Table of Contents 

PC.GIF     NC.GIF    

Alfred de Grazia: Discovering National Elites





The premise of elite theory is that the great majority of human actions are not random, but are structured and directed. The structure and direction are provided by the roles, references, or identifications that all people possess. A man has one or more such roles, usually a number of them: family, occupation, social class, nation, neighborhood and political party are some examples of them. Those people who share a man’s role or identification together with him make up the group. The group, in turn, is structured - it consists of people who range from the greatly involved to the little involved and who can also be ranked from the 1) most influential (leaders) to the least influential 2) (followers). There is naturally some correlation between those who are heavily involved and those who are most influential. A high degree of activity usually means a higher than average power in a group.

Involvement and leadership resulting in power, are, of course, the criteria that interest the operator, and therefore he seeks out those groups that have important orientations towards matters usually expressed through the government actions. Within such relevant groups, he is interested in the involved and influential personnel. The major problem of target analysis becomes that of discovering who is heavily involved and influential in a politically relevant group. The influentials are, of course, his ultimate targets, but the search for the involved follows closely because the involved are defined by high psychic investment and high activity and hence the involved are 1) likely to be influential in contrast to the less involved and 2) are active in executing policies or preparing matters for determination or decision by the influentials (elite).

The several points just made can be elaborated and clarified in the several paragraphs to follow on the structure of society; the structure of involvement; and the distribution of leadership in groups.

II - 1 The structure of society. The relationship between individuals in a society which produce its structure can best be illustrated by considering a hypothetical model. Suppose that every man in a society has an equal chance of knowing every other person in the society, that is, that the only universe is the society and that the society is unstructured.* In that case, the probability that John Doe knows any given Richard Roe is determined by the number of Doe’s acquaintances, divided by the total population of the society. If this is limited to the adult population of the United States and if Doe has 500 friends, then Doe has 500 divided by 100,000,000, or one chance in 200,000 of knowing Roe. If Doe is very active in making friends and has 2,000 of them, then his chance of knowing Roe becomes one in 50,000.

Although the probabilities of mutual acquaintanceship are thus far very low, they increase by gigantic strides when a) the number of acquaintances is increased, and b) an intermediary, or third man, is introduced. Suppose Doe and Roe each have 500 acquaintances. What is the chance that through these acquaintances they will have an acquaintance, Q.E. Smith, in common? The chance is one in 200. If Doe and Roe each have 1,000 acquaintances, the chance becomes one in 50; and if each has 2,000, the chance is one in 12.5. Now suppose a fourth person, A. B. Jones, is considered, and the question is: if Doe knows 500, and Roe knows 500, what is the chance that each will know somebody who will know somebody else who will know the other? The chance now is "better" than unity. It is certain the Doe knows somebody who knows somebody who knows Rod in this fictional society.


*The theory here presented has been developed primarily by Dr. Ithiel de Sola Pool, Massachusetts Institute of Technology.

However, no society consists of random-moving, atomistic individuals. Instead, the society is structured itself as a group into involved vs. less-involved, the influential vs. less-influential, and many other functional components. Moreover, besides the society as a group, there are numerous other role or identification groups, such as the neighborhood, family, profession, club, or religion. The effect of these two facts on the theory of networks in the randomized universe is twofold: 1) some contact chances between persons are increased; 2) other are lessened.

Contacts and contact chances are increased by social separatism when the two individuals selected for testing share an identification which sets them apart in the universe under study. (Suppose they are both government employees.) The contact chance in the shared group increases inversely with the size of the group. (Suppose they are both Bureau of the Budget employees.) If the men belong to one shared group, they may belong to others, in which case also the chance goes up. (Suppose they are both Bureau of the Budget employees and live in Georgetown - a second common identification that would increase slightly their chances of encounter.)

Contact chances are lessened by social separatism when the differences between individuals’ identifications keep them moving along separate channels and through different networks. For example, a minor labor leader of the garment industry and a Department of Agriculture farm agent in Iowa would have few common groups and might even lack intermediaries. A merchant in Paris may know more merchants in London than he does workers who live a mile from his home.

II-2. The structure of involvement. The inner structure of any grouping is also not atomized but is composed of people who can be ordered according to their involvement in the grouping. Involvement can vary greatly. An internal measure is the extent of one’s spiritual and mental life that is bound up in a role; an external measure, more useful in studying elites, is the extent of his behavior or action-output that is oriented towards or done in the context0of a role. One might ask, for example, how active a man is in his professional group, or in his political community. A recent study of the political activity of the American population asked a representative sample a series of questions about participation in their political role.*

Responses were scored in the following manner:


Percentage of total sample qualifying Scoring points credited Total possible score for channel
Once or more in last four years7513
Three times or more471
Five times or more211
TYPE OF POLITICAL ACTIVITY: Discussing public issues with others
Discusses frequently and takes an equal share in the conversation2112
Discusses frequently and usually tries to convince others he is right61
TYPE OF POLITICAL ACTIVITY: Belonging to organizations that take stands on public issues
Belongs to one or more3112
Belongs to two or more71
TYPE OF POLITICAL ACTIVITY: Written or talked to congress-men or other public officials giving own opinion on public issue
One or more times in past year1312
Two or more times in past year71
TYPE OF POLITICAL ACTIVITY: Worked for election of a political candidate in last four years
 11 2 2
TYPE OF POLITICAL ACTIVITY: Contributed money to Party or candidate in last four years

* Julian L. Woodward and Elmo Roper "Political Activity of American Citizens," XLIV American Political Science Review (1950). 872-885.

The distribution of scores for the whole sample population on the scale of 0 to 12 follows:


ScorePercentage making score

Figure 1


When this distribution is plotted on a graph, the figure of a J-curve is obtained (see Figure 1). That is, there is a progressively smaller group involved in political activity as one moves towards greater participation. This phenomenon, here illustrated by one study, is found in all groups for which information is available.*

As the operator examines every new grouping that comes to his attention, he can expect that the J-curve will fit the distribution of activity therein. To the degree that it does, his problem of target-selection is greatly limited; most of his targets will fall among the high-scorers, as succeeding sections of this manual will show.

II-3. Distribution of political leadership. Whereas involvement is an index to who are leaders and who are symbol-receivers and symbol-conveyors, leadership is the direct indicator of influence and is the bulls-eye of the target. A J-curve fits generally the known distributions of leadership in groups, but it should be remembered that there is not a perfect overlap between those who score high on involvement and those who score high on influence. The leaders have a proved influence on the behavior of others within the group. The involved have not.


* Alfred de Grazia, "Political Activity and Leadership in Politics, Lad and Administration" (publication forthcoming).

An example of the J-curve of leadership is presented on the next page, based upon a study by Bryce Ryan of a random sample of 25 per cent of all farm operators residing in four townships of Iowa. Respondents were asked to name people to whom they would go for 1) advice, 2) organization, and 3) representation on five local problems of rural schools, farm taxes, scarcity of farms, land use, and local roads. The combined nominations of all counts give a kind of composite index of community leadership. The distribution of replies in presented below in Table 3 and plotted on Figure 2. It is noticeable that, as in the study of involvement, very few individuals emerge as the leaders. This is the expected situation regarding leadership in any group, and the operator, in moving into a new situation or into the examination of a new, relevant social grouping should again aim at the top influentials as the most profitable targets of his messages. The few who have a disproportionate amount of influence can do more for him than can the rank-and-file; they can act on his cues more decisively if convinced, or block them if hostile; they can pass on his cues with greater force.

Often, of course, the operator will discover that the elite is more hostile or difficult of persuasion than the rank-and-file. He will often have to direct messages at the lower leadership (and activity) echelons to make any impression whatsoever. But this retreat must be planned and self-conscious; he must realize he is forced to abandon the choicest targets because they are inaccessible, and not rationalize his retreat under illusory slogans about the "greater potency" of the rank-and-file members of a group.


71 persons(53%)were named 1 time
20 persons(15%)were named 2 times
11 persons(8%)were named 3 times
08 persons(6%)were named 4 times
05 persons(4%)were named 5 times
10 persons(7%)were named 6 to 10 times
07 persons(5%)were named 11 to 15 times
03 persons(2%)were named 16 or more times

* Graph *

Being practical does not mean being visionary or wishful. It means being as rational or calculating as possible under a given set of circumstances.

The so-called mass scarcely exists from the point of view of propaganda. Every component of society is structured. Within each element or group distinction between elite and mass is one of graduation of influence. Unless a person has some significance as an influence upon others or a target. The foregoing paragraphs have shown that in practically all groups, persons without influence are a very large fraction of the total who are identified with the group. What unfortunately and frequently happens in propaganda analysis is that the phrase "appealing to the masses" is used loosely to refer either to a kind of blunderbuss scattering of messages on the principle of "hope-it-hits-something," or to the targeting of a part of the population that is large an has potential leaders, or leaders not formally recognized as powerful in the institutional striker of the government. The inefficiency of the farmer case is self-evident; the leaders in the latter case can, and should, be found by appropriate target analysis of the elites, and their importance measured by the degree of their influence, not by the degree of their accessibility.

PC.GIF     NC.GIF    

 Table of Contents