I was very much impressed by Andrew Stark's profound structural interpretation of the question of “Why Political Scientists aren't Public Intellectuals” (PS&P, September 2002). I do not want to rehearse or respond to his particular argument which I find quite persuasive. But, it seems to me, there is more to the story, and Stark himself provides the clue in a single sentence: “Yet if the problem is structural, the solution must be the individual.” It is this thought I want to address.

In a strictly theoretical and methodological perspective, the Columbia studies of personal influence—conducted in the 1940s and early 1950s—are today of largely historical interest as particularly self-conscious and sophisticated examples of social-scientific discovery. Yet, there are indications that these studies are once more coming to scholarly attention and their long eclipse, so symptomatic of discontinuity in social-scientific research, may be coming to an end (Scheingold, 1973). There is a growing interest in describing and explaining electoral and related patterns of behavior in terms of the “social networks” to which people belong. The contribution of the Columbia studies to research on the effect of social networks in voting behavior and public affairs seems therefore worthy of retrospection.

Project RADIR (“Revolution and the Development of International Relations”), conducted at Stanford University’s Hoover Institution in the years following the Second World War, was probably the first large-scale, quantitatively-based empirical research enterprise in contemporary political science. Inspired Dy the theories of Harold D. Lasswell about world revolutionary developments in the modern era, the project was in some respects the successor to work done in wartime Washington at the Library of Congress by Lasswell’s Experimental Division for the Study of Wartime Communications, and by the Organizations and Propaganda Analysis Section that Lasswell had helped to set up in the Special War Policies Unit of the Department of Justice. When published in the early fifties, the Hoover Studies, as they came to be called, were noted not only for their substantive findings but also for their theoretical suggestiveness and their methodological boldness of overall design. Most pre-war empirical, quantitative work in political science had been hyperfactual, but the Hoover Studies represented a genuine break with that tradition. Some early reviewers, if not ignorant of quantitative analysis, had trouble with the project’s theoretical formulations, and others were methodologically innocent.^ The methodological novelty of the Hoover Studies consisted in the demonstration that quantitative data could be generated out of the verbal contents found in newspapers and conventional biographies. Although some political scientists had made use of aggregate data reported in the Census, government budgets, legislative roll-calls and electoral statistics, the notion that one can “create” one’s own massive body of data was still relatively new at mid-century. Public opinion polls had come into existence only in the middle thirties and had yet to be exploited fully for scientific purposes. It was in this research context that the Hoover Studies were published.

Hyperbole is the trademark of writing on technology and its impact on human affairs. My indiscretion in using a mysterious or sensational title may be a case of epidemic infection. My intent, of course, is not to be mysterious or sensational. Rather, it is irony that I have in mind. Although technology is usually the manifest target, it is a fear of the politics of civility rather than of technology that pervades much contemporary writing on the social impact of technology. To irony must be added paradox. The politics of civility is feared not only by those who dislike technology, but also by those who see in technology the tool for curing the social problems that stem from its impact.

The meaningful participation of the electorate in the selection of public officials is, in many respects, a defining characteristic of democratic politics. It is, of course, not the only characteristic; and it is, moreover, a characteristic that is not independent of other social phenomena that shape the process of leadership selection. In particular, social background broadly conceived as including status at birth and educational or occupational achievement, as well as the intervention of a small politically active stratum in the selection process affect the decision as to who gains control of government. Even in the most democratic society, the electorate does not choose from among all its members. It chooses from among a pool of eligibles disproportionately drawn from the higher social-status groups in society; the availability of these eligibles as political candidates very much depends on the intramural maneuverings within the active stratum by which some are recruited to and others eliminated from candidacy.

Scholars interested in theorizing about political representation in terms relevant to democratic governance in mid-twentieth century America find themselves in a quandary. We are surrounded by functioning representative institutions, or at least by institutions formally described as representative. Individuals who presumably “represent” other citizens govern some 90 thousand different political units—they sit on school and special district boards, on township and city councils, on county directorates, on state and national assemblies, and so forth. But the flourishing activity of representation has not yet been matched by a sustained effort to explain what makes the representational process tick.

Despite the proliferation of representative governments over the past century, theory about representation has not moved much beyond the eighteenth-century formulation of Edmund Burke. Certainly most empirical research has been cast in the Burkean vocabulary. But in order to think in novel ways about representative government in the twentieth-century, we may have to admit that present conceptions guiding empirical research are obsolete. This in turn means that the spell of Burke's vocabulary over scientific work on representation must be broken.

To look afresh at representation, it is necessary to be sensitive to the unresolved tension between the two main currents of contemporary thinking about representational relationships. On the one hand, representation is treated as a relationship between any one individual, the represented, and another individual, the representative—an inter-individual relationship. On the other hand, representatives are treated as a group, brought together in the assembly, to represent the interest of the community as a whole—an inter-group relationship. Most theoretical formulations since Burke are cast in one or the other of these terms.