Linking the prepositions: using power analysis to inform strategies for social action

ABSTRACT This article reviews longstanding debates about the relationship between power over and power to – often posed as the tension between domination and emancipation. It then turns to several frameworks which integrate these approaches to inform strategies for social action. In particular, it focuses on recent empirical studies which apply one such framework, the ‘powercube’, to glean insights into how social actors navigate across multiple forms, spaces and levels of power. In so doing, we gain clues into how relatively powerless groups develop the capacities for agency and action which challenge domination and in turn give new possibilities for emancipation.


Introduction
Many years ago, as a young would-be activist confronted with the need to understand power relations in a desperately unequal and exploited rural mining Valley, 1 I asked the question that was to become the basis of my book, Power and Powerlessness: Quiescence and Rebellion in an Appalachian Valley (Gaventa 1982): 'Why in a situation of glaring inequality where one might intuitively expect upheaval, does one instead find, or appear to find, quiescence? Under what conditions and against what obstacles does rebellion begin to emerge?' Drawing upon the three-dimensional model of power developed by my supervisor, Steven Lukes (1974), I argued that over time, elites were able to shape decisions over resources to their advantage (first dimension), use their position to gradually keep key issues affecting their interests off the agenda (second dimension), and instil a sense of powerlessness, acceptance and quiescence in the face of an unjust status quo (third dimension).
Later, colleagues and I were to develop the work further in the framework now known as the 'powercube', 2 which suggested that Lukes' three dimensions of power were, in fact, only three aspects of a single spectrum of power. How the dimensions of power (relabelled as the visible, hidden and invisible forms of power) play out also vary according to the spaces (closed, invited and claimed), and the levels (household, local, national and global) of power. Moreover, the forms, spaces and levels of power constantly interact, in turn affecting the possibilities and strategies for change. (Gaventa 2006). of powerlessness, and seen as 'empowerment lite' (Cornwall 2018), or 'em-ment', (empowerment with the notion of power removed) (Christens 2019, p. 51) or power 'defanged' (Schutz 2019, p. 13).
Yet the separation of the idea of empowerment from ideas of power as domination has not always been the case. When first used in the United States in the 1960s, theories of empowerment were 'anchored in a philosophy that [gave] priority to the points of view held by oppressed peoples, enabling them not only to express themselves, but also to gain power and overcome the domination to which they were subject' (Calvès 2009, p. 4). In the early feminist thinking in the field of international development, writers also 'stressed that empowerment was a socio-political process, that the critical operating concept within empowerment was power, and that empowerment was about shifts in political, social, and economic power between and across both individuals and social groups' (Batliwala 2007a, p. 559). Building on this feminist tradition, Eyben et al. (2008) later added the idea of imagination: 'empowerment happens when individuals and organised groups are able to imagine their world differently and to realise that vision by changing the relations of power that have been keeping them in poverty ' (p. 6).
Despite these more radical roots, over time, the word empowerment gradually lost its association with the transformation of power, often driven by mainstream donor and institutional co-optation of the term (a process further described by Batliwala 2007b, Cornwall 2018, Schutz 2019. However, other traditions continue to insist on the need to link these two strands of power, or to bring them back together once again. For many feminist thinkers, the concepts remain inextricably intertwined (Batliwala 2007b, Nazneen et al. 2019. Within the field of community power, Christens (2019) argues that while recognising that empowerment 'is now often used ambiguously to indicate nearly any type of positive development amongst members of a group' (p. 13), other traditions have tended to define empowerment as the process through which groups can 'gain greater control over their circumstances' (p. 14), thus retaining an element of power within the concept. Moreover, he suggests that 'in addition to the cross-disciplinary disjointedness between scholars on these topics, there is a research-practice chasm that has limited the application of the research on community power and empowerment. There is therefore a great need to draw the parallel strands of research on community power and empowerment together, toward greater integration' (Christens 2019, p. 14).

The prepositions of power
One avenue for both linking the themes of power and empowerment, as well as bridging the research-practice divide, is found in how we bring together the multiple prepositions often used with the word power, including the often cited power over, with, to, within and (more recently) for. One fundamental debate on the meanings of power, for instance, has focused on the differences between power over and power to. In this sense, power over refers to the control that actors may have over others, growing from the formulations of the power of 'A over B', found in the work of Lukes and others before him. Other traditions have focused on the idea of power to, or the ability to act, a concept similar to empowerment, if understood as the process through which people gain control over their circumstances. Power over theorists often consider power as 'zero sum' and therefore conflictual -for one party to gain power, others must give it up -whereas power to theorists see power as potentially positive sum, or accumulative (see, for instance, Haugaard 2012).
While these are often seen as separate and competing definitions or processes, other scholars have argued that power over and power to need to be seen as interrelated. Haugaard suggests, for instance, that 'the same empirical processes, which Lukes theorises as three-dimensional power, have the potential to emancipate' (Haugaard 2012, p. 34). Similarly, Pansardi suggests that 'power to, just like power over, should be understood as consisting in social relations, and, moreover, that the social relations on which power to is necessarily based are specifically relations of power over' (Pansardi 2012). Kashwan et al. (2019) develop a 'power in institutions matrix' which links the two concepts. For them, the key difference between power over and power to, 'resides in whether power is exerted by the constraining of the opportunities and control that actors and agencies previously had (power over), or by creating new opportunities -i.e. new resources, structures and institutions -and relatively greater than individuals and groups of individuals enjoyed previously (power to) . . . Instead of associating "power over" exclusively with more powerful actors and "power to" exclusively with the subordinate actors, we present them as distinct components with the portfolio of power that all actors and agencies may simultaneously deploy in various combinations ' (p. 137).
In many of these debates, the prepositions over and to have been complemented by others as well. Writing about the politics of environmental transition, Partzsch argues for the importance of power with which she links to 'learning and cooperation', in addition to the concepts of 'power to (resistance and empowerment) and power over (coercion and manipulation)' (Partzsch 2017, p. 193). Power with, she suggests, is often linked to Hannah Arendt's (1970) definition of power, which 'always refers to a group or to a collective of individuals: Power corresponds to the human ability not just to act but to act in concert' (Arendt 1970cited Partzsch 2017'. Power with thus brings a more horizontal dimension to power, and points to the importance of social solidarities, alliances and coalitions (Allen 2000). For many feminists, just as important as power over, to and with is the notion of power within, through which actors develop awareness of their capacities for action. In her pioneering book Reversed Realities: Gender Hierarchies in Development Thought, Kabeer (1994) wrote that 'the multidimensional nature of power suggests that empowerment strategies for women must build on the power within [emphasis added] as a necessary adjunct to improving their ability to control resources, to determine agenda and make decisions' (cited Rowlands 1997, p. 21). Building on work by Freire (1997), power within often emphasises the importance of 'conscientisation', the process through which oppressed groups develop awareness not just of their own power within, but also a more critical understanding of the forces of that shape the power over their lives, and the possibilities of power to and with others.
While perhaps only recently coming together in academic literature on power, these ideas of power over, power to, power with and power within have long been used by gender activists and others in the development field (e.g. Rowlands 1997, VeneKlasen andMiller 2002). More recently, faced with a growing recognition of the way that the deeply engrained values, beliefs and ideologies shape the possibilities of action, Bradley and her colleagues at Just Associates have also added to the lexicon a new preposition -power for. Arising out of a concern to confront power over, with transformative power, which 'indicates the goal of fundamental change in power dynamics at all levels' (Bradley 2020, p. 107), power for 'refers to the combined vision, values, and demands that orient our work. It inspires strategies and alternatives that hold the seeds of the world we want to create. Power for provides a logic to transformative power -motivating the sustained movement-building efforts that generate power to, with and within as building blocks for change (ibid., p. 108).' Taken together, these concepts give us a basis for an approach to social action that begins to transcend the often separate streams of thinking and practice on power and empowerment, and which focuses not just on power as domination, but also power as challenge to that domination. In this more unified approach, these forms of power can be seen as interrelated, such that empowerment becomes a process through which relatively powerless groups develop a sense of power within, and the capacity for power with others, in order to challenge the power over their lives, and gain the power to determine their own futures, guided by their vision of a different world, as in power for. Though we can describe this process in a linear way, activists also know that these processes are often more iterative and messy, with setbacks along the way. Sometimes it is through challenging the power over in the first instance, that people, through that action, develop a sense of their power within and with others. Or the vision found in power for often emerges through the processes of gaining power within and being able to articulate a new worldview collectively. As people gain the power to act with others, they may face backlash, reprisals and repression, which can weaken the power within, or alternatively strengthen its resolve. As I have previously argued, without the building blocks in place of power within and power with others, the power to act and challenge the power over are likely to remain limited (Gaventa and Barrett 2012). But wherever one enters the cycle, this more holistic understanding of linking the multiple forms of power shows more promise for understanding how change happens than the focus on any one form of power on its own.

Applying power frameworks for action
If we are concerned with working across these prepositions of power to develop a more transformative vision of power, what strategies of action can be used? Fortunately, in recent years, a number of strategic frameworks for addressing power have emerged which attempt both to cross the academic and activist divide, and which also link the analysis of power to intentional organised efforts that challenge unjust power relations. While there are a number of these frameworks, here we are concerned in particular with those that have built upon Lukes' theories of the faces of power, as well as the more integrated approaches that link the forms of power 'across prepositions'.
One such framework for linking power analysis to social action is known as the 'power matrix', developed over the years by the activist group Just Associates. 4 First published in 2006 (Miller, et al), this is elaborated further by Bradley (2020, pp. 111-112) in a chapter in the book Power, Empowerment and Social Change (McGee and Pettit 2020). 5 One side of the Power Matrix features the different forms of power which arise from Lukes' three dimensions, which they articulate as 'mechanisms' of power, including 'visible' power (making and enforcing rules), 'hidden/shadow' power (setting the agenda) and invisible power (shaping meanings, values, and conceptions of what is considered 'normal'). Across the top of the matrix, the framework links each dimension of power to the prepositions of power discussed above, simplified to power over, challenging and resisting, and building and creating transformative power. Bradley's argument is that 'strategy in a movement context is by necessity concerned with efforts to both resist and challenge power over and to build transformative power. In fact, it is often the case that doing one well contributes to the other. ' (p.109). In this sense, struggles against power over are inextricably linked with struggles for power to.
In a somewhat similar 'matrix', Kashwan et al. (2019) develop a 'power in institutions' matrix (p. 138) that highlights a number of strategies that can be used to shift power, especially in the context of global neoliberalism. Arguing that much of the work on institutional change focuses too strongly on bargains amongst elites, influenced by what is often called the 'political settlements' framework (e.g. Kelsall 2016), they showcase 'efforts by less powerful groups to gain a foothold in decision-making processes' that go beyond more elite forms of policymaking (p. 133). Also, building on the concepts of power over and power to, they outline strategies in each for challenging the three dimensions of power, as developed by Lukes, which they call 'overt power' (first dimension), 'agenda power' (second dimension) and 'discursive/ideational' power (third dimension) (p. 138). They offer strategies by which less powerful actors may gain the power to, and influence the forces of power over. These include 'co-optation from below', which through which subjugated groups take advantage of spaces created from above to further their own interests; 'counter-power and resistance', involving collective action and mobilisation; and 'crafting institutions', through which citizens 'organise their own alternative bottom-up institutional arrangements of collective action (such as self-help associations) (p. 138).
In his short book, Empowerment: A Primer, Schutz (2019) also draws on these concepts to discuss strategies for empowerment and organising, largely in a US context. He argues that any strategy for empowerment needs to think about where and how it engages across a range of continua including types (moving from power to to power with to power over), spaces (moving from open to invited to closed) and forms (moving from visible to hidden to invisible). He then examines five strategies for empowerment, which can be used across these continua, including individual empowerment, collaborative empowerment, counterscript, 6 solidarity 7 and civil resistance, often found within social movements.
Though not focusing as much on the Lukes' framework of the dimensions of power, Fung gives a slightly different, though related, framework in his article 'Four Levels of Power: A Conception to Enable Liberation' (2020). While he argues that many scholars of power have 'sought to understand how power produces domination' (p. 131, italics in original), this article attempts to develop a 'mental map', to enable public leaders and activists to 'seek to alter patterns of asymmetric power in order to achieve liberation from domination' (ibid.). In particular, he points to four important levels 'at which to understand power and alter the terms of its exercise' 8 (p. 132). These include: • Everyday power, 'the interactions with power faced by individuals on a regular basis'; • Policy power, the 'general laws and policies from governments and other organisations) that make it more or less difficult for individuals to advance their interests'; • Structural power, 'the rules of engagement -the parameters and terrain -that govern contests between groups and organisations that advocate for individuals at the first levels and seek to shape the covering laws and policies that constitute the second level'; • Ethical power, 'the content and distribution of ideals, values, public narratives and norms in society'.
Drawing a number of lessons from this analysis, Fung ultimately looks for a more integrated approach, concluding that 'the most stable and powerful changes come from alignment at all four levels of power. There is no Archimedean level of power from which to make stable and lasting change. Victory or domination at one level can be limited or overturned by dynamics at the other levels' (p. 157). Taken together, these four relatively recent frameworks (Bradley 2020, Kashwan et al. 2019, Schutz 2019, Fung 2020) provide a range of strategies through which activists may challenge power relations, from forms of mobilisation and resistance to challenge power over, to strategies of creating alternatives, as power to, to ways of working with others and collaboration as power with. Strategies of change in each need to address and transcend manifestations of power in everyday issues, to the more policy-oriented levels, to its structural and the discursive forms. In so doing, the distinctions of power over and to, as well as with, within and for give way to more integrated and iterative approaches.
But what do we know about how these strategies work in practice? Each of these frameworks mentions or bears some resemblance to another framework which I and colleagues have developed over the years, known as the 'powercube'. 9 This framework has now been applied and used dozens of times both for the study of power as well as for informing strategy and social action. These applications thus provide an empirical lens through which to understand how strategies for action work in practice across settings and issues. By looking more closely at the evidence from these applications, linked to the frameworks above, we can gain further insights into how to build transformative strategies for change.

Empirical applications of the powercube: what do they tell us about how to challenge power relations?
In previous publications, I have outlined the emergence of the 'powercube' framework, and described its concepts in depth (Gaventa 2005(Gaventa , 2006(Gaventa , 2007. 10 To recount briefly here, the framework builds on my earlier work Power and Powerlessness (1982), which had been heavily influenced by my then supervisor, Steven Lukes and his classic book, Power: A Radical View (1974). In this book, Lukes argues that power can be seen in three dimensions, from the more visible to its hidden and invisible forms. Over the years, as I continued to work in this field, I began to realise that these forms of power are a continuum but reflect only one dimension of power (Gaventa 2006) and that a more complex approach was needed.
Influenced by the work of other colleagues at the Institute of Development Studies (e.g. Cornwall 2002, Hunjan andPetitt 2011), we began to realise that power manifests itself differently in different spaces, which ranged from the closed, to invited to claimed, and that power was relative -those who were perceived by themselves or others as powerless in one space might be seen as more powerful in other spaces. Concerned also about how in an increasingly globalised world power becomes disconnected from territory, we were clear that it was also critical to move beyond the realm of 'community power', to examine the dynamics of power and citizen action across levels, from the household, to the local, to the national to the more global (Gaventa 2007).
Building on the dynamic metaphor of the Rubik's cube, the powercube approach suggests that we must examine these three aspects of power (forms, spaces, levels) not only separately, but in their interactions. Moreover, each dimension of the cube also reflects a spectrum of possibilities which interact with one another, opening and closing the entry points for influence and change. Power strategies which only focus on one element, or one dimension, often simply reproduce or strengthen power in another. Really transformative change occurs, the work argues, when social actors (e.g., movements, civil society organisations, donors) work across all aspects of the cube, necessitating the emergence of coalitions and networks of actors, which themselves are affected by power dynamics.
Over the years, the powercube model has been picked up and used widely by a broad array of academics and activists, including international NGOs, local NGOs, social movements, think tanks, universities and donors, and others -in ways far surpassing our expectations. Many of these applications analyse power in relationship to participation, policy and governance issues (the areas from which the first applications emerged). But the powercube has proven itself to be highly versatile and relevant to other fields and issues as well, including digital inclusion, economic justice, environmental issues, trade (including fair trade), health, housing, humanitarian relief, human rights, hunger and nutrition, legal empowerment, mental health, peacebuilding, water and other natural resources. 11 These extensive empirical applications now give us the chance to explore the original propositions about the powercube, and how change might happen across its dimensions.

The forms of power
In 1961, Robert Dahl famously asked about New Haven, Connecticut, 'Who Governs'? His book was one of the best known in a genre of work on community power in the US at the time, launching an extensive debate on who had power, and indeed how one understood power in American democracy. Critiquing the argument that power could be studied by observing who prevailed in decision-making arenas, Lukes (1974) argued that power must be understood not only in terms of who participates, but also in terms of who does not. Power he argued, had three faces -the public face which Dahl and others had studied, a hidden face, which served to keep issues off the agenda of decision-making arenas (Bachrach and Baratz 1962), and an even more 'insidious' third face, through which the relatively powerlessness came to internalise and accept their own condition, and thus might not be aware of or act upon their interests in any observable way. Later, these began to be referred to as simply the 'visible', 'hidden' and 'invisible' forms of power (VeneKlasen and Miller 2002). Put simply 12 : • Visible power -focuses on who participates and predominates in observable decision-making. Contests over interests are assumed to be visible in political institutions and policymaking processes, which in turn are presumed to be relatively open.
• Hidden power -focuses on how certain issues and voices are kept out of the decision-making process through a prevailing 'mobilisation of bias' (Schattshneider 1960) or rules of the game which favours certain interests over others. • Invisible power -focuses on how the internalisation of ideologies, norms and values keeps issues and contests from emerging, and leads to the acceptance of an unjust status quo.
Very usefully elaborating on each of these concepts, Hathaway (2016, p. 118) reframes these slightly by focusing on 'how actors can realise their interests through decisionmaking and the control of resources (visible power); backdoor machinations and institutional organisation (hidden); and the structural-discursive empowerment of the actor and the creation and use of discourse (invisible)'(2016, p. 118). While a host of studies have now applied these concepts to the study of power more generally, for the purpose of this article, we are most interested in the implication of these forms for strategies of social action that attempt to challenge power relations based on domination. For instance, as suggested in the widely applied Power Matrix, developed by Just Associates (e.g. Miller 2006, Bradley 2020, if we are concerned with the visible forms of power, then the emphasis for action may be on policy and rulemaking institutions, through advocacy, as well as mobilising for accountability, building alliances with key policymakers, and focusing on institutional reform. If the focus is on hidden power, then strategies for action involve mobilisation to bring voices and issues into the public arena, as well as to protect collective actors from backlash for doing so. The focus may also be about building collective power through the media, grassroots organising and movements that challenge what Fung (2020) would call not only 'policy power', but also the 'structural power' around it. And if the concern is with 'invisible' power, strategies for change involve interrogating and disrupting social norms, beliefs and traditions that legitimate an unjust status quo, as well as building new narratives and critical consciousness, in what Fung might refer to as the 'discursive' levels of power.
In practice, very few organisations or movements have the skills or capacity to address each of these forms of power, at least not simultaneously. Often for funding reasons, social change groups may be encouraged to specialise, so that some will focus more on winning the issues through public advocacy, others will mobilise from below to challenge the rule-making process itself, and yet others will do the long hard work of changing norms and values. The effect of this separation, especially for those who take a more robust view of power, is to lead to ineffective forms of challenging dominant decisionmaking processes. VeneKlasen (2020, p. 24), for instance, writes of 'de-politicised advocacy', in which efforts become 'focused more on specific issues and advocacy (visible power) "on behalf of" but disconnected from organised voices and the demands of constituencies directly impacted'. Fung (2020) similarly argues that second order power (policy) cannot easily be changed without also shifting third order power (structural). Such arguments are also echoed in Kashwan et al. (2019), who suggest that a policy focus can end up largely focusing on political settlements amongst elites, often in highly technocratic and instrumental terms, rather than broader forces of domination which may shape these processes.
Various empirical studies of the interactions of visible, hidden and invisible forms of power seem to support this warning about the risks of not addressing these interrelationships. For instance, one of the most thorough applications of the powercube framework, Human Rights, Power and Civic Action (Andreassen and Crawford 2013), focuses on how structures and relations of power shape human rights advocacy, as well as the extent to which civic action has been able to challenge, alter or transform such power. The authors compare a number of human rights struggles led by non-governmental actors in Cambodia, China, Ghana, Kenya, South Africa and Zimbabwe. Analysing the manifestations of each form of power (visible, hidden, invisible) as they are played out in these struggles, the authors offer numerous rich examples of how various forms of power constrain struggles for human rights, across contexts.
Moreover, they argue that 'human rights advocacy was found to be constrained by all three dimensions of power' (p. 227), and that each form of power is 'nested' within other forms of power, reinforcing one another. For instance, in efforts to change land policy in Kenya, the visible power regime 'was also nested inside clientelist networks of hidden power' (p. 220), or in battles over the passage of a domestic violence bill in Ghana, the visible power of the minister was nested within his own internalized attitudes and behaviours, representing more invisible power. In order to challenge these 'nested' forms of power, strategies of change had to be directed across all forms -visible and hidden power, but also 'the insidious and invisible forms of power that are inherent in hierarchal structures and encourage deference and passivity to power actors' (p. 241).
A similar argument is made by McGee (2020) who focuses less on struggles to change policies, and more on efforts to hold decision-makers to account for their actions. In recent years, efforts for transparency and accountability have gained traction around the world as strategies for reducing corruption, increasing the responsiveness of powerful actors, or indeed of deepening forms of democratic engagement (Gaventa and McGee 2013). McGee insists, however, that a power analysis leads us to distinguish between 'tactical accountability' and more 'strategic' forms of accountability. Tactical accountability sees the problem of lack of accountability largely in Lukes' first-dimensional terms. If there is power over, it is largely that service providers have acquired too much unaccountable power over service users and constituents.
Within this tactical approach, the remedy to this power asymmetry is seen largely in terms of informational terms -if citizens simply receive, through rules and processes of transparency, more information, then they can mobilise to hold power to account. Such an approach, she argues, fails to understand the 'hidden' forms of power which limit or capture ways of mobilising and organising around information, and even further, fails to recognise forms of invisible power such as 'marginalised peoples' 'low expectations of their entitlements, or their apparent passivity or apathy in the face of perpetual service deficits' (McGee 2020, p. 61). New, more robust strategies of accountability claiming are needed.
Her article goes on to give the example of the Workers' and Peasants' Empowerment Organisation Masdoor Kisan Shakti Sangathan (MKSS) in India, which adopted a much more 'structural' and 'agential' approach, addressing each of the forms of power, and organising public fora which exposed corruption. MKSS mobilised and built countervailing power from below, challenging over time the norms and behaviours which had historically 'normalised' the status quo.
In these two examples of struggles for human rights and for accountability, the more transformative and successful strategies for change seemed to be the ones that saw the forms of power not as separate and distinct, but as 'nested', one within the other. These two studies are re-enforced by a number of others, which make similar points, especially about the importance of change strategies which address the role of norms and beliefs as forms of social power which turn in re-enforce the workings of hidden and invisible power (e.g. McCollum et al. 2018). An important corollary of this, however, is not to reject the importance of the policy and law-making arena, but also to recognise that policy reforms and policy protection must be accompanied by changes in other forms of power, if they are to be sustained.

The spaces of power
One of the innovations of the powercube is that it adds to forms of power an assessment of power across different spaces for action. As discussed and defined in earlier work, spaces are seen as 'opportunities, moments and channels where citizens as social actors can potentially challenge and change policies, discourses, decisions and relationships which affect their lives and interests' (Gaventa 2007, p. 213). While these can be expressed along a continuum, they include 13 : • Closed spaces, where decisions are made behind closed doors, without any pretence of broadening the boundaries for inclusion.
• Invited spaces, where people are invited to participate in public arenas but within set boundaries. Invited spaces may be regularised; that is, they are institutionalised, ongoing or more transient, through one-off forms of consultation (Cornwall 2002). • Claimed/created spaces, where less powerful actors claim or create their own spaces, where they can shape their own agenda or express their own voices more autonomously. These spaces range from ones created by social movements and community associations, to those simply involving natural places where people gather to debate, discuss and resist, outside of the institutionalised policy arenas.
Similar to levels, working in different spaces may require differing strategies and entry points for change. If the concern is how to open up closed spaces, the focus may be on movements for greater transparency, or 'openness' or the right to information. If the concern is with how to engage effectively in 'invited spaces', then strategies may focus on evidence and research for persuasion, or greater negotiation or deliberation skills. Or, if the concern is building claimed or created spaces, then the strategies may shift to those of resistance, non-violent or otherwise, movement building, or more unruly forms of disruption and change, including cultural expressions, or what Schutz (2019) refers to as 'counterscript' strategies. While much literature exists on each of these strategies, in practice, differing groups tend to need to focus on one or the other, often with some tensions across actors around which approach is the most appropriate. Again, a number of empirical studies now help us to understand further how groups may employ these differing strategies. For instance, the work by Andreassen and Crawford (2013) on human rights advocacy also analyses these strategies in relationship to closed, invited, and claimed or created spaces. Some of the groups they studied used more collaborative strategies, to engage in invited spaces or to try and open up closed spaces. Others built networks and alliances 'in preparation for a more confrontational strategy with powerful actors' (p. 233). Others used various tactics to lever open closed spaces, including occupying meetings, or of claiming and creating new spaces for action. As they observe, 'different combinations of strategies and participation in spaces are possible ' (p. 233).
Empirical studies which apply the spaces framework point to the dynamism of these spaces, their multiple tactical uses, and the importance of working across the spaces over time. Several key lessons emerge: Spaces for change open and change over time. For those promoting participation and democracy, recent years have taught us very quickly that the nature of what appears to be a democratic space can change very quickly (Hossain et al. 2018). While for several decades, we saw a gradual expansion and deepening of democracy in many contexts, more recently, we have seen very quick reversals, and closing of civic spaces through legislation, direct intimidation and violence, or attacks on the legitimacy of civil society actors -processes that have often intensified in the context of the Covid-19 pandemic.
The dynamism of opening and closing space can be seen within communities as well. In a PhD thesis focusing on community struggles in Belfast around the construction, and ultimately demolition of the Divis Flats complex, which threatened one of the oldest historic ethno/religious working-class areas in central Belfast, Webb (2016) shows that even in the same place, over a twenty-year period, the campaign went through distinct phases, each of which changed the nature of the spaces which the community mobilisers could use to continue their struggle. By shifting tactics over time, as the spaces changed, Webb argues that political and social ruptures provided 'strategic opportunities for community action' (p. 269).
Spaces can be used for multiple purposes, and action in one can leverage power in another. In her earlier important work, Cornwall (2002) warns that every space is itself filled and affected by the forms of power within it. Invited spaces, in particular, may give the appearance of greater voice and engagement, but forms of hidden and invisible power may simply mean that even in these spaces, voices become echoes of what powerful actors want to hear. While these warnings are important, several studies point to how actors can use spaces for a number of purposes, and can themselves co-opt the spaces created by the powerful for their own ends, similar to what Kashwan et al. (2019, p. 138) discussed earlier, when they refer to strategies of 'co-optation from below.' For instance, in their work on ownership dynamics in multi-stakeholder initiatives, Biekart and Fowler (2018) argue for understanding a non-linear view of how the dynamics of spaces might evolve: 'In particular, this may occur if what starts as spaces of invitation by the more powerful -typically a government agency with coercive authority -transform into spaces that are co-created by other stakeholders, suggesting that a broader "societal" ownership is taking hold. Conversely, governments can join created spaces and then proceed to take them over by, for example, starting to exclude stakeholders they do not like ' (p. 6).
In other interesting work, Discetti et al. (Accepted/In press) apply the powercube framework to the Fair Trade Towns (FTTs) movement. They conceptualise FTTs as 'campaign spaces' that are claimed 'through a synergy between local grassroots effort and NGO organisational capacity' (p. 9). While they recognise (as did Cornwall) that over time, these spaces, though originally shaped as democratic spaces, can become less participatory, on the other hand, they give examples of how campaigners have also entered invited spaces and then used them to expand the messages of their campaign, as well as to link up to new allies across a broader environmental and trade justice movement. They observe that it is the capacity to create the boundaries of the space that is critical: 'What is relevant to these dynamics is the capacity of campaigners to shape the boundaries of campaign spaces starting from the local level . . . decisions over boundaries -who is allowed to speak and participate in a determined space -are crucial to the democratic permeability of the space' 14 (ibid., p. 10).
Another example of the ways in which spaces evolve is found in the Guatemalan mesas de concertación -fora for consultation and follow-up in the peace process (Idler et al. 2015). While the 'mesas' originally were established as claimed spaces to enable local groups to have a voice in the process, these were then seized upon by the UN and other national actors who sought to formalise them as spaces for consultation and dialogue across multiple actors. The authors argue that while the mesas reproduced some of the 'structural and cultural violence at the root of the armed conflict and were used by various actors to pursue their own agendas (p. 6)', they nevertheless succeeded in 'raising awareness about the significance that support across different levels of society has for the successful implementation of the Guatemalan peace agreements' (p. 1).
Thus, while in the original notion of the powercube, invited spaces created by the powerful could be seen as a way of retaining power over, studies of various campaigns and struggles reveal the capacity of relatively powerless groups to use their agency to subvert these original purposes, often with a great deal of sophistication and intentionality. In work with activists in Colombia and Guatemala on what strategies could be used by social actors once they got into an 'invited space', Pearce and Vela (2005) came up with over 20 reasons that activists could give for engaging within them, including, for instance, decision-making, scrutiny, negotiation, and as a form of protest within the space. Or as Idler et al. (2015) found, the powercube approach helped to 'grasp the manifold expressions of power at play in a local peace initiative, not only power as domination but also power as empowerment, participation and resistance in the face of domination' (p. 38). 15 Working across spaces. Earlier research in the field of participatory governance has shown that simply creating new spaces for previously excluded groups to participate is not enough -such spaces are held open by the political will of champions inside the state (sometimes from within their closed spaces), and by effective mobilisation from the outside by citizens who insist on accountability and scrutiny of what goes on inside the invited space (from within their claimed spaces). In addition, these invited spaces may be more 'empowered' when they are well designed and facilitated, and backed by legal provisions (Cornwall and Coelho 2007). Arguably, a good example of what happens to such spaces in the absence of these conditions is found in the well-known case of participatory budgeting in Porto Alegre, Brazil. At its best, this space had support from the ruling parties (at both state and national level), had grown out of well-organised and mobilised civil society in the city, and had developed a clear and well-designed process for engagement. However, 'after the Workers' Party was defeated in 2004, a politically conservative coalition took power . . . [and] maintained superficial features of PB [participatory budgeting] while returning the actual functioning of the administration to more traditional modes of patronage and privileging of local elites' (Baiocchi and Ganuza 2016). Moreover, as much of civil society attention had moved to engagement in the invited space, it had lost the base outside of the invited space to keep pressure on the process.
Spaces in authoritarian settings. Finally, it is worth noting that while one might assume that in more authoritarian contexts, we would find more 'closed' spaces, and in more democratic settings, more open and invited ones, studies show that there may be a multitude of types of spaces across these regime types, depending in part on the issue at hand or the level of entry. In one study on how NGOs navigate power dynamics in China, Lay Lee (2012) shows how NGOs were able to claim or create spaces themselves, and once they did so, this could lead to recognition by the state, and to 'invitations to enter closed spaces' (p. 347). Moreover, by working across levels of power, NGOs were able to gain legitimacy at the global level, and use this to influence national and local levels. Thus, much of what happens in a given space may also be affected by the forms of power which surround it, as well as the dynamics and pressures from other levels above or beneath it. This takes us to the final dimension of the powercube, that of the 'levels of power'.

The levels of power
While much of the earlier work on power focused on 'community power', a large literature on global governance warns us of the dangers of focusing only on the 'local', or the 'national' in a globalising world, arguing that governance has become 'multilayered', ranging from the subnational, to the national to the supranational (e.g. Keohane 2002, Gaventa 2007. The powercube asks us to look at the relations of power both within and across these levels along a continuum, including 16 • Global -formal and informal sites of decision-making beyond the nation state; • National -governments, parliaments, political parties or other forms of authority linked to nation-states; • Local -subnational governments, councils and associations at the local level; • Household -the micro-level, which may be outside of the public sphere but which helps to shape what occurs within it. Just as in the forms and spaces of power, a number of studies also look at how power works across levels. In particular, in a global world, studies point to how building movements and campaigns across levels can be a massive challenge (Gaventa and Tandon 2010). However, a failure to work across levels of power can serve to pervert or limit outcomes of reform initiatives. For instance, Brugger (2017) uses the powercube to examine power relations in the global production network for Fairtrade Himalayan Tea. While this process is supposed to be one in which workers' rights can be strengthened and protected by a certification process and by enabling consumers to use their purchasing power to also promote 'fairer' trade, this analysis sheds light on the 'black box' that exists between consumers and producers of Fairtrade Orthodox Himalayan Tea. The certification process is found to be a buyer-driven and top-down process which recreates colonial dependencies in production and trade. Bargaining processes over the welfare for workers and small farmers often take place in closed spaces at a supra-local level, while taking away power for the workers and small farmers to demand adequate welfare on local levels and through created spaces such as mobilisations through labour unions or political ventures. Another study also shows how the absence of local claimed spaces can lead to topdown policy processes, in which local groups have neither presence nor voice. Using the powercube to study the making of nutrition policy in Zambia, Harris (2019) finds that nutrition policy priorities were 'shaped by a global epistemic community relying on the hidden and invisible power of technical language and knowledge to frame policy options which resonated with their own beliefs about malnutrition . . . Striking in their absence from either invited or claimed spaces of power were the malnourished themselves, or their communities or representatives, who did not have a clear voice in Zambia's nutrition policy process and therefore were without power' (p. 121).
This example in Zambia, however, can be contrasted with another study which focuses on food security and children's rights in Canada (Blay-Palmer 2016). Also taking a powercube approach to analyse the government's approach to food security, the study illustrates how activists were able to work across levels, in particular leveraging global human rights agreements to gain more voice at the national and local levels. While 'the government used both visible and hidden power to privilege discourse around free markers and downplay the need to address social justice issues' (p. 9), as Canada was also a signatory to the UN Convention on the Rights of the Child (1991) as well as the International Covenant on Social, Economic and Cultural Rights (1976), local civil society organisations were able to mobilise to demand their rights through 'new spaces' and organisations which insisted on a social justice agenda. 'Using power across multiple scales allowed for the improvement of access to more healthy food for more children and opened up spaces for access to local healthy food at the provincial scale' (pp. [9][10]. These three examples -fair trade in the Himalayas, nutrition in Zambia and food rights in Canada -each have differing outcomes. A critical difference was the ability of activists to work across levels, to hold or create spaces locally or nationally. The challenges of doing so are elaborated in a further study of ten different civil society campaigns or initiatives that attempted to link citizens' voices from local to national to global (Gaventa and Tandon 2010). In only some of these, were these linkages successful. Much depends, the cases find, on the role of effective intermediary actors, institutions and processes that help to achieve vertical links across scales of power. For instance, in one study of a relatively successful campaign that worked across these levels, the Global Campaign for Education for All, deliberate attempts were made to a) build on local and national experiences and organisations in the global work; b) develop representative governance across all levels; c) pay attention to inclusive framing of the issues from local to global; d) always acknowledge the contribution of actors at each level, and e) attempt to allocate resources to lessen competition amongst organisations at each level (Gaventa and Mayo 2010).

Conclusion
In my earlier book on power (Gaventa 1982), I argued that 'dimensions of power and powerlessness may be viewed as accumulative such that each dimension serves to re-enforce the strength of the other'. As powerful actors develop their power -gradually asserting control over decisions, agendas and beliefs -each dimension can be used to support the other: 'The power of A is also strengthened by the fact that the powerlessness of B is similarly accumulative, and that power and powerlessness may each re-enforce the other towards the generation of B's quiescence ' (1982, p. 22). However, I observed at the time that the reverse may also be true: 'Just as the dimensions of power are accumulative and re-enforcing for the maintenance of quiescence, so, too, does the emergence of challenge in one area of a power relationship weaken the power of the total to withstand challenges by more than the loss of a single component . . . Once patterns of quiescence are broken upon one set of grievances, the accumulating resources of challengeorganisations, momentum, consciousness -may become transferable to other issues and targets ' (pp. 24-25).
Inherent in this process is the constant interrelationship and tension of the push to accumulate more power, or power over and to resist or challenge the boundaries of that power, power to. Rather than being seen as separate forces, as much literature in recent years has tried to do, this essay has argued that these processes are inextricably intertwined. 17 Moreover, while much of the work on power, including my own work, focuses on the repressive effects of power over, we have found that even in cases of extreme inequalities of power, dominated groups have found ways to exert their agency, pushing back and constantly challenging such power. In so doing, they use multiple strategies -resistance from the outside through claiming their own spaces, engagement within invited spaces, challenging dominant discourses and articulating new prefigurative possibilities for change, and more, each of which may be re-enforcing the other. These strategies and practices of social action work across and beyond the 'prepositions' of power which have been so visible in power debates in recent years. In various combinations, engagement strategies involve and support the development of the power within, through which disenfranchised groups recognise their own agency, the power for, through which they develop their own visions and imagination of how circumstances could be different, and the power with others through alliances and solidarities -all as part of using their power to challenge the power over their lives and circumstances. Transformative change involves working with all of these strategies, and with all of these forms of power.
The powercube framework implicitly builds on this understanding of the iterative and accumulating processes of power. In this article, we have examined how this process happens more deeply in practice through reviewing a number of empirical applications of the powercube framework. The review of these empirical applications offers several key insights, some of which re-enforce earlier arguments about the powercube and its utility. Several key themes emerge.
First, the examples re-enforce the point that while each dimension of the cube may be analytically distinct, in practice they are 'nested' within one another, as described in particular in the study of human rights struggles by Andreassen and Crawford (2013). Visible power contains within it the norms and beliefs of invisible power; the boundaries and possibilities of invited spaces are shaped by the other spaces which surround them, global levels of power are experienced locally, and so forth. Not only are the elements of each dimension nested within one another, in addition, each of the powercube's dimensions interact with one another to open and close possibilities for change. Invited spaces for change which may look like new openings for engagement may in fact be filled with the norms and assumptions of hidden and invisible power. National and local policymaking spaces may appear independent, but in fact their agendas may be set by powerful forces at other levels, and so on.
While the 'nesting' and interacting of each element and dimension of the cube may serve to strengthen the power of dominant groups, as in the argument above, they also create the opportunities and cracks through which less powerful groups may gain foothold for change. As Scott (1990) has argued, systems across levels, spaces and forms are rarely so hegemonic that there are not possibilities for resistance, even though such resistance may be less visible to the power-holders. We have seen multiple examples in the studies reviewed in this essay -the experience of participating in invited spaces can be used to build awareness of power, breaking down invisible power, or to build alliances with others (Brosnan 2012, Biekart andFowler 2018). Or as we saw in the case of NGOs in China, while controlled at national levels, spaces for engagement may be opened at local levels, in order to maintain the same national power (Lay Lee 2012). A powercube analysis may help to find the 'cracks' in the system that can be used to reverse the accumulative effects of power over, and to strengthen the possibilities of power to.
In that process, in order to build more sustained and transformative 'countervailing' power (Andreassen and Crawford 2013), strategies for change need to work across all the elements and dimensions of the cube, e.g. challenging closed spaces while entering invited spaces and claiming autonomous spaces, building links from local to global, engaging with visible power, while challenging hidden and invisible power. When this happens, groups may also be able to work across the levels of power suggested by Fung, linking the struggles around everyday policy, structural and ethical power (2020). The obstacle often becomes that it is very difficult for any campaign, movement or collective action to work across all of these spaces, forms and levels alone, or certainly not at once.
Critical therefore is the process of how to build alliances and coalitions which work across these dimensions in concert, recognising that such alliances are often themselves filled with power divisions and conflicts. This involves also understanding further the role of intermediaries -individuals, organisations, parties, processes -which can connect the positive forms of power across the dimensions of the cube. It is only through such connections that transformative change across the forms, levels and spaces of power will happen.
Over time, as scholars of power, we have seen an evolution of our understandingfrom recognizing the three dimensions of power, as articulated by Lukes in 1974, to linking these forms of power to the spaces of levels of power, as in the powercube (Gaventa 2006), to understanding how interrelated and nested these dimensions are (Andreassen and Crawford 2013). As our understanding of power changes over time, so do we learn that power itself is not static. Even as relations of power over are being challenged from below, so too the forms, levels and spaces through which these are manifest may also be changing. For instance, while with globalisation, there has been growing work in recent decades to understand power at the global level and how it affects the national and local levels of power, so too have global actors learned to recreate and strengthen their soft power in new ways (Gallarotti 2016). As we have learned about the need to understand the spatial dimensions of power, the Covid-19 pandemic, with its enforced lockdowns around the world, along with the rise of new forms of digital engagement, have challenged us to reconsider our received understandings of the nature of public and private spaces themselves (Chenoweth et al. 2020).
For activists, the dynamism of power points to the need for similar agility in strategies for collective action, recognising that these must constantly be re-evaluated in light of the new ways in which power re-shapes itself. And for scholars of power, this suggests that our own field must equally remain dynamic. In this sense, perhaps the most important form of power (as observed by researchers using the powercube with peasant organisations in Guatemala), is 'staying power', i.e., the ability 'to move across spaces of engagement over time, to retain links with groups working with other spaces, and to have the different capacities for engagement demanded by different spaces in differing moments' (Gaventa 2005, p. 20). In that process, in a highly unequal and constantly changing world, the challenge before us is continuously to deepen, adapt and change both our understandings of power and our strategies to challenge unjust power relations as well.

Notes
14. Though to my knowledge, they do not make a direct link, the point about boundaries reinforces the work by Hayward who 'suggests that we might understand power 'as the network of social boundaries that delimit fields of possible action'. Freedom, on the other hand, 'is the capacity to participate effectively in shaping the social limits that define what is possible' (Hayward 1988, cited Gaventa 2007. In this sense, participation is not only the right to participate effectively in a given space, but the right to define and shape the boundaries of that space. 15. Brosnan (2012) makes a similar argument in her work with user committees in the field of mental health, arguing that through applying the powercube service users could develop greater awareness of hidden and invisible forms of power, and their potential to influence decision makers. 16. Adapted from www.powercube.net. 17. As Haugaard (2012) has argued: 'The fact that normatively desirable power and domination are constituted through the same processes is not chance: the effectiveness of power as domination is parasitic upon power as emancipation ' (p. 33). See also the empirical study by Andreassen and Crawford who make a similar argument in their study of human rights advocacy (2013, p. 240).