case of coal policy discussions, in order to avoid disagreement, the strategy involved limiting the access of important stakeholders [45]. Yet, this strategy proved to be detrimental, with negative consequences outside the MPS. Interestingly, when power is analyzed from a relational perspective [5], it seems easier to persuade powerful parties to become involved in collaborative action as they may lose their power otherwise (in this case, by not learning). When power is conceptualized from a possession perspective [45], avoiding change in MPCSs is a viable option, as power cannot be easily shifted. However, these approaches lead to different systemic outcomes, as it will be further discussed.

Goal setting and goal-directed behaviors are also issues addressed by the two theories. One element is proposed only by SDT [12]—those with high levels of power influence the goal-setting process. Powerful stakeholders do not only set goals but also define the issues MPCSs should focus on, such as moving the focus from social inequalities towards the development of profitable business [44]. In asymmetric power contexts, powerful stakeholders prioritize their own goals, for example, by making a list of discussion points, while other parties have to adapt [39]. For example, Hardy [46] shows that, in the context of MPCSs dealing with refugee situations, the powerful tend to accentuate goals such as security and sovereignty over refugee rights. In addition, in the context of internet use, Antonova [48,49] proposes that those with high levels of power set goals that are in accordance with their needs, without paying attention to the powerless. Hardy and Philips [20] discuss that, in order to have their goals represented, stakeholders must either have power or be able to attain the goodwill of the powerful parties. Tello-Rozas and collaborators [19] state that, at least in the beginning of the collaboration cycle, powerful stakeholders tend to set relevant goals in order to subsequently assure the empowerment of other less powerful stakeholders.

Powerful stakeholders are also thought to show more goal-directed behaviors [12,14] and to be more proactive towards accomplishing their goals as compared to the powerless. Powerful stakeholders show more behavioral activation and engage in more interaction with others in order to serve their own goals, without being driven by prosocial motivation [9]. However, even when the powerful aim to empower the other stakeholders, they still have to take action at least in the beginning of the collaboration process by setting the stage [19]. There are also goal-directed actions that are openly self-promoting at the expense of the other parties. For example, in one study, powerful stakeholders limited the access of other relevant parties to the collaboration process by not inviting them to the plenary discussion session [6]. Other studies report findings showing that powerless stakeholders may not be included in the collaboration process from the very beginning, when their goals are perceived as not aligned with the powerful stakeholders’ goals [39,44,45,49]. In addition to limiting access, other goal-directed behaviors include setting the location of the meeting, the time of the meeting [39], or proposing a solution [44] and strategies [49].

With respect to the level of cognitive processing displayed during interactions, stakeholders with high levels of power seem to worry less about the feasibility of their goals and chosen behaviors [12]. On one hand, those with low levels of power may complement this orientation with their controlled, planned cognition [12,14]. In the context of climate change negotiations, Larson [4] shows that stakeholders with low levels of power have a transformative effect in adapting strategies, as they are able to take into consideration contextual restraints. Translating cognition into behavior, the strategies adopted by those with low levels of power tend to be more cooperative than those implemented by stakeholders with high levels of power [4]. On the other hand, as low power stakeholders engage in in-depth processing, task conflict and cognitive dissent arise within the system, with positive implications for the outcomes of the collaboration situation [8]. Thus, as power is conceptualized from a possession point of view, low-power stakeholders engage in adaptative processes that are tolerated by the powerful [4]. When power is conceptualized from a relational point of view, low-power stakeholders use their controlled cognition to enhance the whole collaborative process with positive conflict [8].

The explanation for the goal-oriented behaviors of the powerful stakeholders also differs between the two theoretical perspectives. SDT [12] proposes the construal level as a plausible explanation, whereas AIM [14] proposes reward orientation as a viable explanation. Curșeu and Schruijer [6] suggest that, as powerful stakeholders focus on their view on the problem, they will not take into consideration other perspectives and end up excluding low-power parties from the collaboration process. Thus, the level of abstraction and disinterest towards the views of the powerless stakeholders may represent a viable explanation, which is in line with SDT [12].

Self-control is the last element discussed in SDT and AIM, with the two approaches holding conflicting views. SDT [12] states that those with high levels of power have higher levels of self-control, while AIM [14] states that high-power individuals do not have self-control. This issue was not explored by any of the empirical studies included in the review and may warrant attention in the future.

4.4. Systemic Dynamics of Power

Due to the multi-level nature of MPCSs, individual goal-setting behaviors such as goal prioritization, goal sacrifice, or goal contagion present cross-level effects [31]. Mutuality of goal-setting (i.e., jointly discussing and prioritizing goals) can be achieved especially when power is conceptualized as relational—for example, when focusing on civil society’s empowerment, which leads to higher levels of empowerment of those involved in the interaction [19]. In order to achieve mutuality, trust and psychological safety play an important role, as they foster positive interaction among the parties involved [8,18,44]. As stakeholders interact, if a base of trust is not set, the results are subpar because goal interdependencies are not explored, limiting the possibility of mutuality. However, as most of the times parties do not acknowledge the positive interdependence, trust and safety do not emerge within the system, and goal mutuality is not possible. Hardy and Phillips [20] show that, in the context of the UK refugee system, power disparity brings about cooperation by compliance on the part of low power individuals, a strategy that restricts change. In a study on the internet governance domain taking a possession view on power [48], when the powerful stakeholders are reluctant to engage in mutuality of goal prioritization, sacrifice and contagion emerge and the entire system is negatively affected, leading to failure. One action undertaken by powerful stakeholders in order to avoid mutuality is to deploy a top-down strategy and exclude parties perceived as low in power and not useful [50]. As already discussed, this strategy may backfire [45], as it influences the entire systemic dynamics, leading to lower overall goal attainment [6]. When certain stakeholders are excluded, the problem is not comprehensively analyzed, and goal formulation is not inclusive for all parties [6]. When power is conceptualized from a possession perspective, the goals of the powerless parties are sacrificed by restricting their visibility [39]. If the context is well-structured and characterized by clear rules and governmental support, powerless stakeholders have to engage in goal sacrifice or goal contagion [46]. When the collaboration situation is seen as underorganized [59], powerless stakeholders may be able to secure their goals in the system, resulting in change [46].

Decisions made in MPCSs have consequences outside of the system and reverberate into the institutional field. The quality of interaction and goal alignment influences the extent to which the system’s decision influences the entire institutional field. The findings of one study dedicated to learning show that by building and formalizing knowledge, stakeholders that are initially powerless in the system manage to gain visibility and, thus, become important players [5]. When power is conceptualized from a relational point of view, episodes in which power shifts in order to facilitate learning and development lead to positive consequences in the entire institutional field, such as permanent change [5]. In contrast, Gray and Hay [45] show that when the stakeholders involved in the collaborative situation overlooked the participation of those with high systemic levels of power, their ability to implement their proposals was restricted. Thus, the extent of the influence of a collaborative situation may or may not extend to the institutional field based on elements such as the inclusion of relevant stakeholders.

AIM [14] proposes that powerful stakeholders tend to assume responsibility for the outcomes of the collaborative situation. One study shows that high levels of power (seen as high levels of power self-attributed and attributed by others) lead to internal causal attributions (i.e., perceiving internal causes for the results that were obtained during collaboration). Those with low levels of power (seen as high levels of power attributed to others) tended to make external causal attributions for the outcomes of the collaboration, on the other hand [9]. As powerful stakeholders assume responsibility for the outcomes, they feel in control and responsible for the rest of the parties involved. Thus, they perceive the climate as auspicious and they are willing to engage in future collaboration.

The results of the present analysis are summarized in

Table 2. Individual and systemic power dynamics in MPCSs according to SDT and AIM.

	Emotional Dynamics	Cognitive Dynamics	Behavioral Dynamics
Individual level	No studies directly addressing emotional dynamics. Findings regarding relational conflict, climate, trust and safety [8,9,51] offer support for AIM predictions regarding the effects of power on the valence of emotions.	Little support for stereotyping tendencies, proposed by both theories [46].	As AIM proposes, powerful stakeholders engage in interaction guided by self-serving reasons [5,6,9,18,47].
		Support for instrumentality of low-power stakeholders, as proposed by both theories [18,47].	Powerful stakeholders influence low-power parties [19] and resist persuasion unless clear incentives are at stake [5,45].
		Brief support for low motivation to attend to the powerless, a mechanism proposed by both theories [18].	As SDT proposes, powerful stakeholders engage in goal-setting behaviors [19,20,44,46,48,50].
		Goal orientation explained by the SDT proposed mechanisms of abstract thinking [6].	As both theories propose, powerful stakeholders engage in goal relevant behaviors [6,9,19,39,44,45,49].
		Low-power stakeholders present more controlled cognition and attention to others, with differences in their approach based on the conceptualization of power. From a possession perspective on power, low-power individuals adapt to the powerful by the means of their attention to others. When power is conceptualized as relational, less-powerful stakeholders engage in task conflict, with positive consequences for the system [4,8].
Systemic level		Powerful stakeholders assume responsibility for the outcomes [9].	Mutuality is achieved when stakeholders operate from a social perspective on power [19]. When power is conceptualized as possession [48], it is not achieved. The lack of mutuality negatively impacts trust [18,20,51]. Exclusion is used to avoid mutuality [6,45,50].
			The results may be included at higher levels, outside the situations [5] with the condition of involving relevant stakeholders [45].

.

5. Discussion

The aim of the current review was two-fold. Firstly, we set out to explore how power is conceptualized in research on MPCS. Secondly, our goal was to analyze the effects of power in MPCSs through the lenses of two major psychological theories on power—The Approach/Inhibition Theory [14] and The Social Distance Theory [12].

Building on understanding power as the ability to exert influence over others [13], literature proposes two main approaches on power—the possession view and the relational view. Studies on MPCSs adopt both perspectives, with a focus on the possession view in earlier works [4,18,20,21,39,45,47] and a focus on the relational perspective in the more recent studies [6,8,9]. MPCSs typically address complex or wicked problems, with high levels of interdependence [60], in which stakeholders have to integrate multiple, diverse perspectives on the issue at hand. Within this context, the relational perspective may be more suited for analyzing power dynamics.

The second goal of the study focused on exploring the effects of power dynamics for emotional, cognitive, and behavioral levels by integrating the predictions of The Approach/Inhibition Model of Power [14] and The Social Distance Theory [12]. Given the complexity of MPCSs [6,8,10,11], power dynamics influence both the stakeholders and the system as a whole; therefore, our results are discussed both at the individual as well as at the systemic level.

SDT [12] proposes that powerful stakeholders perceive higher levels of social distance and AIM [14] proposes that powerful stakeholders have an activated approach system. The main theoretical divergence refers to the prediction concerning the type of engagement powerful stakeholders have in the collaborative situation. Results support the idea that powerful stakeholders become involved in the interaction, yet with self-directed motives such as securing their powerful initial positions [6,18], therefore supporting the predictions of AIM.

At the stakeholder level, the results point towards overlapping elements between the two theories, such as the fact that powerful stakeholders use powerless agents in an instrumental way [18,47] and display goal-directed behaviors [6,19,39,45,48], while powerless stakeholders display systematic analysis and increased depth of information processing [4,8,9]. Other elements tackled by the two theories are complementary such as the fact that powerful stakeholders tend to be more persuasive [19], as proposed by AIM, while, in certain contexts, they are more resistant to persuasion [45], as proposed by SDT. Other findings seem to provide more support to one theory over the other. For instance, the results pointing towards the specificity of the valence of emotions experienced by the powerful versus the powerless lend more support to the predictions of AIM [14]. Moreover, the fact that powerful stakeholders tend to prioritize their own goals [19,39,46,48,49] is in line with SDT predictions [12].

At the systemic level, interesting results were related to goal-setting in MPCSs. When stakeholders with different goals engage in interaction, processes of goal prioritization, goal sacrifice, or goal contagion could arise on one or multiple parties’ behalf [31]. When power is conceptualized as more relational, stakeholders have higher chances of engaging in mutual goal-related behaviors [19]. When mutuality does not arise, it ultimately reduces the collaborative effectiveness of the entire system [6,48]. From a possession perspective on power, mutuality appears to a lesser extent [39,46], as the dependency of the powerless stakeholders on others constrains their freedom of action. In addition, at the systemic level, stakeholder exclusion and inclusion play an important role regarding the long-term consequences of their collaboration, such as being included in the larger institutional field [5,45]. Finally, it seems that powerful stakeholders do assume responsibility for the outcomes of the collaboration setting [9].

There are some issues that seem to be contrasting in the results presented by the studies included in the review. First, there are differences regarding the resistance to persuasion on the part of the powerful. When power is conceptualized from a relational perspective, powerful stakeholders show more willingness to change, as they are motivated to maintain their status [5,19]. For power conceptualized from a possession view, powerful stakeholders show the opposite pattern, even avoiding situations that could trigger change, such as the presence of certain stakeholders [45].

Another element regards the complementary role of high and low power individuals in goal-directed behaviors within MPCSs. High-power stakeholders take initiative and propose directions of action without taking into consideration their feasibility [12]. On the other hand, low-power stakeholders engage in controlled cognitive processes that enable them to identify weak points in the proposed courses of action. Moreover, due to threat-avoidance tendencies, they are able to adapt these goal-achievement strategies [14]. When power is analyzed from a possession perspective [4,20], stakeholders with lower levels of power deploy cooperation strategies, which are related to compliance and adaptation to the paths proposed by the powerful. However, when power is analyzed from a relational perspective, the possibility of power shifts is related to cognitive dissent and task conflict with beneficial effects for the whole collaborative context [8]. Thus, as the conceptualization changes, the focus of in-depth thinking shifts from others (in the case of the possession perspective) to the whole system (in the case of the relational perspective).

As power can be seen as shared [19] and despite some stakeholders entering the collaboration with higher levels of power, through empowerment, power distribution in MPCSs can be shifted. Hardy [46] also looks into the destabilization of power structures through disorganizing forces. Until now, theories did not include factors that could explain shifts of power in the systems, despite it being a relevant characteristic especially for the relational perspective of power.

5.1. Implications

This review has multiple implications, both theoretical and practical. At the theoretical level, it is one of the first attempts to review and integrate the literature on power dynamics in complex MPCSs. First, we tackled the issue regarding the conceptualization of power in MPCSs. In this regard, we show that the perspective on power has shifted from a possession perspective in the early years [4,18,20,21,39,45,47] to a relational perspective in the most recent years [6,8,9]. These results support the importance of identifying positive interdependence [17] in such complex systems that require constant social exchanges in order to come to joint problem formulations and to identify or formulate common goals. In order to identify common ground and the integrative potential in the multiparty situation, it is essential that the inputs of every party are taken into consideration and integrated with the common superordinate goals. When power is conceptualized as dynamic (i.e., as being transferred and flowing through the system), stakeholders are encouraged to interact and collaborate, facilitating the identification of superordinate goals. Such a dynamic perspective on power opens valuable venues for research and applications in the context of MPCSs.

Second, the review integrates two main theoretical perspectives on power, namely SDT and AIM [12,14], in order to understand the effect of power on collaboration effectiveness in the context of MPCSs. These results fill the gap regarding the need of a theoretical framework dedicated to power dynamics in MPCSs. According to the insights of our review, it is not the level of systemic power itself that influences collaborative effectiveness but the manner in which power asymmetries are dealt with. For example, excluding less powerful parties hinders the likelihood of identifying the integrative potential of the MPCS and the existence of positive interdependencies between stakeholders, leading to a narrow problem formulation and a suboptimal decision [3,6].

Both high- and low-power stakeholders can have important systemic contributions. On the positive side, as SDT [12] and AIM [14] propose, powerful stakeholders take initiative in goal-setting behaviors in MPCSs [4,6,8,39,48], which could play a beneficial role in the initial stages of the collaboration context. Powerless parties may foster collaborative effectiveness due to an in-depth exploration of the issue at hand. Curșeu and Schruijer [3] state that diversity (in this case, power diversity) is important for MPCSs as it fosters task conflict, an essential prerequisite of collaborative effectiveness. As Fleștea and the collaborators [8] show, those with lower levels of power are an important source of dissent and task conflict, making their input vital for MPCSs. By suppressing input from low-power parties, MPCSs runs the danger of reaching a suboptimal decision due to false consensus [3,61]. Thus, there is an equilibrium needed between the idealistic goal setting processes favored by powerful stakeholders and the realistic and diverse perspectives of the powerless.

In addition, if powerful stakeholders display resistance towards persuasion [45], they may ignore the relevant insights stemming from the systematic analyses performed by the less powerful parties. As they try to hold on to their powerful position [45], powerful stakeholders will not be open to change, which may hinder the effectiveness of the collaboration. The same limiting effect stems from the powerful stakeholders’ tendency to view others in a stereotypical and instrumental manner [18,46,47] because, as Lammers and collaborators [62] discuss, stereotyping by the powerful is characteristic of perceiving social independence between stakeholders. As MPCSs are contexts in which social positive interdependence is vital for performance, these tendencies hinder the possibility of reaching an integrative conclusion.

The persuasive nature of powerful stakeholders [14] may play out as a double-edged sword. On one hand, it may play an important role in motivating powerless stakeholders to become involved, aiming toward the goal of future empowerment [19]. On the other hand, the powerful may use persuasion as a tool for gaining resources and even more power [22], or to convince the less powerful parties that their interests are aligned with the ones of the powerful [51]. Thus, the mechanisms related to persuasion may either promote the expression of diversity, with positive systemic consequences, or the suppression of diversity, with negative consequences for collaborative effectiveness.

One systemic element with potentially positive consequences is the propensity of powerful stakeholders to assume responsibility for the outcomes of the collaboration setting [14]. Those with high levels of power will see themselves as responsible for the outcomes in the system, which will make them consider the climate as auspicious and become motivated in being involved in collaboration [9]. Oftentimes, the most powerful actors assume leading positions in the collaboration process [6]. Yet, establishing effective collaborative leadership in MPCSs is a highly complex endeavor, as stakeholders that assume such leading positions also have their own interests in the issue at hand [7]. A powerful actor that assumes a leading position in a multiparty system has to balance the neutrality needed for such a complex collaborative process with the proclivity to follow its own interests in the matter [63].

Thus, our review regarding power dynamics in MPCSs through the lens of SDT [12] and AIM [14] does not only show the effects of power and power discrepancies in such complex systems. More importantly, our endeavor shows that power itself and power discrepancies are not inherently advantageous or disadvantageous to certain parties or to the systemic results of MPCSs. The interaction between power distribution and attempts to uncover conditions of positive interdependence dictate the direction of these effects. A systemic identity and involvement in identifying positive interdependencies would lead to harvesting the benefits of power diversity. In contrast, an individualistic approach would lead to negative effects and likely suboptimal decisions.

A third contribution regards identifying areas proposed by SDT [12] and AIM [14] that have not been covered by studies up to this point, yet that could further explain the way power influences collaborative outcomes. On one hand, future studies should focus on disentangling the effects of power and power discrepancies at the emotional level, as emotions may influence the individual, parties, or the entire system [16,64]. Moreover, as MPCSs are multilayered, studies should focus on multilevel and cross-level dynamics in order to mirror the complexities of such interactions. In the same vein, future studies could focus on self-control, exploring whether the powerful exercise more self-control or whether they act impulsively from their power position with the intention to be instrumental in goal formulation and achievement. On the other hand, the mechanisms through which power influences collaborative outcomes have not yet been fully investigated. Even if SDT [12] and AIM [14] propose mechanisms related to abstraction levels, motivation, or cognitive resources, no empirical studies on MPCS have directly investigated these mechanisms. Some of the reviewed research on power in MPCS did not even make explicit reference to theories of power. From a practitioner’s perspective, we do understand the bottom-up, a-theoretical approach in studying the dynamics of power in such systems, yet we also see value in using the insights derived from scientific theories of power that received substantial empirical support in the literature. We call for future research that addresses directly the theoretical propositions derived from SDT [12] and AIM [14] in MPCS. Lastly, power is often a double-edged sword. In this case, future studies should focus on uncovering moderating variables that may modulate the effect of power in MPCS. We pointed out the perceptions of positive interdependence as a likely contingency for how power differences impact the collaborative effectiveness of these systems.

Fourth, there are some concepts that the proposed theories do not cover. On the one hand, multiple studies discuss the relevance of trust in MPCS [8,18,20]. Although trust can be discussed in relationship with emotional dynamics [8], it is also a factor with implications for coalition formation [18] and goal-attainment for low-power stakeholders [20]. Moreover, in newly formed MPCSs, stakeholders typically start from a context in which they distrust each other; therefore, distrust has to be tolerated and worked with until, eventually, parties manage to develop mutual trust [16]. As a consequence, future frameworks for multiparty collaboration should take into consideration the importance of working with distrust and building trust in MPCSs. Working with distrust and trying to build trust in MPCSs requires collective efforts and collective action, yet parties with varying degrees of power may approach the initial distrust differently. Differences in social distance between powerful and powerless parties may lead to unequal engagement in sorting out initial distrust. More powerful parties could, for example, tolerate higher levels of initial distrust as compared to powerless ones. Coupled with the negative emotional experiences tied to the lack of power, such a distancing of the powerful may be interpreted as withdrawal and disengagement by the powerless parties that will further breed distrust rather than reducing it. We see a rich area of future research that taps into the dynamics of power in dealing with initial distrust in MPCSs.

At the practical level, these results may be used as guidelines in establishing a sense of positive multiparty interdependence and interaction. Thus, setting the mindset that power is not immutable, that it actually flows, that it can be transferred and exchanged in the system may lead to more actions from those entering the collaboration with perceived low levels of power. This perspective may lead to more integration in the overall systemic outcomes, reflecting the interests of multiple parties invested. Thus, framing power from a relational perspective may lead to more involvement on the part of low-power stakeholders [8] and more willingness of high-power stakeholders to participate [5]. Moreover, raising awareness of how power impacts cognition, emotions, and behaviors in MPCS may help participants alleviate the negative consequence of power differences. Such awareness may increase the inclusiveness of those with low levels of power and will enable stakeholders to harvest benefits regarding diversity.

Moreover, as stakeholder diversity plays an important role in generating the necessary task conflict for collaboration [3] and power levels represent an important source for diversity, directly related to conflict [8], our results recommend an integrative selection process. Thus, a collaborative situation should involve stakeholders with different power levels, alongside a set of norms that would encourage open discussion and the expression of diverse points of view in order to discover or generate integrative potential in multiparty contexts [3].

In addition, as multiple studies highlight the importance of trust [8,18,20] at the onset of the collaboration process, practitioners should explore interventions aimed at building it. Oftentimes the initial interactions in MPCSs are marked by distrust rather than trust, and stakeholders have to work with their own and others’ distrust and ultimately develop trust [16]. From a systemic power dynamics perspective, trust interventions should be aimed at establishing that powerful stakeholders share common goals with low power stakeholders and empower them to engage with the system. Building trust should also focus on setting the idea that powerless stakeholders will not be instruments for those with high levels of power [47].

One last practical implication resides in highlighting the importance of social positive interdependence amongst stakeholders. While the situation whereby parties are brought together typically hints at positive interdependence, stakeholders tend to often frame it in negative terms [6]. Thus, proposing a goal structure that pinpoints positive interdependence will lead to the expression of diversity, which in turn will enable stakeholders to harvest the positive consequences of power disparity.

5.2. Limitations

This systematic review has some limitations as well. One limitation pertains to using the English language for the search and selection of the papers, which may have limited the number of studies included in the analysis. However, English is considered the global publication language [65], with the majority of papers being published in English. Second, our paper integrates the literature on power in MPCSs using SDT and AIM theories building on the idea that these psychological theories can shed light on how power differences in MPCSs impact individuals and the system as a whole. Other perspectives on power—for example, the systems–psychodynamic perspective (see for example [61,66,67,68])—could also provide useful frameworks for integrating the literature on power dynamics in MPCSs collaborative systems. Another limitation is related to the literature included in the theoretical review, as we were only oriented towards the so-called “white literature” (peer-reviewed papers) and we ignored the “grey literature” (unpublished literature or reports from practicians), which could have offered more descriptive insights in power dynamics in MPCSs [69]. Fourth, the studies included in the present review cover vast contexts and methodological approaches, from quantitative studies based on negotiation tasks [9,18], decision making tasks [8], or unstructured tasks [6] to real-life settings regarding refugees [46], internet governance [48,49], and environmental issues [4,45]. This heterogeneity of contexts may influence the applicability of the present results; thus, there may be contextual factors affecting the dynamics that we did not pinpoint in this review.

6. Conclusions

The present paper represents an effort to understand the manner in which power is conceptualized in MPCSs. We build on a theory-driven approach to integrate insights on power dynamics through theoretical propositions stemming from AIM [14] and SDT [12]. We have analyzed 18 papers and the results showed that the conceptualization of power in MPCSs evolved from a possession view towards a relational view, in line with the positive social interdependence underlying the dynamics of such complex systems. Regarding the propositions of AIM [14] and SDT [12], results showed that power does influences collaboration at multiple levels, cognitive, affective, behavioral and overall systemic, with power being a double-edged sword, that could hinder the interactions within the system or intertwine benefits tied to the cognitive and behavioral tendencies displayed by low and high-power stakeholders.