7
Studying Power
John Scott
There are three dominant methodological traditions through which power is studied in empirical political sociology. The reputational approach looks at those who are believed to have power. Increasingly, however, it is thought that this is evidence only of images of power on the part of those asked. Structural approaches focus on strategic positions in the central organizations and institutions of a society. Decision-making approaches are based on the claim that the reputational and positional approaches ignore what actually happens when decisions are made. Scott favours the structural approach, arguing that it can, and should, incorporate the insights of the others: decision making can only be studied where there is understanding of the important structures within which decisions are taken; and perceptions of power can best be understood where there is independent knowledge of the positions people believe to be powerful.
The principal approaches to the study of power have generally been seen as bitter rivals and as offering mutually exclusive paradigms of research (Scott 2001). They have each come to be associated with quite distinctive methods of research and analysis. Indeed, it has even been claimed that the theoretical starting point determines not only the choice of research methods but also the substantive conclusions that can be drawn from the research (Walton 1966). To see things in this way is to put the point far too strongly. While there are certainly affinities between theoretical approaches and research methods, leading to distinctive research traditions, these are not tight and rigid connections. The merits and demerits of the various research methods can be considered independently of the particular theoretical approach adopted. In this chapter I concentrate principally on the virtues of one research tradition, but I take it as axiomatic that the theoretical approaches associated with these traditions must be seen as complementary perspectives rather than as all-or-nothing rivals (Moyser and Wagstaffe 1987).
The Wiley-Blackwell Companion to Political Sociology, First Edition. Edited by Edwin Amenta, Kate Nash, and Alan Scott.
© 2012 Blackwell Publishing Ltd. Published 2012 by Blackwell Publishing Ltd.
70 JOHN SCOTT
Table 7.1 Traditions and research methods
Preferred research methods
Research tradition
Paradigmatic study
Data collection
Data analysis
Object of analysis
Reputational    Hunter (1953)
Expert judgement,   Voting, ranking,
interviews and rating scores
Documents Frequency
Images of power
Structural
Mills (1956)
Positions of power
distributions, social network analysis
Decision making
Dahl (1961)
Observation, Policy outcomes
interviews
Agencies of power
Three dominant research traditions have generally been identified in the study of power. These are the reputational approach, the structural approach and the agency or decision-making approach (Crewe 1974). Each of these traditions is associated with a study that exemplifies its research methods and techniques and that has provided a model for later researchers. Table 7.1 presents a simplified summary of the links between the research traditions and their preferred methods of research.
The reputational approach to power has as its main concern those agents who are reputed to be powerful. While it has often been assumed that this method can give direct evidence on actual power relations, it has increasingly come to be realized that, in fact, it evidences only images of power. Structural approaches to power focus directly on the attributes of strategic positions in the central organizations and institutions of a society. These positions are held to be central to the control of the resources that are the basis of power, and the occupants of these positions are the central actors in the exercise of power. Decision-making approaches have been based on the claim that reputational and positional approaches have been overly formalistic. They have looked at formal, official definitions of power and have ignored what really happens when decisions are made. Not all of those who occupy positions of formal authority will be equally involved in all the various stages of decision making, and the only proper way to investigate power, it is held, is to do so directly at its point of exercise.
My own position is that the structural approach has the most to offer to researchers on power and that it provides a basis for incorporating the insights of the rival approaches. It is possible to study decision making only if we have an understanding of the structure within which these decisions are made, and perceptions of power can best be studied if we have some independent knowledge of what it is that the participants are trying to perceive. The starting point for any study of power, then, must be a structural analysis.
Each tradition relies on particular techniques of data collection and data analysis, these techniques being very widely used in the social sciences. I will not attempt to give a comprehensive coverage of such techniques as survey methods, interview methods and the use of documents. Instead, I will concentrate on the features of these research methods that are most particular to the study of power and that raise particular issues
STUDYING POWER
71
in power research. As the focus of this discussion is on studying power, I have not discussed the research methods used in elite studies more generally, where the focus is not on power but on elite attitudes, values and behaviour (Putnam 1973; see also Moyser and Wagstaffe 1987, eds).
Power can be studied at a number of levels of analysis, and these will figure in this discussion. Some research has focused its attention on the national level, investigating power relations in and around the nation-state. An important tradition of research, however, has been concerned with power at the community level, in towns and cities within nation-states. There are, of course, important theoretical and substantive issues that surround the choice of an appropriate level of analysis, as well as about the extent to which global power relations should be considered alongside the national and the local. However, the research issues that arise in each of these areas are, in general, similar, and there is little need to make explicit reference to the level of analysis here.
Images and Decisions
The paradigmatic study for the reputational approach is that of Floyd Hunter (1953), for whom the central concern in a study of power was to identify those people who, according to general opinion in their community, exercise the greatest amount of power. It is perceptions or images of social positions and their occupants that are of interest to Hunter. In this respect, Hunter's work is similar to studies of images of class (Warner 1949; Lockwood 1966; Bulmer 1974) and of images of society more generally. In his work, however, he tends to gloss the distinction between images of power and the actual exercise of power. Hunter's 'positional' approach to power saw it as 'the acts of men [sic] going about the business of moving other men to act in relation to themselves or in relation to organic or inorganic things' (Hunter 1953:2-3, emphasis removed). (Note that all the writers considered in detail here followed the sexist practice of referring to 'men' instead of 'people', and there is little or no discussion of the practices through which women have been excluded from power. In the direct quotations used in the rest of this chapter, I have left the argument in the words actually used by the researcher.) The resources that made such power possible were seen as being tied to social positions, and so the focus of any investigation must be on those who occupy prominent positions in various types of groups or associations. This starting point is the same as that of the structural approach, but Hunter wanted to move from structures to reputations.
In his study of community power in the financial, commercial and industrial centre of Atlanta, Georgia (called 'Regional City' in the original report), Hunter aimed to identify powerful individuals in four arenas of power - business, government, civic affairs and 'society leaders and leaders of wealth' (Hunter 1953:169). He sought key informants in the leading organizations and associations in each of these arenas, asking them to name the chairmen and other leaders in the principal organizations in each of the four arenas of power. Many such office holders could, of course, have been identified from published documentary sources, as has been the case in more explicitly structural research, but Hunter was keen to tap into the knowledge and opinions of his key informants from the beginning.
72
JOHN SCOTT
The lists produced by the key informants were given to a panel of 'judges', whose job it was to use their knowledge to reduce them to a more manageable 'top ten of influence' in each arena. The panel of judges was supposed to be representative of the community in terms of religion, sex, age and ethnicity, and they were also supposed to be representative of business and the professions (though no attempt was made to ensure that they were representative of other occupational groups). The community influentials - seen as the holders of power - were defined as the 40 people who received the largest number of votes from the panel of judges. (Note, however, that while the panel of judges was supposed to be representative of middle-class opinion, there is evidence that they were far less representative even than this. Hunter found that no African Americans appeared on the list. This reflects, of course, the lack of power - real or reputed - held by African Americans in the southern states of America in the 1950s, but it also seemed, to Hunter, to reflect the unwillingness or inability of his key informants to recognize those African Americans who did achieve positions of power. To overcome this, Hunter made an ad hoc extension to his research by carrying out a parallel sub-study within the black community, arguing that there was a divided structure of black-white power. This argument is analogous to Warner's claims about black-white class relations in the Deep South (Warner 1936; see also Davis, Gardner and Gardner 1941).
This was, of course, an arbitrary limitation, and Hunter's claim that these people were typical of a larger group of powerful persons (Hunter 1953: 61) highlights a problem that occurs in all projects where only a sub-set of the powerful are studied. This is the problem of sampling. When a researcher does not cover the whole of the target population, whether by accident or design, it is important that the nature and representativeness of the resulting sample is examined. In general, it is preferable to use explicit sampling criteria in the first place, though this may not be possible when the size and composition of the target population are unknown or unspecified.
While it purports to investigate the actual holders of power, the reputational approach, at best, provides evidence on images of power. The images disclosed are those of the expert judges, or the larger social groups of which they are representative. As such, it is important for a reputational study to identify clearly its target group: Is the aim of the research to identify those that a whole society rates as the most powerful, or those that one class, sex or ethnic group within it rates as the most powerful? Such questions can be answered only on the basis of some knowledge about the actual structures of power and the wider social structure.
The paradigmatic study for the decision-making approach to power is that of Robert Dahl (1961), who was one of the earliest critics of Hunter and of structural approaches. Structural, or positional, approaches, he argued, presuppose that an elite exists, and a methodology that concentrates on top positions will inevitably conclude that an elite does, in fact, exist. The whole process, he argued, is circular. For Dahl, the existence of an elite had to be demonstrated through the direct investigation of decision making. He holds that 'A has power over B to the extent that he can get B to do something that B would not otherwise do' (Dahl 1957:202-203), and this is studied by measuring the actual participation of position holders in specific key decisions (see Lukes 1974 and the 'Introduction' and the reprints of key contributions by Dahl, Lukes, and others in Scott 1994).
STUDYING POWER
73
Paradoxically, Dahl also began his research with the identification of structural positions. His study of New Haven, Connecticut, in the 1950s identified a large number of positions that he thought had the potential for power and influence in the community. These included office holders in the city administration (elected and appointed), local businessmen and various 'social and economic notables'. The latter were large property holders and directors, and those active in 'Society' activities. While this starting point looks little different from that of Hunter, Dahl was not using it to delineate a group of actual power-holders. Rather, he wanted simply to identify a large population of potential holders of power, so that he could then go on and identify which of them were involved in the active exercise of power. This was the question that was to be investigated through an examination of their participation in the making of key decisions in the community.
Dahl's study of politics and his decision-making methodology have been emulated by many other political scientists and sociologists, though few have undertaken the kind of detailed and careful investigation of processes and policy outcomes that Dahl himself undertook. At the level of community power are studies by Vidich and Bensman (1968), Birch (1959) and Wildavsky (1964), while at the national level there have been Rose (1967) in the United States and Hewitt (1974) in Britain.
Dahl concentrated on a number of 'issue-areas' - urban redevelopment, local schooling, and nominations for political office - and within each of these he looked at specific decisions such as the formation of a Citizen's Action Commission, the redevelopment of particular streets and squares, the introduction of eye tests in schools, changes to educational budgets, policies for dealing with delinquency at school, nominations for election as mayor and proposals for a new city charter. Dahl and his researchers sought to use interviews, observations and documents to identify who proposed particular alternatives, who spoke in discussions, when and how proposals were modified or rejected, and who voted for each proposal when a final decision was arrived at. He concluded that a great many people were involved in initiating or vetoing proposals, and that they tended to be actively involved only in those areas where they had particular professional or occupational interests. Only the democratically elected politicians were centrally involved in more than one proposal (Dahl 1961:181-183). The positional resources of the economic and social notables gave them only the potential for power, but very few of them either tried or succeeded in converting their potential into actual influence in decision-making processes. He further argued that political decisions were shaped by the lobbying and pressuring activities of a variety of groups. The outcome of decision-making processes did not uniformly express the interests or advantages of any one group. Power in New Haven was 'pluralistic' rather than elitist (see also Polsby 1980).
The problems with the decision-making approach are, of course, that there is no certainty that researchers will either get access to those who really make decisions or be able to uncover the key participants. To the extent that decisions are made behind closed doors, away from the glare of public scrutiny, then political scientists and sociologists are unlikely to be able to observe these decisions or to interview those involved (Bachrach and Baratz 1963, 1975). This critique points to the need to investigate the 'non-decision-making' processes that occur behind the scenes and that serve to keep some issues out of the overt decision-making process. From this
74
JOHN SCOTT
standpoint, the 'potential' power inherent in structural positions has a far greater significance than Dahl allowed.
The necessity for a structural framework is also apparent in the need for an objective criterion for identifying which decisions are the most important or strategic in a community. Which decisions are important, and which are not, is a matter that can be decided only in relation to the overall structure of the society and the distribution of advantages and disadvantages within it. Without such information, the researcher may end up looking only at the marginal and unimportant decisions that the real rulers could safely leave to others. The implication of this kind of criticism, then, is that the very structural concerns that Dahl sought to eliminate must, indeed, find their place in a comprehensive investigation of power.
Structures of Power
If structures of power are to form the centrepiece of power research - and both Hunter and Dahl began with the identification of structural positions - how is this to be carried out? The paradigmatic study for this approach is that of Mills (1956), who used the positional method to study national-level power in the United States. Where Hunter and Dahl identified positions of power simply as their starting points, Mills saw this as central to the whole project. Power, he held, resides with all those 'who are able to realize their will, even if others resist it' (Mills 1956: 9). While the identity of the particular individuals is recognized to be important, it is the attributes of the positions that they occupy that are seen as more fundamental to power relations. Someone exercises power as an occupant of a particular position, subject to the constraint exercised by the occupants of other positions. Without their positions, individuals have no significant power.
Power is located in the top positions of the institutional hierarchies that define the social structure of a society, and the distribution of power varies with the shape taken by this structure. As the institutional hierarchies of a society become more centralized, so the distribution of power becomes more concentrated: 'As the means of information and of power are centralized, some men come to occupy positions in American society from which they can look down upon, so to speak, and by their decisions mightily affect, the everyday worlds of ordinary men and women' (Mills 1956:3). Mills' central concept of the power elite follows from this view of power. The institutional hierarchies form a structure of power, and it is the overlapping and interlocking of their top positions that forms a power elite. A power elite, then, comprises the 'men whose positions enable them to transcend the ordinary environments of ordinary men and women; they are in positions that allow them to make decisions that have major consequences' (Mills 1956: 3-4).
Mills saw three institutional hierarchies at the heart of the power elite in the United States of the 1950s. These were the economic, the political and the military hierarchies. As the identification of positions of power was to be the heart of his study, Mills sought to be as comprehensive and as systematic as he could in his use of evidence. Instead of relying on the knowledge of key informants, he went directly to the documentary sources that provided a full coverage of these positions. Although his precise selection criteria varied from case to case, Mills did make great attempts to be
STUDYING POWER
75
systematic and rigorous in his data collection. In most cases, he collected data for the full set of positions over three generations.
Within the economic arena, Mills noted the twentieth-century growth of the corporate sector at the expense of personal, privately owned enterprises. He therefore focused his attention on those positions that formed what he called the 'corporate rich'. This category included holders of substantial wealth (termed the 'very rich') and holders of corporate office (the 'corporate executives'). The very rich were operationally defined as those men and women with assets of $30 million or more, and lists of names were compiled from a variety of official, corporate and secondary sources (on the use of documentary sources in power and other studies see Scott 1990). He defined 'corporate executives' rather loosely as the 'top two or three command posts in each of those hundred or so corporations which... are the largest' (Mills 1956: 126), and similar data on them were collected by his PhD student, Suzanne Keller.
This definition of corporate executives highlights a general problem in positional studies of power. This is the problem of defining and bounding the positions that are to be studied, sometimes referred to as the problem of system boundaries. While any such decisions are likely to be arbitrary, it is important that the criteria are both clear and consistently applied. For example, we must know whether the category of 'top' corporations includes the largest 50,100,200 or 500 corporations, and we also need to know by what criterion 'size' is measured. Similarly, we must know which actual positions are to count as the 'top' positions within them. Do we include just the president (chief executive), all the office holders or all the directors? There is no simple answer to such questions, as the boundary criteria that need to be used will vary from one situation to another.
In the political and military arenas, Mills focused his attention on what he called 'the political directorate' and 'the warlords'. The political directorate is a category that includes all the leading positions of state: President, Vice President, Speaker, Cabinet members, and Supreme Court Justices. His list also included a number of positions that had grown in importance in executive decision-making over the course of the century. These were the Under Secretaries, Directors of Departments, Members of the Executive Office of the President, and White House Staff. The warlords were all generals and admirals, including - most importantly - those holding office in the Pentagon. Like the political directorate, these office holders were identified from official documents that listed the positions and their occupants. As with the corporate rich, the boundaries of the 'top' positions in the political and military hierarchies were, inevitably, drawn arbitrarily, as a decision must always be made about which positions are important enough to include. An attempt to set out a framework for such matters in relation to identifying a political elite can be found in Giddens (1973a).
Mills' power elite comprised the overlapping groups of the corporate rich, the political directorate and the warlords. This emphasis on the analysis of overlapping memberships has been a central characteristic of structural studies of power. These studies have investigated the overlap among positions of power by the more or less systematic use of methods of social network analysis. Hunter had used these same techniques rather more systematically. He used rudimentary methods of social network analysis to construct sociograms of interaction among the reputedly powerful, concluding that there was evidence for the existence of various 'crowds' or 'cliques' within the leadership group (Hunter 1953: 77-78).
76
JOHN SCOTT
The systematic use of social network analysis has gradually become more central to structural research on power, as the advanced techniques developed since the 1960s have allowed more rigorous investigations into the formation of cliques and other sub-groupings. In social network analysis, individual positions are represented as points in a diagram or as rows in a matrix, while the social relations that connect these positions are represented as lines connecting the points or as the individual cells of the matrix. Mathematical techniques are now available to chart the size and structure of social networks through such measures as density, centralization and fragmentation (Scott 1991b; Wasserman and Faust 1994; de Nooy, Mrvar and Batagelj 2005). Density measures the coherence or integration of a network - how closely connected its members are. Centrality, on the other hand, concerns the relative prominence of members in the network. At an overall level, centralization measures examine the extent to which a network is organized around focal units. Particularly important measures in structural analysis are those that identify cliques, clusters and other sub-groupings that cross-cut the formal boundaries of institutions (Knoke 1994).
The most systematic and theoretically sophisticated examples of the use of the structural approach can be found in the work of Domhoff (1967,1971,1979,1998, 2009) and those who have been influenced by him. In these studies, structures of powerful positions are investigated in relation to the social background and policy preferences of those who occupy them. Domhoff has explored the consolidation of capitalist class power through the formal and informal networks involved in the special-interest process, the policy-formation process, the candidate-selection process and the ideology process. In Britain, a similar approach has been used in works by Guttsman (1963), Miliband (1969) and Scott (1991c). Scott has shown that the 'old boy' networks of British politics can be explored through the structural analysis of power blocs and the structure of intercorporate relations in business. Such work has recently been enlarged in the growing number of studies into policy networks (see Marsh 1998).
The approach has been especially important in analyses of economic power in large corporations (see Mizruchi 1982; Scott and Griff 1984; Mintz and Schwartz 1985; Stokman, Ziegler and Scott 1985; see also Scott 1991a. Some of the key studies of political and economic elites using these methods can be found in Scott (1990, ed.). Such work has examined interlocking directorships and intercorporate shareholdings, showing the organization of economic power around structured relations between industrial and financial interests (Scott 1997). Central to many such studies has been a critical examination of the managerialist ideas of writers such as Burnham (1941) and Berle and Means (1932), who share many of the assumptions of the pluralist writers. Rejecting this point of view, the works of Mintz and Schwartz in the United States and Scott in Britain have documented the existence of structures of bank centrality through which finance capitalists are able to coordinate the affairs of the numerous corporate boards on which they sit. Through their interlocking directorships, these multiple directors become the most important force in the corporate power structure.
Conclusion
Each of the traditions that I have reviewed has produced important work, showing the potential and the value of the particular methods used to study power. However, each
STUDYING POWER
77
also has its limitations, and I have tried to sketch these out. The trite conclusion is undoubtedly that no one tradition has a monopoly of the truth, and they must, ideally, be combined in a single research design (Dowding 1996: 58ff and see also 1995 where he downplays the significance of structural concerns in an otherwise useful survey. This seems to be based on his appraisal of the limited results appearing in the relatively new area of policy network research). This is not to say that they carry equal weight. I have argued that the structural approach provides the best basis for integrating the results of research on participation in decision making and the images of power that motivate participants. It provides powerful techniques for mapping and measuring power relations, and it provides the essential framework for understanding processes of decision-making power.
Further Reading
Scott, J. (ed.) 1990: The Sociology of Elites, 3 vols. Cheltenham: Edward Elgar. Scott, J. (ed.) 1994: Power, 3 vols. London: Routledge. Scott, J. 2001: Power. Cambridge: Polity Press.
Wartenberg, T.E. (ed.) 1992: Rethinking Power. Albany: State University of New York Press.