Вы находитесь на странице: 1из 17

Journal of Power

ISSN: 1754-0291 (Print) 1754-0305 (Online) Journal homepage: http://www.tandfonline.com/loi/rpow20

Agency and structure: Interpreting power


Keith Dowding

To cite this article: Keith Dowding (2008) Agency and structure: Interpreting power relationships,
Journal of Power, 1:1, 21-36, DOI: 10.1080/17540290801943380

To link to this article: https://doi.org/10.1080/17540290801943380

Published online: 26 Mar 2008.

Submit your article to this journal

Article views: 1684

Citing articles: 10 View citing articles

Full Terms & Conditions of access and use can be found at

Journal of Power
Vol. 1, No. 1, April 2008, 21–36

Agency and structure: Interpreting power relationships

Keith Dowding*

Australian National University

Many of the debates over the ‘true’ nature of power relationships concern the importance given
2008 (online)

to agents or to structures in describing those relationships. Writers who discuss the power
structure or system tend to concentrate upon structures; those who write about the power of
agents or power relationships tend to concentrate upon actors. The agency-structure
relationship goes deep into many seemingly different issues in very different approaches to
power. I argue that whilst the agency-structure divide is false, our interpretation of the world
– the way in which we describe it – means we cannot fully transcend the structure-agency
divide using natural language. Whether we choose to use the language of structures or of
agents depends upon the questions we are seeking to answer and commitments we wish to
make in assigning responsibility. Those latter commitments do make power assignations
normatively driven, though that does not, in itself, entail that they cannot be objectively made.
Ultimately, structural and agential accounts can describe the world in non-contradictory ways
though the choice of description demonstrates the sorts of commitments the describer has
towards changing the world.
Keywords: agency; power; resources; structure; structuration; rational choice

Agents or structures
There are many different accounts of the nature of social or political power in society.1 One of
the deepest divides between those accounts is whether ‘power’ is predicated on agents or on struc-
tures. Within each segment of the divide reside other divisions, notably over what constitutes an
agent, and what is the nature of the structure of society. These divisions are deeply implicated
within the broader divide I have identified. Some accounts of agents are themselves deeply struc-
tural, and some accounts of structure implicate agents as the holders of power. In this section,
I will begin with examining how agents are defined with relation to power holding, then examine
different definitions of structure, and how these affect the nature of power holding.

According to the basic idea of methodological individualism in social theory, the only agents are
individuals where ‘individuals’ means biological human beings. Methodological individualists,
in the main, have been happy to state that whilst we often write of other agents – such as firms,
political parties, pressure groups, and so on – their actions can always be decomposed into the
actions of biological human agents. Thus, we might explain a political party adopting a specific
programme and election manifesto in order to appeal to the median voter and so try to win the
plurality vote and the election; the actual process of the political party adopting that programme
can be explained in terms of the specific actions of members of that party. Those members may
not all have adopted that programme in order to appeal to the median voter. Some members might

*Email: keith.dowding@anu.edu.au

ISSN 1754-0291 print/ISSN 1754-0305 online

© 2008 Taylor & Francis
DOI: 10.1080/17540290801943380
22 K. Dowding

have supported that programme because they thought it was the ‘right one’. Some members might
have supported that programme because they realized during the debates that it was the best they
could hope for despite in their view it not being the ideal policy. Others might have supported it
because they thought it would command most support in the country and they wanted victory (their
way of expressing the academic thought of ‘appealing to the median voter’); others still opposed
the programme and, at best, they will only give it lukewarm support during the election campaign.
In this manner, the methodological individualist will argue that we can explain the actions of
the political party as an agent, in terms of the actions of each of its members that led the party to
that programme.2 Part of that story will include not only the beliefs and desires of the members
by which we explain their actions, but also the rules and institutions of the party that ‘structurally
suggest’ what actions its members might take in producing that programme. That is the rules of
the organization will help determine the strategies of the agents. Of course, the rules will, struc-
ture the powers of different agents also. One party might be highly democratic and its programme
adopted following a series of votes involving all the members of the party; another might be
hierarchical and its programme adopted by the leader supported by a small coterie of advisers.
Nevertheless, the core idea of methodological individualism is that only agents act. Collective
agents can be decomposed into biological human beings (together with institutional rules) and
thus all social theory can be reduced to the actions of human agents.
The methodological individualist need not deny that there are other agents that are causally
important in social affairs. A prime minister might fall because of some natural disaster. Foot and
mouth is a viral disease affecting cloven-hoofed bovids, notably cattle and sheep. The virus, of
course, can be seen as an agent that was causally important in a prime minister’s fall. Without the
outbreak of the disease, he would have continued as prime minister for many years. However, for
the individualist, as far as social theory is concerned, foot and mouth is an agent exogenous to
social explanation. What is important for social explanation of the prime minister’s fall is his (and
his government’s) reaction to the disease (Boin et al. 2005). All sorts of natural disasters such as
AIDS, floods or earthquakes might have massive economic and political effects, but they are
exogenous to social explanation. In social explanation we explain the fall of governments due to
natural disaster not in terms of the agency of disaster itself but rather in terms of the state
responses to that disaster: state responses which can be decomposed into the actions of biological
human beings.3
Of course, we might have a completely structured answer as to why some prime minister
failed if there were no actions whatsoever that could have saved him. We might consider this to
be a form of determinism where events conspire to bring down a prime minister and there is
nothing that she, or any one else could have done to stop that fall. We note here that this form of
determinism is fully compatible with free will or autonomy in the sense that the agent can be in
full command of their faculties. Their beliefs and desires have not been determined for them.
However, given the events surrounding the agent, that agent is unable to stop the process that
leads to her fall. There was nothing she could do. Even this ‘fully structured’ explanation of the
fall of a prime minister is compatible with a strong form of individualist explanation, since the
fall is still the result of events caused by something exogenous to the social reality, but those
events still followed the actions of others in society. The ‘prime minister’s fall’ is an event that
gets its meaning in a social context (of prime ministers, of leaders, of failure) that requires social
action by agents. The prime minister’s fall might be due to the action of some virus (in cattle, or
in the prime minister herself – she resigns because of illness) but the social explanation is couched
in terms that are reducible to (sets of) individuals as agents.
The idea of ‘structural suggestion’ here is that given the biological individuals’ interests the
environment they face – both the social or institutional rules and the interests of other people –
will structure their behaviour. For social scientists, the methods of revealed preference theory
Journal of Power 23

should allow us rationally to reconstruct individuals’ interests from their behaviour given the stra-
tegic situation that we can model. What might differ from the commitments of most methodolog-
ical individualists from those of structuralists is the provenance of interests. Whereas
structuralists (as we see below) tend to assume that individuals’ interests are determined by the
structure, individualists tend to assume that individuals’ interests are not so determined – at best
they also are ‘structurally suggested’. We might assume that there are some base or fundamental
underlying interests – such as the desire for warmth, food, status and so on – common to all people
(and if caused then caused by our biology) and how we promote those interests will be suggested
to us by our place in society. Exogenous interests are those so suggested, endogenous ones, ones
we choose more independently, such as a preference for detective over science fiction stories for
example (Dunleavy 1991, Dowding 1991).
We note briefly here that structure is not completely absent even in the most individualist of
methodological individualist explanation because the structure of an organization will affect the
outcome no matter what the preferences of the actors are. That is, a party composed of the same
individuals might reach different policy programmes dependent upon the institutional rules
governing how policy is adopted. This will happen because both different preference aggregation
mechanisms produce different outcomes and because different mechanisms will affect the strat-
egies actors adopt in order to promote their favoured outcome. However, in terms of the power
structure, methodological individualists will predicate power to the individual agents. The power
of the political party (to win an election) will be a result of its strategies in relation to the strategies
of other parties. The power of its members with regard to policy formation within the party will
be a result of the strategies they can adopt to help promote the outcomes they want; and their
power to win office is based on their ability to adopt the best programme that furthers that aim.
A further manner in which mainstream ‘methodologically individualist’ explanation is struc-
tural is the fact that in most models of social and political processes biological humans are not the
agents. Rather agents are composed of social roles such as ‘leaders’, ‘bureaucrats’, ‘consumers’,
‘voters’ ‘middle class professionals’ and so on. As we see in the next subsection, one of the major
ways in which ‘structural’ explanations diverge from individualist ones is in the provenance of
individuals’ interests. Structuralists understand individual interests as being caused by social
structure, whereas individualists tend to see agents as reacting to structure in defining their inter-
ests. However, in most social explanation, agents are seen as social roles and those roles are
largely defined by structure in terms of the role of expectations and the interests of the agents
occupying those roles. Thus, in a second way virtually all individualist explanation has structure
deeply implicated within its very form.
Still we can maintain that individualists see power as a predicate of individuals and we can
mention many analyses of power that view power in this manner. Weber (1978) defines power in
terms of an agent’s ability to get what he wants despite resistance. Dahl (1957, 1968) sees power
as the ability of one agent to bring about changes in the activities of another agent. Oppenheim
(1981, ch. 2) likewise defines social power in terms of how one agent (person) affects the activities
of another agent (person). Such examples could be multiplied. The importance of individualism
in all these analyses is the assumption that only agents act and power is a causal notion that is
applied to the causing agent. Now structure, as we have seen, cannot be irrelevant to social expla-
nation and it cannot be causally irrelevant. Two political parties might adopt different policies,
not because the individual agents have different views over policy, but rather because of the insti-
tutional rules by which those views are aggregated. In addition, unlike natural disasters, such
mechanistic structuring cannot be considered exogenous to social explanation. After all, we might
well want to explain socially why each party adopted different institutional rules. However, therein
lies the rub. For the individualist, the explanation of the social rules will have an individualist
explanation, even as those rules affect specific outcomes today. For the individualist, structures
24 K. Dowding

are not irrelevant to social explanation, but ultimately, they will also have individualist explana-
tion. We return to this point, and its converse, repeatedly.

Most simply a structure is the relationship between variables. Take a set of red bricks and put
them together one way, and one gets a bungalow, place them another way and a two-storey house:
same bricks, but different structure. However, the two-storey house will also require a staircase
that will not be needed in the bungalow. Breezeblocks are a different type of brick, and could be
placed in relatively different relationships to build a bungalow or a two-storey house, though
these houses would not be identical to the redbrick ones. We might make the two houses look
identical, by perhaps putting red bricks as fascia around breezeblocks, and using plasterboard and
skimming inside to make both outside and inside identical. The cottages and houses might appear
identical with each other, but there will be qualitative differences, noted perhaps in heating bills
or during renovation.
Society could be viewed the same way, with people taking the role of the bricks and social
structure the form of the relationships between people. Thus genetically identical people might
be formed into different social structures – one hierarchical and one egalitarian for example. The
different social structures, which place genetically identical people into different roles vis-à-vis
one another, will form a different system of power. A structuralist would examine the relation-
ships between the variables ‘people’ and thereby describe the power structure. A structuralist here
would predicate the term ‘power’ to the structure for it is the structure that both constrains and
enables the options of agents. Those relational factors determine the outcomes that result. The
individualist would interpret those relational factors as properties of the agents. Whilst those prop-
erties are given to people by those different structural relationships the individualist sees them as
properties of individuals. Using the language I have used in the past, each agent’s resource would
be dependent upon the structure of relations, and those agent’s resources would depend upon the
way in which we could both identify and provide some measure of their power (Dowding 1991,
1996, 2003, see also Morriss 2002). We can see in this representation of the agent-structure divide
a commonality that distinguishes one from the other only by where the emphasis is laid. If we
concentrate upon the set of relationships that constrain and enable then we privilege structure; if
we concentrate upon how those relationships constrain or enable the agents to do what they want
– that is we concentrate attention upon individual agency – we examine the power of individuals.
The structural forces of one account constitute the full set of resources of the other account.
In Dowding (1991), I argued for a resource account that firmly puts ‘power’ into the agency
camp using a ‘redundancy’ and a ‘conceptual’ argument against making power a predicate of
structure. The redundancy argument claims that we can always remove ‘power’ from sentences
where it is used in terms of structural power and produce sentences with the same empirical impli-
cations. The use of the term ‘power’ here is redundant. I also claimed that conceptually the term
power is closely aligned to cause and thus to agency. Since agency is naturally aligned to agents,
the natural use of the term ‘power’ is predicated to agents and not to structure. At the same time
however, I believed my rational choice account of power crossed the structure-agency divide. As
many others have also claimed their account crosses the agent-structure divide (see below). I still
believe that ‘power’, at least in the English language, is more naturally used as a predicate of
agents than structure. However, my two arguments only really apply to this first cut at the agency-
structure relationship. When we look at a deeper account of structure, the two arguments lose
some of their bite.
The analogy of bricks and building with people and structures can only be taken so far. Whilst
bricks might be affected by the structure of relationships in which they are bound – loading
Journal of Power 25

factors would vary across the structure, perhaps causing some bricks to crack or fail – we would
expect human agents to be affected to a much greater extent. Moreover, people are affected in two
ways. One is ‘of the moment’. The situation in which they find themselves might affect their very
being. For example, a sportsman’s confidence might be affected by his opponent’s demeanour.
That confidence leads to greater effort and eventual victory. Here we might say one person’s
action is (a part of) another’s structure. Seeing the lack of confidence in one’s opponent brings
confidence and that creates a new person for oneself: a newly confident person. That, of course,
leads to the second way in which structure affects people; indeed, largely structure makes them
who they are. To continue the sportsman analogy: victory of our sportsman brings greater confi-
dence still and renewed effort in training and so making him into a champion. What can happen
is that a sequence of luck can turn one person into a winner with a winner’s temperament and turn
another into a ‘choker’ who fails at the last hurdle. Both people genuinely do have the qualities
with which they are labelled (‘winner’ and ‘choker’) but these have become part of their identity
because of events that occurred: events due to the situation in which they found themselves and
perhaps a smattering of chance. In other words, history makes the person. It does not make the
person alone, perhaps, but in the way agents themselves are structured by their environment and
that environment over time, history makes them. It is in this manner that structuralism goes deep
inside of agents. I will call this ‘deep structuralism’ and it is deep structuralism that forms the
strongest basis for my account of structural power.4
In most individualist analyses of power, an agent’s power is measured by what she could
achieve given the circumstances in which she finds herself. The individual might ‘bring some-
thing’ herself to the achievement – skills and talents provided by her genes say – but the even here
what a person’s genes can provide is determined by the environment.5 A breeze block brings
something different from a red brick to structure into which it is built – the breeze block might
withstand forces differently from the red brick, and might provide different thermal properties –
and in the same way genetically different people might respond differently to social structures.
For example, some people might behave more responsibly than others might in an egalitarian
society; some more than others might be more prone to abuse their positions within hierarchical
systems and so on. However, our expectation is that structural relationships have bigger effects
upon human agents than they do upon physical entities such as bricks. To a much greater extent,
human agents are constructed by their relationships to one another. For example, both the original
formulation of the authoritarian personality (Adorno et al. 1950) and its more recent version in
the work of Altemayer (1996) argues that submission to authority and the expectation of how to
dominate when in authority are given by how one has been treated within a household during
one’s early life. The argument here is that we are partially built by our environment, a view few
would dispute, though many would dispute the analysis of the authoritarian personality (for a
good review and analysis, see Martin 2001).
There are two ways then in which individual human agents and structure might work together
to explain power in a society. One way is to assume that the human agents are fixed in some
manner – we take their desires and beliefs (which together form their preferences) as given – and
then examine how they behave given the environment in which they operate. Formal results in
social and rational choice theory demonstrate that identical preferences lead to different
outcomes given different aggregation rules. It follows that we can expect to see different
outcomes from hierarchy than from bargained decision-making; different results from different
electoral systems and so on. Now it must be recognized, even at this first level, that examination
of these fixed preferences needs to take into account the structural or institutional relationships
that exist. Individual agents can be expected to respond to different institutional relationships by
the different strategies available to them, that is, they respond strategically. Any institutional
form can be manipulated in one way or another and different forms of manipulation will work
26 K. Dowding

best in different circumstances (Gibbard 1973; Satterthwaite 1975) and different tactics of
manipulation will be adopted given the possibilities open to both the powerful and the weak
(Goodin 1980). However, we can imagine here that people have underlying preferences, and
these together with the institutional form lead to their behaviour.
At this level of analysis, we should acknowledge that different structures have different effects
upon behaviour. Some institutions are constructed of expectations about ways of behaving.
Norms are constructed about how we should respond to one another in different social contexts.
Different societies and cultures generate different sets of expectations. In game-theoretic terms,
different equilibriums have been reached by agents playing essentially the same game and
complex games often have multiple equilibriums. Equilibrium strategies are maintained by ratio-
nal players since it is in each player’s best interest to maintain that equilibrium strategy. Some
equilibrium will be more stable than others will, since out-of-equilibrium action or exogenous
shocks might lead rational players to play new strategies, or return quickly to the original equi-
librium strategy. Games that are straightforward assurance or coordination games need little
explanation of why the particular strategies are adopted. Other cooperative games where free
riding is possible might require more policing either by the players themselves or external agents.
Games where conflict is more endemic are likely to require greater policing still. The point
remains, however, that different institutions are more or less stable, and thus more or less endur-
ing. Some structures are more enduring than others are. Those which do not require agents to have
‘power over’ others are more enduring than those which do require such power usage, since struc-
tures that do not invoke ‘power over’ can endure with each agent merely using their ‘power to’
get what they want especially in mutually beneficial ways. The greater the amount of ‘power over’
needed to keep a structure in place the less enduring or stable it is likely to be, for it is more prone
to suffer when exogenous shocks shake the power structure, or when losing groups collectively
organize to change the power structure.
Game theoretic analyses of institutional forms can explain a given equilibrium and can
explain why some institutions (those with more stable equilibriums) endure longer. However,
why one equilibrium solution has been adopted rather than another will not be told by the game-
form itself but requires particularistic historical explanation (Dowding 1994). One set of equilib-
rium strategies might have been adopted historically by pure chance perhaps, or might have
served some interests better than others. Most obviously, rules governing property rights can
secure the rights of those who already hold large estates, or can be rewritten to spread property
over a wider community. Which interest prevails is likely to depend upon specific historical loca-
tions. Acemoglu and Robinson (2006) explain the growth of democracy based upon the specific
needs of securing the continued interests of the governing elites. The rich elite controls the state
in non-democratic societies and the poor can only contest power through revolution, especially
when the opportunity cost is low – that is, during very hard times. The elite can buy off the poor
through redistribution but can only credibly do so through democratizing. Where society is rela-
tively equal, redistribution will be limited; where it is highly skewed toward the rich, redistribu-
tive policies will make them much worse off, encouraging them to counter-revolt through elite
coup after a period of democratizing. For Acemoglu and Robinson this explains the essential
difference between Latin America and Northern Europe. The latter had a more equal society at
the crucial democratization stage.
The Acemoglu and Robinson story shares features with many other historical accounts that
examine large differences in the fabric of social structures. They posit a key variable within a
social structure, in their case relative equality, to explain historically diverse courses of history.
Such path dependencies can reinforce existing structures or they can lead to big changes, as in the
democratization process. The essential elements of such stories are that individual actors respond
to the incentives provided by their environment in diverse ways leading to completely different
Journal of Power 27

sets of outcomes. We can explain what is going on in the histories by the relationship between
agent and structure seen as this first level. Agents respond to their environment and their actions
might reinforce existing structures or, at crucial or critical junctures, lead to changes in those
structures. Such changes might be planned by agents (the transformation in the Republic of
Turkey by Mustafa Kemal Atatürk might be a particularly dramatic example) or they might come
about through the unintended consequences of a series of agents’ actions.6
A second level of relationship between agents and structure exists however. Not only do
structures create incentives for agents to behave differently given some underlying set of prefer-
ences; structures can be thought to delve deeply into agents themselves to create those underly-
ing sets of preferences. This I refer to as the deep structure. The deep structure not only provides
incentives for agents to act in certain manners given their objectives, they create those objec-
tives for them. One way of making this point is that it is not that individuals have created their
own propositional attitudes towards the world but those attitudes are created by the structures.
So it is not that I believe x, desire y and given structure Z, I will do α to achieve y: the individu-
alist agency story. Rather, structure Z creates in me my beliefs x and desires y. In this sense,
structure becomes the agency as it creates the neurological activities that lead biological human
beings to act. The ‘agents’ have become mere transmission agents for the structure. Thus, in its
most radical form, deep structuralism takes agency from individuals and gives it to structures.
At the level of deep structuralism, the issue of structure and agency is in fact one of free will or
Now it is not clear to me whether there are any or, at least, many social theories that are deep
structuralist in this sense. To be sure, the issue of autonomy is broached and discussed at the edges
of the agency-structuralist debate. Theorists accuse others of not allowing room for agency or
autonomy, but few explicitly proclaim that human beings have no autonomy or free will. Few
proclaim deep structuralism explicitly. Most authors who write in the field tend to proclaim that
they are transcending the agency-structure divide, or that they are providing a solution to the
agency-structure problem. It is to the solution of the agency-structure divide that I turn in the next

Beyond agent-structure
Most writers who have considered the subject argue that social analysis needs to bridge the
agency-structure divide. Often it is rather difficult to see precisely how they differ in their
attempts to cross the divide (despite their critical comments about others’ attempts). In many
cases the major differences seem to be in terminology rather than substance; the rhetoric of their
arguments differ more than the analytics.
Talcott Parsons sees individuals working within a physical environment to which they
responded and within a social environment, which they both help to create (and recreate) and to
which they responded. In The Social System Parsons (1952) examines the pattern of relations in
terms of the systems theory that was prevalent at that time. The essential element of the relation-
ship between structure and agency is one of role development. Individuals create understandings
of their role in society by internalizing standards that they have learned by seeing other struc-
tured behaviour. The structure has a two-fold ‘binding in’: first, an internalizing of the standards
of conformity that has instrumental or expressive value to the person and, second, through the
structuring of the reactions of others to the agent’s conformity with the standards expected of his
role. Value patterns are thus created by the reactions of people; first, in how they see their own
roles and, secondly, in how others view that role and how they view the agent’s view of that role.
The latter also reacts to the agent’s playing of the role through negative and positive sanctions,
which then modify the agent’s future playing of the role. The attempt here is to see how structure
28 K. Dowding

(expectations about roles) affects agency while also allowing agents to interpret their roles in
ways that change the structure, especially given how others react to those interpretations.
On my reading, Pierre Bourdieu (1977) attempts to reconcile the concepts of structure and
agency though his own further concepts of ‘field’, ‘habitus’ and ‘capital’. The agent is socialized
into a set of fields. A field defines a set of roles and relationships within given sets of social
domain. These fields are relatively autonomous social spaces, which socialize humans into roles.
These roles and relationships change or evolve over time and their worth or capital is defined by
the prestige, the monetary rewards and the status afforded to each role within a field. Thus, differ-
ent activities might be differentially rewarded within different fields, and actors within those
fields would develop different value systems based upon those rewards. Nevertheless, each field
is envisaged as a competitive environment for the different types of worth or capital. We can view
these fields as professions such as ‘academia’ or as more generalized social spaces, in which
agents compete. The agent then learns her role and the relationships that role has within the field
by internalizing a set of expectations and normative understandings of her role. Her internalized
understanding of what is expected of her and how her role fits with others in the field develop
over time and these form her ‘habitus’. Habitus is the system of dispositions, or enduring ways of
looking at the world, that an agent develops in response to the objective conditions in which they
find themselves (the structure of the world around them). The objective structure thus enters into
the subjective awareness of the agent who then processes that structure into her interpretation of
the expectations of her role. In this manner, objective structure enters into the very agency of a
person, but the person is still the agent so-to-speak. The habitus becomes important as the dispo-
sitions imbued create the actions of the agents and these, of course, constitute the field and give
it meaning. So habitus creates the structures of the field, and the field then mediates between habi-
tus and practice.
Critics of Bourdieu contend that habitus leaves no room for individual agency since it seems
to determine the whole of an agent’s disposition to act in one way or another. Defenders suggest
that Bourdieu leaves room within the fields of expectation for individuals to interpret what is
expected of them. I have taken the latter line, as it seems to be the way in which the agency-structure
divide can be bridged. However, Bourdieu does not write much about the autonomous side of
agency and in that sense; he is usually regarded as a deep structuralist. Indeed, he suggests that
agency cannot be directly observed but can only be experienced in practice or through habitus.
However, that is not so far away from revealed preference theory where we interpret an agent’s
preferences (beliefs and desire) to reconstruct their agency. For ‘internalists’ in the theory of mind
we try to track what is ‘really’ going on in the mind that is not observable. For externalists all we
have are interpretations of actions where the subject has no more natural authority over the correct
interpretation than the observer (Dowding 2002; 2008a). In some circumstances the subject might
be the best interpreter of her agency, in other situations the observer might be better placed. What
matters is how much information the interpreters have to go on.
I do not consider Bourdieu’s methodological comments on practice here other than to suggest
they allow room for the observer to interpret action structurally by using the world around agents
in order to understand the agent’s behaviour. In that sense, the process seems hermeneutic and it
could be construed in either an internalist or externalist sense.
One aspect of Bourdieu’s analysis important for considerations of power is his notion of
‘symbolic violence’. ‘Violence’ is a rather dark and powerful word in this context, and we might
lighten it in English by replacing it with the term ‘force’. Some of the elements of the value that
agents see within roles are what he calls ‘symbolic capital’. These are such things as prestige or
honour that are conferred upon people. These elements of symbolic capital (or specific types of
resources) are crucial sources of power for Bourdieu. Any use of the symbolic power successfully
used against another person to change his or her actions is, according to Bourdieu an exercise of
Journal of Power 29

symbolic force. Therefore, where someone acts according to duty, or in ways that they feel they
are expected to behave, they are acting through symbolic force. For instance, if a woman clothes
herself modestly in order to comply with what she believes her family desires, Bourdieu would
interpret her actions in terms of the symbolic force of her family. So symbolic force is the impo-
sition of ways of thinking about the world and one’s place it in. At one level, symbolic power
seems to be a predicate of agents. In our example the woman acts because of the symbolic force
of her family – what she believes they want her to do as communicated by their actions (even
though they do not explicitly tell her what to wear). However, power might also be seen as a pred-
icate of structure, since the imposition of ways of thought seems to be the incorporation of struc-
tures (other desires and actions) into the thought and behaviour of the agent. It is embedded into
our very ways of thinking.
Anthony Giddens argues that all approaches to the agency-structure problem fail adequately
to deal with it because they fail to transcend the ‘dualism’ of agency and structure. They either
give precedence to one or to the other when dealing with any aspect of society. He claims to
replace this dualism with a ‘duality’, which manages to truly transcend the conflict. He claims to
produce an approach that is neither agent nor structure-centred because his approach does not
merely recognize the interface between agency and structure but produces a social ontology
where structure is both the median through which action is made possible and structure is repro-
duced through social practice itself. He claims others see structure as merely constraining but his
notion is that it also enables action. His account he labels ‘structuration’, which is simply intended
to convey the idea that social structures create social action even as social action creates those
structures (Giddens 1979, 1984). Structure for Giddens is the rules and resources that enable
people to act and interact. The modality is the way in which structure is translated into action, that
is, through people whilst interaction constitutes the activity of people within the structure. He
discusses three types of structures: signification – the meaning and interpretation we place on our
behaviour; legitimation – the moral order of ethical norms and so on; and domination – the exer-
cise of power through the control of resources. The three work together as we place meaning upon
some actions given how they are legitimized and underpinned by the control of resources. Power
is central to his account since all social relations involve some kind of power, and this power
encompasses both conflictual but more importantly consensual power as collective action.
Whether Giddens has provided a new approach to the agency-structure question that tran-
scends it to any greater extent than others is a moot question (Archer 1984; 1995; Bryant and Jary
1996). It is not my intention to provide a full account and critique of anyone’s attempts to recon-
cile agency and structure, though a few comments on Gidden’s claims are appropriate for my
argument. Giddens claims that his approach is the true ontology of society where both aspects –
structure and agency – are brought together into one entity. In fact he claims to bring three ‘dual-
isms’ into three ‘dualities’, namely the dualisms of (a) free will and determinism, (b) subject
(people) and object (society) and, (c) static analysis from dynamic analysis. Now it might well be
that the basic ontology of the universe is some seamless flow, but Giddens fails to explain how
we explain that flow without breaking up the social world into manageable pieces. If we cannot
distinguish the explicandum from the explicans or, in the language of statistics, the dependent
from the independent variables, then we cannot manage to explain any outcome, event or institu-
tion since we cannot distinguish it from that which is meant to explain it. Furthermore, whilst we
all acknowledge that dynamic models are generally superior to static ones, it is often necessary
to solve the static in order to get a handle on the dynamic. (Moreover, comparative statics can be
very revealing.) In fact I suggest that, despite his grand claims, Giddens does not offer much more
of an analysis of the agency-structure ‘dualism’ than those who had gone before. For his illustra-
tions of the dualism of free will and determinism, he examines the habits of routine action
(constraint) and the opportunities allowed by structural facilitation. Not only do the illustrations
30 K. Dowding

not transcend the dualism, but also they do not provide insight into when one occurs and the other
does not (Archer 1984, p. 459) and they certainly do not provide any insights into free will and
Similarly his claim that each action is part of the social structure is in one sense trivially true
but in another unenlightening, if not false. We can define the structure of any society as the
conglomeration of all actions within that society; but to contrast two different structures we need
to examine the forms of them that are different – an egalitarian ethos versus a competitive one for
example. However, one would not expect every action in the egalitarian society to be non-
competitive or every actor in the competitive society never to display egalitarian tendencies. In
other words, some aspects of structure are more important than others are, and they are more
important for explanatory purposes. By concentrating upon the ‘flux of the universe’, Giddens
loses sight of the fact that social theory has to be provided in the logical form of explanation. This
is no more apparent in his claim for dynamic over static models. He seems unaware that even
dynamic models have independent and dependent variables that need to be distinguished concep-
tually and (often) temporally.
In reality, I claim, everything that makes sense in Giddens approach to the agent-structure
question, takes us no further than those that went before him. Yes, agents are implicated in struc-
ture and structure in agents, but in social explanation, we pare them apart in order to be able to
explain their relative influence on the outcomes, events or institutions in which we are interested.
This failure is particularly pertinent to the account of power. Power in one form enters as the abil-
ity of people to take the opportunities open to them enabled by structure. However, this certainly
predicates power to people yet does not seem to allow structure enough to say about the relative
capacities of people both to take their opportunities cooperatively. His account of domination
does not here analyse the power inequalities that exist through the inequality of resources that
most social structures engender in one form or another. Contrary to Giddens’ claims, structuration
neither overcomes the dualism of structure and agency nor provides explanatory power of earlier
All approaches to the structure-agency divide note that people act, but they act as constrained
and enabled by others. Their views, interests, beliefs and desires are formed through their inter-
actions with others, and with ideas learned from people in the past. We gain ideas from the past,
modify them with others, form habitual patterns of behaviour, out of which we sometimes break.
The roles in which we find ourselves, sometimes thrust upon us, at other times partly chosen, give
us specific interests, direct our beliefs and actions which they affect others reverberating back into
the future, creating structural forces affecting others living now and in the future. These patterns
of behaviour are not entirely accidental. We can explain many conventions by their enhancing
and enforcing mutual interests. Quite why one society has one convention and another, an entirely
different one that solves the same collective action or coordination problem might be simply
fortuitous. However, it is the mutual benefits of some sets of conventions over others that lead
societies to converge on those sets. In the metaphor of Parikh (2002), such conventions act as
social software enabling societies to run more smoothly. Moreover, humans can examine and
rewrite that software, just as they can with computer software, to help society run more smoothly.
In addition, like computer software they, with their changes, unwittingly can introduce bugs that
cause breakdown under specific conditions.
As part of that social software, people take on specific roles, which further structure their inter-
ests, strategies and interactions with others. In all of these attempts to examine the inter-relationship
of structure and agency, theorists have taken very similar lines. The major differences are in
language used – itself dependent upon these theorist’s backgrounds. Anthropologists tend to read
different books from sociologists, and sociologists tend to have a different background to political
scientists who, in turn, differ from economists. The tools each use differ also. The technical nature
Journal of Power 31

of game theory allows for a close analysis of the mutual benefits of conventions and shows why
equilibrium strategies are adopted even in areas of conflict. Sociologists and anthropologists might
be more interested in the provenance of the preferences more often taken for granted by game theo-
rists and economists. Whereas the former concentrate more upon mutual benefits, the latter query
whether those benefits are so mutual, and are more prone to examine critically whether the conven-
tions and equilibriums benefit all equally. Delving more into history, we can see that conventions
often benefit one set of people over others so entrenching privileged interests. Such historical
accounts lead us to ask whether structured benefits in this manner are formed through power struc-
tures that inevitably favour some groups over others.
Structure and agency are deeply entwined but if some explanations seem to privilege structure
over agency then that is because for that explanation we look to the structure to explain the nature
of the agency. If other explanations seem to privilege agency over structure then that is because
we look to an agent or set of agents to explain structural change. However, the fact that specific
explanations seem to privilege one over the other does not mean we have to believe that the social
theory from which that particular model or explanation was derived always privileges one over
the other. Particularly so, as I have argued, that individualist models generally have structure built
deeply into their conception of the agent (by social role) and structural explanations generally
leave room for agent discretion. Whilst I have argued that power is best seen as a predicate of
agents, that is only because it seems a more natural use of English, agent-centred and structured
accounts can be translated from one into the other through the medium of agent’s resources which
are themselves structurally defined. Nevertheless, preferring the language of structure over
agency or vice versa might reveal some value-commitments as I discuss in the next section.

The valuational element of power

It has long been argued that power is an essentially contested concept and any attempt to define
it will necessarily reveal specific value commitments (Lukes 2005). As an empirical concept that
could, in theory, be measured across sets of people or compared across societies, I believe that it
need not be so contested. To be sure, there might be different accounts of power, but if they have
empirical application, they should be translatable one to another (Dowding 1991, ch. 8; Dowding
2006). However, there is little doubt that the line one takes on agency and structure does reveal
certain interests and values of the inquirer. I have tried to argue that the agency-structure divide
has been bridged, though in specific explanations one or the other might appear privileged. The
divide is no longer a puzzle; indeed, once the relationship between agency and structure is laid
out it is puzzling why it was ever considered a puzzle.8
Now that is not to deny that there are some minor conflicts. I have tied my own flag to the
mast to say that the power is best predicated (at least in English) to agents, but we must also see
that the type of explanation which we offer in social science has to be structural as much as agen-
tial. This is so because the interests of social agents in social explanation are often defined almost
exclusively in structural terms, and can be seen as proceeding from or caused by structures. I do
not exclude structural accounts of power as such, they are not meaningless, but argue that they
can be translated into my preferred form of power analysis without loss. In other words, when we
predicate power to individuals we can see the source of that power within structures and further
recognize the structural influence upon individuals’ interests. Similarly, when we predicate
power to structures we can still leave room for human agency. However, why might we some-
times want to concentrate upon power as structural and sometimes as agential? Whilst these are
not incompatible views, each does suggest different value commitments.
Mainstream political science has almost left ‘power’ behind.9 When discussed, it is largely in
terms of the ‘power of the president’ or ‘power of the prime minister’ or discussion of the relative
32 K. Dowding

power of countries (at the supra-national level) or between organizations within a state or across
states.10 The more structural or social issues that once held rapture in the ‘community power
debates’ have been left behind in network, regime or other similar analyses. It would be incorrect
to suggest that issues of structural power are not implicated in these terms, often explicitly so (see
Stone 1980, for example),11 nor that there is not a vibrant sub-community of those whose central
topic is power within political science (I am one of them). However, mainstream political science
has left power studies behind (as Jordan and Richardson (1987) suggested, in order, perhaps, to
concentrate upon topics more easily graspable). One reason might be that mainstream political
science examines policy outcomes at a detailed level. In order to explain those outcomes it brings
in all the resources and events that agents have in order to show how an institution or policy
outcome formed. In this agents and institutions (or structure) both play a part, but the concentra-
tion is upon the specific processes that lead to policy outcomes. Alternatively, where political
science examines institutions as dependent variables, it does so again at a concrete level of how
a specific institution form was chosen rather than another.
Given that mainstream political science tends to view power as a predicate of agents, and
agential power is essentially about what agents could do, and not simply about what they actu-
ally do, power does not become a focus of study or a term within the analysis. Analysing the
power structure at a given moment in time is to analyse a game form and not the game itself.
That is, it is to examine the possible strategies and outcomes, and not those that are played, given
the specific interests in play at any one moment in time (Dowding 2008c). Thus, the agential and
particularistic focus takes us away from explicit power considerations. Alternatively, to put the
point another way: once one has analysed an outcome through the interplay of agential
resources, there is no need to mention power at all. It has become redundant; it has been analy-
sed away (Moe 2005; Dowding 2008c). Thus, positive or empirical analysis tends to analyse
situations without mentioning power at all. Power is still normatively important, in terms of
what could have been, or in what should be the case. We can say precisely the same for econom-
ics – in fact, here power is even further analysed away (see Dowding 2008c for further argument
on this issue).
Anthropology and sociology, however, are still involved in broader accounts of power.
Sociology has less of a focus upon specific outcomes (and where it does what I have said about
political science tends to apply). These more general concerns of the effects of culture, socializa-
tion, socio-economic class and so on upon behaviour examine the structural side of the agency-
structure relationship. The questions posed in sociology are more likely to be concerned with the
effects of social processes on human behaviour, such that the agency is the dependent variable
and the structures the independent variable. Thus, these studies tend to concentrate more upon the
structural side.
Dividing the nature of the social inquiry into ‘political science/economics’ and ‘anthropol-
ogy/sociology’ as I have done might be misleading. The analytic point is that where the depen-
dent variable is an event such as a policy outcome, or an institution, then the independent
variables will include agents as well as structural factors. Where the dependent variable is behav-
iour of a set or class of people, then the independent variables will largely be composed of struc-
tural variables. Thus, we see the nature of the question posed directing the concentration upon
one or other side of the agency-structure divide. However, the fact that interest is posed in this
manner does not entail that researchers cannot recognize the truths contained in seeing structure
and agency as a duality or structuration bridging the agency-structure divide. Both agential and
structural variables are important in social explanation. Too often, when theoretical debates
become embroiled in agency-structure debate writers choose examples from one general tradi-
tion over another to illustrate their argument, thus making the division appear greater than it
perhaps is.
Journal of Power 33


In this paper, I have examined the agency-structure problem. I have argued that the manner in
which most individualists conduct research is compatible with seeing agents in a highly struc-
tured manner. I have also suggested that most ‘structuralists’ leave some room for human agency.
I further suggested that many authors have ‘solved’ the agent-structure problem in very similar
ways, and the differences between these solutions are more rhetorical than analytical, and minor
differences need not distract us from the fact the problem is solved. That is not to say that there
are not real differences between accounts of power that are tied to individuals and those tied to
structure. I have reiterated my claim that ‘power’ is best seen as a predicate of individuals since
that is its more natural form in the English language. Accounts rendered in a structural form can
be translated into an agency form without loss and a similar translation is possible the other way.
However, that is not to say that the choice to concentrate upon the structural over the agent-based
(or vice versa) conception of power is not without value. Or, rather, the specific research question
that one is posing might naturally lead to a concentration upon either structure or agency. In that
rather weak sense, there is an evaluative element to seeing power in agency or structural terms.
The translatability of one form to the other, however, suggests that this does not entail any essen-
tial contestability in the concept of power.
There is an important caveat to my argument about the translatability of agent-based and
structure-based accounts of power. I distinguished between a structure seen as merely affecting
the strategies of agents given their interests and structure affecting the very nature of those inter-
ests. The latter form of structuralism I described as ‘deep structuralism’. There can be shallower
and deeper forms of structuralism. One that leaves no room for agent autonomy is the deepest
form of structuralism. I suggested that structural accounts are deeper when structure is seen to
affect action not simply through the strategy that actors develop in order to achieve their aims,
but also when it affects those aims themselves. Structure could affect those aims at an ever-deeper
level. Explaining an individual’s action using the intentional stance (Dennett 1987), we give the
agent’s ‘reasons for action’. Indeed that is how we interpret our own actions. Now our reasons for
action, to a greater or lesser extent, might be formed through customs, habits, ethical rules and so
on. These elements structure our reasons for action. If we fully explain a person’s reasons for
action in terms of these structures then we leave no room for agent autonomy or free will. Even
when there appears to be an opportunity for choice, then the choice itself will be determined by
the relative structural weighting in the person’s mind. It has been my argument that, in fact few,
if any, writers have explained structure in quite the deep sense: ‘deep’ because the structure goes
all the way into the mind of the subject. One reason, perhaps, that structure is not normally
defended as providing all the explanation is that it is not clear how it can provide agency – how
it can cause agents to act down to their smallest components.
It is hard to see how we can use structure in explanations of individual human action that are
fine-grained enough to predict every aspect of a person’s action. This fact suggests that even if
all of our actions are fully determined by the interaction of our genes and our environment – there
is no free will or autonomy as classically understood – then we would still need to leave room for
individualized ‘reasons for action’ that we cannot perceive how they are structurally determined.
It is for this reason, I suspect, that even the deepest of structuralists wants to leave room for human
agency. Not necessarily, because ontologically speaking there is such agency – but simply
because we do not have the tools to provide completely specified structural explanations of
human behaviour down to the smallest detail. Giving agents’ ‘reasons for action’ is the best expla-
nation, and provides the most efficient predictions (Dennett 1987, 2003).
The fact that we need to leave such room for human agency at the individualized level does
not mean that we cannot adequately explain the large differences between the types of outcomes
34 K. Dowding

we see at the societal level purely in terms of structural features of societies. Many properties of
societies can be expressed only at the societal level; the degree of inequality, the class nature of
social power, the family or network oriented nature of political power, and so on are all social
properties. They affect the amount of power, freedom and welfare of each individual in society
to be sure, but we do not explain those individual properties by simply agglomerating them.
Rather, we do so by categorizing in a structured manner. Our interpretation of power relationships
will largely be determined by social categories as differentially structured in assorted societies.
Nevertheless, we leave room for biological individuals to have different qualities, and thus power
resources, even though they are members of specific classes or sets people we have categorically
structured. There is no agency-structure puzzle here: simply different types of questions that need
to be addressed using different tools and focus.

The author should like to thank Claus Offe for our conversations during the time of writing this
essay that affected its form and content in a manner not entirely within knowledge of the author.

1. I do not make a distinction between social and political power in this paper. Whether or not one consid-
ers political power to be a subset or identical to social power depends upon the breadth of one’s defini-
tion of what constitutes ‘the political’. I use the term ‘power’ here in the social sense.
2. Some will further argue that Arrow’s theorem show that there is no sense in which a collective being
has a collective will (see Riker 1982). For a modern view on how collective can have a common will
despite Arrow (1951/1963) see Pettit (1993).
3. ‘Natural disasters’ might have human causes of course – global warming an example obvious to most
scientists. Other ‘natural’ disasters might also be disasters because of social structures – famine an
example much discussed (for example, Drèze and Sen 1999). Floods might be natural disasters because
of where people have chosen to locate, or due to government decisions about flood defences. My point
however, is to the extent that we can locate an event with important social consequences beyond human
action it is, according to the individualist, exogenous to social explanation. To the extent that it can be
located within human action then the methodological individualist can point to individual action as
locus of social explanation.
4. I argue in Dowding (2008b) that how we identify and measure ‘responsibility’ is directly related to how
we measure luck. If we were to remove all luck from the determination of character, we would also
remove all responsibility. See the penultimate section below.
5. This is so even for physical aspects of phenotype. Newborn rats are blind, and without a source of light
to ‘switch on’ the relevant genes, they remain blind.
6. Systematic attempts to examine how individuals can dramatically change political and social situations
often rely upon not only the skill and perception of the agents, but also the opportunities they grasp often
at a crucial moment in history. Riker (1982; 1986) calls such agents ‘herestheticians’. In his account,
heresthetic politicians forge new coalitions of interests to transform political cleavages into new ones,
and then modify the political or social situation onto a new path. Nevertheless, heresthetic politicians
can only utilize the opportunities provided by such moments. Similarly, the application of ‘punctuated
equilibrium’ to policy formation suggests that periods of stability where policy problems parts caused
by the institutionalized solutions to old problems lead to ‘crisis’ which enables more radical policy
development. Equilibrium is punctuated by rapid change (Baumgarter and Jones 1993, Jones et al.
1998, Jones et al. 2003).
7. We say above an outcome might be fully determined in the sense that a person can do nothing about
stopping something happening, but that in itself does not bear on the free will issue. However, if I cannot
break a habit then perhaps I have no autonomy, but if I can choose or not to take opportunities then
perhaps I do.
8. The puzzle might be, is there free will? Alternatively, in another form, what does agent autonomy mean?
However, the inter-relationship between agents and their environment including the social environment
of past and present actions of others and how this affects agent’s beliefs, desires and actions, is not a
Journal of Power 35

puzzle. In other words, we might not have a clear way of deciding between deep structure and any
approach which combines agency and structures; but we should have no problems in understanding the
latter combinations – even if they are not all identical.
9. Jeffrey Isaac, who has published one of the best ‘realist’ account of power (Isaac 1987) suggested at a
panel on Dahl’s concept of power fifty years on, that ‘power studies were now irrelevant to political
science (Isaac 2006). His claim was certainly hyperbolic, but it does contain some truth.
10. Alternatively, at a more detailed level, in terms of ‘voting power’ of such bodies.
11. It is also the case that power is still discussed more explicitly in urban studies than at the national and
supra-national level.

Notes on contributor
Keith Dowding is Professor of, and Head of Program in, Political Science in the Research School of Social
Sciences at the Australian National University, Canberra. He has published seven books including Rational
Choice and Political Power and Power as well as numerous articles in the fields of political and social philos-
ophy, social choice, urban politics, public administration, comparative politics, and British politics, several
of them exposing balatronic dialect in various forms. He is co-editor of the Journal of Theoretical Politics.

Acemoglu, D. and Robinson, J.A., 2006. Economic origins of dictatorship and democracy. Cambridge:
Cambridge University Press.
Adorno, T.W., Frenkel-Brunswik, E., Levinson, D.J., and Sanford, R.N., 1950. The authoritarian person-
ality. New York: Harper and Row.
Altemeyer, R., 1996. The authoritarian specter, Cambridge, MA, Harvard University Press.
Archer, Margaret, 1984. Morphogenesis versus structuration: On combining action and structure, British
Journal of Sociology, 33, 455–483.
Archer, Margaret, 1995. Social theory: The morphogenic approach. Cambridge: Cambridge University
Arrow, K.J., 1951/1963. Social choice and individual values. New Haven: Yale University Press.
Baumgartner, F. and Jones, B., 1993. Agendas and instability in American politics. Chicago: Chicago
University Press.
Boin, A, ‘tHart, P., Stern, E., and Sundelius, B., 2005. The politics of crisis managment: public leadership
under pressure. Cambridge: Cambridge University Press.
Bourdieu, P., 1977. Outline of a theory of practice. Cambridge: Cambridge University Press.
Bryant, Christopher G.A., and Jary, David, eds. 1996. Anthony Giddens: Critical assessments. London:
Dahl, R.A., 1957. The concept of power. Behavioral Science, 2, 201–215.
Dahl, R.A., 1968. Power. In David L. Sills ed. International Encyclopaedia of the Social Sciences vol 12.
New York, Free Press, 405–415.
Dennett, D.C., 1987. The intentional stance. Cambridge, Mass: MIT Press.
Dennett, D.C., 2003. Freedom evolves. London: Allen Lane.
Dowding, K., 1991. Rational choice and political power (Aldershot: Elgar).
Dowding, K., 1994. ‘The compatability of behaviouralism, rational choice and “new institutionalism”.
Journal of Theoretical Politics, 6, 105–117.
Dowding, K., 1996. Power. Buckingham: Open University Press/Minnesota University Press.
Dowding, K., 2002. Revealed preference and external reference. Rationality and Society, 14, 259–284.
Dowding, K., 2003. Resources, power and systematic luck: Reply to Barry. Philosophy, Politics and
Economics, 3, 305–322.
Dowding, K., 2006. Three dimensional power: a discussion of Steven Lukes power: A radical view second
edition. Political Studies Review, 4, 136–145.
Dowding, K., 2008a. Defensa de la preferencia revelada. Revista Internacional de Sociología 1.
Dowding, K., 2008b. Luck and responsibility. Critical Review of International Social and Political
Philosophy, 11, forthcoming.
Dowding, K., 2008c. Rational choice perspectives. In Stewart Clegg, S. and Mark Haugaard eds. Handbook
of power. London: Sage.
Dreze, Jean, and Sen, A., 1999. The Amartya Sen and Jean Dreze omnibus. Oxford: Oxford University
36 K. Dowding

Dunleavy, P., 1991. Bureaucracy, democracy and public choice. Hemel Hempstead: Harvester Wheatsheaf.
Gibbard, A., 1973. Manipulation of voting schemes: A general result. Econometrica, 41, 587–601.
Giddens, A., 1979. The central problems of social theory. London: Macmillan.
Giddens, A., 1984. The constitution of society. Cambridge: Polity.
Goodin, R.E., 1980. Manipulatory politics. New Haven: Yale University Press.
Isaac, J., 1987. Power and Marxist theory. Ithaca, NY: Cornell University Press.
Isaac, J., 2006. Thoughts on the concept of power. APSA, Congress Philadephia, September.
Jones, B., Baumgartner, F., and True, J., 1998. Policy punctuations: US budget authority, 1947–1995. Journal
of Politics, 60, 1–33.
Jones, B., Sulkin, T., and Larsen, H., 2003. Policy punctuations in American political institutions.
American Political Science Review, 97, 151–169.
Jordan, G. and Richardson, J., 1987. British politics and the policy process. London: Unwin Hyman.
Lukes, S., 2005. Power: A radical view second edition. Basingstoke, Palgrave Macmillan.
Martin, J.L., 2001. The authoritarian personality; fifty years on: what questions are there for political
psychology. Political Psychology, 22, 1–26.
Moe, T., 2005. Power and political institutions. Perspectives on Politics, 3, 215–234.
Morriss, P., 2002. Power: A philosophical analysis. Manchester: Manchester University Press.
Oppenheim, F., 1981. Political concepts: A reconstruction. Oxford: Basil Blackwell.
Parikh, R., 2002. Social software. Synthese, 132, 187–211.
Parsons, T., 1952. The social system. London: Tavistock.
Pettit, P., 1993. The common mind. New York: Oxford University Press.
Riker, W.H., 1982. Liberalism against populism: A confrontation between the theory of democracy and the
theory of social choice. San Francisco: W.H. Freeman and Co.
Riker, W.H., 1986. The art of political manipulation. New Haven: Yale University Press.
Satterthwaite, M., 1975. Strategy proofness and arrow’s conditions: Existence and correspondence theo-
rems for voting procedures and social welfare functions. Journal of Economic Theory, 10, 187–217.
Stenner, K., 2005. The authoritarian dynamic. Cambridge: Cambridge University Press.
Stone, C.N., 1980. Systemic power in community decision-making: A restatement of stratification theory.
American Political Science Review, 74, 978–990.
Weber, M., 1978. Economy and society vols 1 and 2. Berkeley, CA: University of California Press.