Some of you might want to do your own studies of national or local power structures. This document provides both the general rationales and the specific steps and sources for doing such studies. It may sound too simple when I say this, but power structure studies are really a combination of network analysis and content analysis. Both methods are easy enough to understand in principle, and they are very respectable in the social sciences. They can be used for large and complex studies by employing software that will be mentioned as this document unfolds. If you start simple with a small project and do the first steps by hand, you will have a good intuitive understanding of what is really going on if and when you advance to huge databases and fancy software.
A network analysis traces out the people and organizations that make up the power structure, and then figures out how they connect to and influence government. Content analysis is the term for the systematic study of the power structure's ideologies, policies, and plans, which are learned about through the careful study of documents such as the texts for speeches, policy statements by organizations, and drafts of legislation.Before I begin, I want to stress that the methodological approach outlined in this document makes it possible to discover any concentration or configuration of power. Researchers of any theoretical persuasion can use it because it is not biased for or against any given theory. It contains only one assumption: there is a power structure of some kind or another, no matter how weak or fragmented, in any large-scale society or social group. The method can discover that power is highly concentrated or more dispersed, depending on the degree of difference between rival networks on the power indicators. It can show that some groups or classes have power in one arena, some in another arena. It can reveal changes in a power structure over time by changes in the power indicators. In the United States it usually has led to corporate power structures at the national level and landowner/developer/real estate power structures at the local level, but the results could be different for other countries and they are sometimes different for cities in the United States.
The empirical study of power begins with a search for connections among the people and organizations that are thought to constitute the powerful group or class. This procedure is called membership network analysis. It starts with a study of people and all the organizations they belong to. Or conversely, you could say the study starts with a list of organizations that includes all of their members. It leads to the same result -- that's why it's called membership network analysis. The results of a membership network analysis are usually presented in the form of a matrix, as shown in Table 1. The people are listed from the top to bottom and the organizations are arrayed from left to right. The "cells" or boxes created by the intersection of a person and organization are filled with "relational" information such as "member," "director," "owner," or "financial donor." The attitudes the person has toward any given organization in the matrix also can be included, such as "supporter" or "opponent." Maybe think of attitudinal information as a "psychological" relation to the organization. The information used in filling the cells of the matrix is obtained in a variety of ways described later in this document.
The information contained in the matrix is used to create both organizational and interpersonal networks. Figure 1 displays an organizational network based on the overlapping members in Table 1. It shows that organization A is at the "center" of the network due to the largest number of connections among these individuals. Figure 2 shows the interpersonal network that emerges from Table 1. Note that no one person is at the center of this interpersonal network, even though the organizational network had a center. That may seem counterintuitive, but it is a good example of the fact that these are two different networks, not just some mirror image of each other. Note also that one person is an "isolate," with no connections. That's the situation for 99% of Americans when it comes to the schools, clubs, policy groups, and companies of the upper class and the power elite in the United States.
|Table 1: Hypothetical Membership Network|
|Org. A||Org. B||Org. C||Org. D|
|NOTE: Individual 5 is an "isolate" with no connections.|
|Figure 1: Hypothetical Organizational Network Created By Overlapping Members|
|Figure 2: Hypothetical Interpersonal Network Created By |
Large and complicated membership networks can be analyzed using computer software based on sophisticated mathematical techniques, such as graph theory, matrix algebra, and boolean algebra (which can detect "hierarchies" or "levels" in a large complex network). Most of this software, except the boolean program, is available in UCINET, a DOS program that is menu driven and can be set up to run under Windows. The latest version of UCINET for Windows can be obtained by students for a reasonable price (including a 60-day free trial) from Analytic Technologies, Inc.
Once the membership networks have been established, there are many other types of links that might be analyzed, such as kinship ties or flows of information between organizations. One of the most important of these other types of links concerns the size and direction of money flows in the network. In theory, money flows are another kind of relationship between people or institutions, but in practice it is a good idea to consider them separately because they are socially distinct in most people's minds. And of course they are usually super important in understanding any modern-day power structure. There are four kinds of money flows:
The first finding from network analyses using relational data or money flows is whether or not the hypothetical group or class that is being searched for actually exists as a social reality. If no connections among corporations are found, for example, then it makes no sense to speak of a "corporate community." If there are few or no overlapping memberships among exclusive social clubs in different cities, then it is less likely that there is a nationwide "upper class." If there are no money flows from wealthy people to foundations or from foundations to policy discussion groups, then there is very little basis for talking about a "policy-planning network."
The second finding from membership network analysis concerns various characteristics of organizational and interpersonal networks, such as their density and the existence of central points or subgroups. Some parts in a social network may have more interconnections than others, for example, or some types of businesses might be more central in the corporate community, or there might be moderately conservative and extremely conservative subgroups within the overall policy-planning network.
Since reference was just made to social clubs and the possible existence of a nationwide upper class, this is a good place to add that upper-class organizations, such as private schools and social clubs, can be understood as "indicators" of upper-class standing. That is, if person X went to any the private schools that are listed in Appendix A at the end of this document, or is a member of any of the social clubs listed in Appendix A, then she or he is inferred to be a member of the social upper class. That's a fairly safe inference, because previous studies have shown that members of these clubs and schools are from high-status families and belong to other social organizations that are known to be upper-class organizations.
There are two things to keep in mind in employing these indicators. First, some of the people from these schools or in these clubs may not "really" be members of the upper class. For example, they may be scholarship students at the schools or honorary members of the club because they hold some elected office. These mistakes are called "false positives," but they are not a problem for us in studies that use reasonably large sample sizes to gain a general picture of what percent of the people on the list -- for example, a long list of corporate directors -- are also members of the upper class. It's also the case that there are always a few "false negatives" on the lists we are studying, which means people that are part of the upper class, but don't mention their prep schools or social clubs in publicly available documents. So there is a tendency for the false positives and false negatives to cancel each other out.
The second thing to keep in mind in using the upper-class indicators in Appendix A is that some of these schools and clubs may have gone out of existence or faded in status. In that regard, the list could be a little out of date even though there is considerable stability in the social institutions of the upper class over generations. At the least, it is a good idea to be on the lookout for more recent schools and clubs that aren't on this list.
Once a membership network is constructed, it is possible to take the next necessary step in the study of social power, an analysis of the ideology and policy preferences of the group or class under scrutiny. This is done by studying the written "output" of strategically located people or organizations in the network, that is, texts of speeches, policy statements, campaign literature, and proposed legislation.
As mentioned earlier, such studies are called content analysis in the social sciences. Content analyses are not always done formally, but content analysis is what investigators are doing whenever they infer on the basis of a speech or policy statement that a person or organization has specific values or policy preferences. That is, many content analyses are informal and intuitive, based on implicit categories that exist in the culture. To insure against personal biases, however, an objective and systematic content analysis is far more useful.
In the past, a systematic content analysis always began with the creation of carefully defined categories that relate to the attitude or issue being studied. Categories can be constructed, for example, to determine a person or organization's stance toward corporations or labor unions. Once the categories are developed, relevant texts are studied to determine the frequency and intensity of elements that fit into one or more of the categories. Then the various frequencies are analyzed by calculating averages or percentages. Finally, the averages or percentages for two or more groups are compared. Then, too, researchers can do studies to determine if there are linkages -- sometimes called contingencies or correlations -- between two or more content categories. Proceeding in this way, the general "world view" of the power group could in theory be established in a step-by-step fashion.
Thanks to the advent of personal computers, computer-assisted content analyses now can be done without a set of predefined categories. Word searches of computerized texts give instant frequency comparisons. Texts also can be compared for telltale phrases that might reveal a connection between a private policy group and governmental legislation. There is also software to determine what concepts or phrases are interconnected in documents, a technique known as "semantic network analysis." One of the best starting points for content analysis using search software and the Internet is Carl Roberts, ed., Text Analysis for the Social Sciences (Mahwah, NJ: Erlbaum, 1997). See also the excellent papers by Roberto Franzosi, "Computer-Assisted Coding of Textual Data: An Application of Semantic Grammars," Sociological Methods and Research, 1990, 19: 225-257, and "Narrative as Data: linguistic and Statistical Tools for the Quantitative Study of Historical Events," International Review of Social History, 1998, 43, supplement 6: 81-104.
Many different sources of information are employed to create membership networks or find textual material for content analysis. First, a wide variety of biographical reference volumes, magazines, and newspapers are used to gather relational information. They include the eighteen different Who's Who's published by Marquis, which contain over 800,000 names. They are online, but you have to subscribe, so try for a one-week free trial or see if the fundraising unit on your university campus will let you use its computer (it has this info at its fingertips because it is extremely useful for money-raising purposes). Then there are the hundreds of magazines that are available on the Internet through your university library information system, along with the many newspapers that are available through the same avenue (often with different names on different campuses.) Then, too, Standard and Poor's Register of Corporations and Directors, the Foundation Directory, The Foundation Grants Index Online (mostly available through public libraries and community foundations), Federal Elections Commission reports on campaign donations, and the annual reports of many organizations can be consulted. Many of these sources are now available online. Simplest of all, just Google every person and organization, and build from there. The Web site Littlesis.org provides much of this information as well.
For an exceptional web site that provides an excellent overview of research on power and links to all of these sources, see Who Rules?: An Internet Guide to Power Structure Research, created by sociologist Val Burris at the University of Oregon. Burris' site provides direct links to internet sources of data that make it possible to study, among many topics: (1) the social backgrounds, economic interests, and interpersonal connections of individual members of the powerful group or class; (2) the internal power structures of major corporations and the political activities in which they are engaged; (3) the flow of money from corporations and wealthy business owners to political candidates and parties; (4) the role of special interests in lobbying congress and shaping legislation; and (5) the role of foundations, think tanks, and business associations in creating public policy. It also contains an excellent discussion of network analysis, a guide to library resources, and a list of suggested readings.
This site, WhoRulesAmerica.net, also hosts a list of Links & Further Reading.
Information for the study of power at the city level is not always as readily available via the Internet as it is for the national level, especially for small towns. For city studies, the following steps can be taken to assemble relevant information in a relatively quick fashion; the starting point is at the reference desk of a library.
It is also possible to study power at the city level through an interview technique called the "reputational method." With this method, the evidence for the power of a person or group is based on a reputation for being powerful, as determined by a series of interviews. It also can be used as a supplement or cross-check to the series of steps outlined above.
A reputational search begins in one of two ways: first, nominations can be obtained from a cross-section of observers who are thought to be knowledgeable about the powerful on the basis of their occupational roles (e.g., reporter, administrator, fundraiser). Or, the people found through steps 1 through 4 who sit on several corporate or nonprofit boards of directors can be used as a starting point. Either way, the people on the list are then interviewed and asked for their nominations as to the most powerful people in the locale being studied, as well as for their opinions regarding the power of the other people on the original list. It is also very important to ask them about their relations to each of the people they mention, and to ask if there are any organizations to which many, or all, of these people belong. This gives you the kind of information (connections) that are useful for a network analysis. Any new nominees are then interviewed and asked for their opinions.
The process ends, usually within two or three rounds, when the same names keep coming up and no new names are added to the list. For brief studies, it can be decided beforehand to do only one or two rounds of interviewing. The method has a further advantage: the people being interviewed can be asked other questions, such as: what are the major issues in the city? how is policy made in this city? what was your role in one or more of the major issues? In fact, it would be a shame, and not very interesting, to just end up with a list of people with reputations for power. You want to know their connections, their policy involvements, and their opinions on the major issues.
The reputational method works best at the community or city level, where it is less expensive and time-consuming to apply than at the state or national level. It is especially valuable for small towns where very little printed information is available. However, the method has been used with good results in two studies of national power in the United States, and in studies of Australia and Norway. The two American studies are by Floyd Hunter, Top Leadership USA (Chapel Hill: University of North Carolina Press, 1957) and Gwen Moore, "The Structure of a National Elite Network," American Sociological Review, 44 (1979): 673-92. The Australian study is by John Higley, Desley Deacon, and Don Smart, Elites in Australia (London: Routledge and Kegan Paul, 1979). The Norwegian study is by John Higley, G. Lowell Field, and Knut Groholt, Elite Structure and Ideology (New York: Columbia University Press, 1976). For a summary of reputational studies at the local level in the United States, see G. William Domhoff, Who Really Rules? New Haven and Community Power Re-Examined (New Brunswick, NJ: Transaction Books, 1978). That final book may seem out of date, but the sad truth is that there haven't been many detailed studies of local power structures in the past few decades. For one recent exception that includes new references, see Richard Gendron and G. William Domhoff, The Leftmost City: Power and Progressive Politics in Santa Cruz (Westview, 2009). See also the section on Power at the Local Level on this site.
Now that membership network analysis and content analysis have been explained, and sources of information have been outlined, it is also possible to provide a generic definition of a "power structure," which might be useful as an introduction to any study that you do, especially if it is for a college course. A power structure is the network of people and institutions in the city or nation under study that stands at the top on the power indicators it was possible to utilize in the study (the power indicators are "Who Benefits?", "Who Governs?", "Who Wins?", and "A Reputation for Power"). See the document in this site entitled "Basics of Studying Power" for a discussion of these power indicators.
Coed and Boys' Schools
Country Day School
Westminster (Atlanta, GA)
Country and Men's Clubs
Burlingame Country Club
Chagrin Valley Hunt
St. Louis County Club
Woodhill Country Club
Brimmer's and May
Convent of the Sacred Heart
Lake Forest Country Day
Louise S. McGehee
Mt. Vernon Seminary
St. Agnes Episcopal
St. Mary's Hall
Mt. Vernon Club
Society of Colonial Dames
All content ©2020 G. William Domhoff, unless otherwise noted. Unauthorized reproduction prohibited.