LINDSAY ROGERS: A LIFE IN POLITICAL SCIENCE

Lindsay Rogers (1891–1970) was a respected scholar and public intellectual who was Burgess Professor of Public Law at Columbia University from 1929 to 1959. Rogers worked in journalism for a brief while: in 1912, reporting on the return of survivors of the Titanic as well as on William Jennings Bryan's “doings and utterances” at the Democratic National Convention.¹

Rogers earned his Ph.D. in 1915 from Johns Hopkins, one of the nation's major doctoral programs in political science (Somit and Tanenhaus, 58). As a scholar, he continued to publish for the public, writing commentaries, book reviews, and editorials in the Christian Science Monitor, Nation, and New York Times. He also involved himself in government service, serving as a consultant to the Senate Foreign Relations Committee (1952, 1956–1959) and as deputy administrator of the National Recovery Administration (1933) (Anonymous 1971).

In his early days, and consistent with the disciplinary traditions of the time, Rogers focused on political institutions and constitutional law. Before completing his doctoral dissertation on the postal powers of Congress, Rogers published a 1914 Yale Law Journal article on the federal government's powers over freedom of the press. In the 1910s and 1920s, he published other law review articles on interstate commerce, regulation of liquor, war powers, and the president's prerogative to sign bills when Congress was in recess. During that time, the American Political Science Review published descriptive and analytical pieces chronicling important developments in Congress; Rogers wrote seven of these between 1919 and 1924. His close observance of Congress was reflected in the third of his seven books, The American Senate (1926), in which he mounted a spirited defense of the Senate's practice of open and extended debate. His views of deliberation emphasized discussion and deliberation among leaders, an institutionally based argument that can be traced from The American Senate to The Pollsters.

THE POLLSTERS

Written with verve, The Pollsters: Public Opinion, Politics, and Democratic Leadership was reviewed widely and greeted with both acclaim and disdain. Although the book raised some typical issues associated with polling, such as sampling and question wording, its approach was such that the Review reviewer called it “an important contribution to the literature of contemporary political philosophy,” and the American Journal of Sociology review assayed Roger's concern as “calling the pollsters to heel—the heel of democratic theory.” In advising pollsters about their wayward path, Rogers wrote before polls and social surveys had achieved an authoritative status (Ginsberg 1986; Herbst 1993; Lee 2002). Critics argued that he did not understand the mechanics of polls, but this contention (which I consider to be incorrect) was off point. George Gallup had promoted polling as a central mechanism for democratic control, and Rogers took aim at Gallup's contentions. With Gallup as his main focus, Rogers also addressed academic survey researchers who contended that polls could reveal mass opinion and thus serve democracy.

In Public Opinion in a Democracy, Gallup (1939, 15) claimed that polls would restore the idea of the New England town meeting. With newspapers and radios widely available to “conduct the debate on national issues,” citizens had access to information, and, through polls, their views could be discerned. “[T]he nation is literally in one great room…[T]hrough the process of the sampling referendum, the people having heard the debate on both sides of every issue, can express their will.” In The Pulse of Democracy, Gallup and Rae 1940, presented polls as a breakthrough that might change the operation of democracy. With an air of near utopianism, they heralded polls as part of the march of progress, a technology that would reshape civic relations (14):

Much of The Pulse of Democracy described how polls work, but at times Gallup and Rae turned to the democratic implications of polling. On the one hand, polls were presented as an aid to representatives: “The continuous studies of public opinion will merely supplement, not destroy, the work of representatives” (Gallup and Rae 1940, 266). However, Gallup and Rae (259) also implied that there should be some sort of “government by public opinion.” For instance, they wrote, “Those against this sort of government arrangement are suspicious of the public, much like dictators who believe that “the common people, because of their numbers, their lack of training, their stupidity and gullibility, must be kept as far as possible from the elite whose task it is to formulate laws for the mass blindly to obey.”

Responding to these sorts of arguments, Rogers made it clear that he supported democracy, not dictatorship. His core complaint against Gallup's vision of polling as enhancing democracy was that Gallup did not appreciate the United States' constitutional design nor its actually existing democracy. Rogers concentrated on four issues: (1) the institutional incompatibility between rule by polls and the American governmental system, (2) the importance of minority rights, (3) the nature of discussion and deliberation and the inability of polls to create an institutional space for their development, and (4) the importance of leadership in educating the public. His argument regarding institutional incompatibility of poll-driven politics and American constitutional design provided a rubric for the other three concerns.

First, Rogers contended that pollsters did not apprehend the American political system and had not explicated their premises regarding the polity. Supporters of this sort of rule did not explain “the nature of the political system in which public opinion should be the ruler” (Rogers 1949a, 12) nor acknowledge the founders' intentions. Polls are not a technological fix for the practical problems with direct democracy, for they cannot grapple with the need to craft compromises. “True it is,” Rogers averred, “that representative bodies often yield to selfish pressures and sometimes enthrone privilege, but representation permits a greater measure of compromise between clashing opinions and interests than when decisions are made by the mass of citizens” (69). Thus, a scheme of rule by polls is to be rejected not only because direct democracy is impractical but also because it is inconsistent with the nature of republican governance.

Promoting polls as quasi-referenda through which citizens can “vote” on policies likewise overlooks the purposes of representative government. Quoting Harold Laski, Rogers held that “the whole theory of a referendum misconceives what an electorate is for. It forms a view on a general web of political tendency…Direct government, in short, is not the same thing as self-government” (Rogers, 79). Criticizing referenda does not mean that one sees the electorate as unintelligent. Instead, Rogers maintained, the generally low turnout for referenda is a signal that “the electorate is so intelligent that it is not keen on voting on measures that it imperfectly comprehends, and a large percentage of it refuses to do so” (Rogers, 79). As governments deal with increasingly complex and technical matters, direct democratic institutional structures are increasingly unsuitable.

Second, Rogers argued that pollsters assume that when polls reveal majority preferences in response to particular options offered, then the majority's views should be adopted in public policy. However, the Constitution was not intended to deliver “majority rule pure and simple” (Rogers 1949a, 88). Instead, it provides for the protection of minority rights.²

In addition, under federalism, national majorities are frequently irrelevant. The U.S. system allows for election by plurality, thus showing that majority rule is neither authoritative nor automatic. Under the electoral college, the winner of the presidential election may receive fewer votes than losing candidates. Small states' parity with large states in the Senate means that senators representing a minority of the country may have a majority of legislative seats. Furthermore, judges can overturn laws, even when majorities agree with these, and amending the Constitution can be done only with supermajorities. Rogers offered these “elementary” civics lessons because “the pollsters ignore [U.S. institutions'] carefully intended complications when they talk glibly of the entire country again having a town meeting” (Rogers, 92).

Third, Rogers argued that an emphasis on polls as a democratic mechanism revealed a misunderstanding of the nature of deliberation. Polls were incapable of creating an institutional space in which deliberation is possible. Deliberation requires discussion, and discussion takes time. Thus, when Gallup called citizens' initial responses to politicians' speeches “sampling referenda,” implying that these immediate responses constituted public opinion, Gallup overlooked the need for time to consider ideas. Furthermore, deliberation must take place within institutional locales. Rogers viewed Gallup's town meeting model as further flawed by its inattention to openness and structure, dual requisites of productive discussion that could be built into institutional norms and rules. In the town meeting, “there could be argument for and against a proposition at the same time in the presence of all who were to vote, and the proposition under debate could be reframed to meet objections that the discussion had brought out” (Rogers 1949a, 71). Majorities often evinced a willingness to meet dissenters half-way. “Usually objections could be met, and the decisions reached could have a wide basis of assent” (Rogers, 71–72). This fluidity in the deliberative process was enforced by the moderator, who would assist in “reframing the question so that it would represent the ‘sense of the meeting’” (Rogers, 72). Polls, with their limited alternatives, allowed no reconsideration of the question at hand.

Although town meetings, properly understood, could allow for discussion more than polls, Rogers contended that representative institutions promote deliberation among public officials which could also provoke citizen discussion. “The pollsters overlook the really vital role of representative assemblies in focusing attention on political issues” (Rogers 1949a, 34). As Rogers discussed in The American Senate, the difficulty of ending senatorial debate could create a crisis and stimulate public interest. “Only the American Senate can act as a ‘teaching apparatus’ or bring about ‘a catastrophe’ of obstruction that will make politics interesting” (Rogers 1926, 249–50). The Senate could allow for open and full consideration of ideas and could overcome the executive's inherent advantage of promoting its views in the media. “[O]nly the Senate can counteract the effects of presidential publicity which lures the country into accepting certain fixed ideas of the executive, and which vouchsafes only favorable information concerning administrative activities” (Rogers 1926, 241). Full and open debate could allow for consideration of ideas beyond those framed by party leaders, including the president.

Elites not only are checked by other elites, but also promote fulsome discussion.

Rogers's view of deliberation did not start and end with elites. Although he agreed with Burke that representatives needed to exert a certain degree of independent judgment, he complained that Burke was “indifferent to the value of discussion, which is indispensable in any community that seeks to govern itself rather than to be ruled” (Rogers, 1949a, 217–18). “Prolonged debates in the Senate on pending legislation or treaties,” whether or we now judge the minority correct or not, have “been extremely interesting in giving the country opportunities to form and express opinions” (Rogers 1926, 246). For Rogers, Senate debate in which minorities could enjoy “freedom of debate, or, as the critics would say, unrestrained, garrulity” (1926, 161) paradoxically protected democracy. The founders sought to protect minority rights and deliberation. Both could be promoted in the Senate; polls advanced neither.

Fourth, Rogers viewed leadership as an essential element of democratic systems. “Polls could provide “data that the statesmen should examine and see whether they suggest opportunities for resistance, correction, education” (Rogers 1949a, 93). By knowing what citizens thought and understood, public officials could respond to citizens' needs. However, this did not mean that leaders should simply do what the public wanted; that was unwise and inconsistent with core institutional principles.

Polls bring with them the danger that elected officials will not provide leadership. “[A] too-ready method of ascertaining opinion might make government weak rather than strong, foolish, rather than wise” (92).

If without leadership, public opinion can be uninformed and latent. In an essay written before the United States entered World War II, Rogers noted that democracies differed from “gangster states,” in their need to attend to public opinion (Rogers 1941, 68–69). Effective democratic leadership surely does not control the population, but might educate. “Every people, it is cynically and untruthfully said, have as good a government as they deserve. But a government may have no better than an electorate than it deserves because it fails to give leadership” (Rogers 1949a, 234). Democracy's political institutions require taking public opinion into account, while seeking to persuade.

ROGERS AND AMERICAN POLITICAL SCIENCE

Rogers's arguments were attuned to institutional arrangements and processes, contemporary and historical dynamics, and normative questions, and were aimed beyond the academy to the attentive public. Although his critique of polling was well read in its time, his views have virtually disappeared from the canon. I see this absence as revealing three political and institutional aspects of disciplinary history: (1) the close connection between academic survey research and commercial polling operations, which was strained but ultimately enhanced by the 1948 polling debacle; (2) the advent of the behavioral revolution, which stressed hypothesis-testing, quantification, and empiricism; and (3) the growth of institutionalized data-gathering organizations that provide these data to quantitative researchers and orient graduate training and scholarly networks and work.

At the time of the 1948 election, commercial pollsters had a good deal of expertise (much of it in market research) and had built close relationships with scholarly surveyors. In the 1940s, pollsters had the capacity to conduct national polls and academics often contracted them to do so. During World War II, leaders in academic survey research either conducted large-scale studies for government agencies or consulted with surveyors in government service. “The few knowledgeable people with experience in commercial, academic, or governmental survey work were in great demand (to help each other, especially) and were linked in webs of frequent contact” (Converse 1987, 163). Even as survey research centers developed after the war, academics continued cooperative research with pollsters. When polling failed most spectacularly—all the national pollsters predicted a certain Dewey win—academics explained.

Survey researchers and pollsters had a shared interest in regaining legitimacy for polls. Concerns about the delegitimization of polls certainly did not prevent academics from raising criticisms (Converse 1987; Igo forthcoming; Meier and Saunders 1949; Mosteller et al., 1949), but a loose group of interdisciplinary scholars assessed the polls and defended the enterprise. Merton and Hatt (1949) studied whether the prediction hurt media elites' views of marketing and social science research. Academics worked closely with the major national polling firms headed by Crossley, Gallup, and Roper. The day after the election, Roper announced, “[W]e are inviting a group of social scientists to come in and examine all of the data on which our conclusions were based, hoping they may find new and important data that will throw light on a situation that has thus far confounded everyone.”³

This group, a committee of the Social Science Research Council (SSRC), the Committee on Analysis of Pre-election Polls and Forecasts, asked for and received complete cooperation from the polling firms. With financial support from the Carnegie Corporation and the Rockefeller Foundation, the committee report was released in late 1948 and a volume containing both the lengthy committee staff report and the previously released committee report was published in mid-1949. In the process of preparing the SSRC staff report, the pollsters were given access to draft materials and were allowed to propose changes. By January 12, 1949, each SSRC committee staff member had been given “a copy of comments by Crossley, Gallup, and Roper on the first draft of their report to take into account in making his revision.”⁴

Expressing concern regarding polls' perceived legitimacy, Pendleton Herring (then head of the SSRC and later, in 1953, president of the American Political Science Association) wrote in the foreword to the SSRC volume that “Appointment of the committee rested upon the judgment that extended controversy regarding the pre-election polls among lay and professional groups might have extensive and unjustified repercussions upon all types of opinion and attitude studies and perhaps upon social science research generally” (Mosteller et al. 1949, vii.). According to the SSRC committee report (Mosteller et al., 294–95), Crossley, Gallup and Roper had done “pioneering work” for which they should be commended, but a new stage of cooperation was now needed: “Public opinion polling is only in its infancy and if soundly developed and properly utilized it will be of great assistance in understanding political and social behavior. Sound development requires the cooperation of the polling organizations and social scientists.” Academic survey researchers hoped that the polling failure of 1948 would not undermine the status of social science in general and the useful tool of surveys in particular.

Scholars and pollsters discussed the committee report and prospects for cooperative research in other venues, including a February 1949 conference at the State University of Iowa. Professor Norman Meier, a major organizer of the meeting, wrote to Roper in November 1948 that, “I'm sure it requires no urging of the point that a public relations effort is immediately in order, or the entire survey tool will suffer serious impairment in the public mind. We believe that the proper place for this is on a University campus.”⁵

In Iowa, technical criticisms of the pollsters were made, but scholars were clear that they came to sustain polling, not to bury it. For instance, Paul Lazarsfeld of Columbia held that the meeting's goal was “mutual exchange of ideas to help along this development of public opinion work from an accidental freak to a very legitimate and well-organized part of social science in general” (Meier and Saunders 1949, 154).

In promoting surveys, some scholars at the Iowa conference championed polling in terms much like Gallup. Samuel Stouffer of Harvard called polls “an instrument a democracy” (214) and Clyde W. Hart of the University of Chicago held that future social science research surveys should “provide bases for the choices” made by society “acting through its representatives” (310). Scholars' normative claims, neither fully argued or debated, took the form of unexamined common wisdom, while the sorts of critiques Rogers made were left unaddressed. As academic survey researchers and pollsters voiced similar visions of democracy, the problems they identified in the polls were methodological, involving sampling, predicting turnout and the vote choice of undecided voters, and the like.⁶

To restore the legitimacy and viability of survey research, academic researchers and pollsters focused on correcting poor technique. To be sure, the pollsters sometimes found the academics vexing. However, by and large, polling's failure cemented previously existing relationships between academic survey researchers and commercial pollsters and prompted the scholars to defend the enterprise of polling.

Moreover, when Rogers published The Pollsters in 1949, American political science was on the verge of enormous change. Rogers questioned the premises underlying the pollsters' democratic theory and called for attention to institutional requisites of democratic deliberation. Within the discipline, a movement to make research propositions testable with quantitative data and to sequester empirical analysis from normative assessments was to move political science away from the traditions Rogers epitomized. Although Rogers's approach bore the marks of a political science enmeshed in political thought, history, and institutional studies, the discipline had already taken steps toward a positivist approach. Scientism, a movement to promote a “new science” of politics within political science, had begun in the 1920s. Charles Merriam called on political scientists to become more scientific, by which he meant generating hypotheses and evidence that could be used to prove or disprove those. Others in this camp called for developing general laws of politics and argued that political science research should be value-free (Farr 1995; Gunnell 2005; Somit and Taanenhaus 1967). Research became more quantitative as statistical techniques were increasingly, if slowly, adopted. “From 1921–1930, four articles and notes employing fairly sophisticated statistical techniques had appeared in the [American Political Science] Review. During the next decade the journal ran 16 such items.”

Scientism gave way to behavioralism in the 1950s, as “The behavioral approach grew from the deviant and unpopular views of a minor sect into a major influence” (Dahl 1961, 766). Somit and Tanenahus (1967, 177–79) identified a number of core characteristics of behavioralism, including the aim of prediction; observation of behavior and what people say; an emphasis on quantifiable data and findings; research that is “theory oriented and theory directed” but not focused around normative concerns from the history of political thought, such as freedom or equality; methodological self-consciousness; and the adoption of methods from other social sciences. During World War II, government officials could see the usefulness of survey research and with the advent of the Cold War, “The very idea of a behavioral political science gained considerable momentum” (Farr 1995, 211; see also Ball 1993). Survey data and its analysis became a more prominent part of the discipline; they were used in 3% of political science journal articles in 1949–1950; 11% in 1959–1960; and 19% in 1964–1965 (Converse 1987, 403).

With this shift came an increasing separation between empirical analysis and normative concerns. Only a dozen years after Rogers brought democratic theory to bear on the polling enterprise, Dahl contended that:

Although Lazarsfeld (1957, 39, 49) noted the “struggles” between “empirical social science” and the “classical tradition” and Rogers' “valid objection” regarding pollsters' inattention to the place of public opinion in democracy, by this point the struggle seemed to be over. Far fewer scholars were equipped both to discover political dynamics and to judge them. Behavioral political science could only study facts and the political philosopher lacked the technical knowledge to appraise what empirical research found. In this context, the concerns of Rogers, previously the traditional concerns of the discipline, had little place.

Another shift in the discipline that marginalized Rogers was the development of institutionalized survey operations and the education of scholars who would use these data. Rogers and other traditional political scientists were classic institutionalists, but they were not institution-builders, and their ideas lacked the organizational basis that the survey researchers created. The earliest survey center, Columbia University's Bureau of Applied Social Research, generated a sense of excitement among students who “believed it was about to remake social science, if not the world” (Lipset, quoted by Converse 1987, 267). Cohorts of graduate students worked in the centers that scholarly entrepreneurs built, where they learned grant-writing, administrative, and methodological skills, and went on to promote this approach among their students and in publications and other professional venues. Today the University of Michigan's Institute for Social Research (which includes ICPSR and the Survey Research Center) and the University of Chicago's National Opinion Research Center exert a significant impact on the work of political scientists and other social scientists. In gathering data and serving as common repositories from which scholars can draw, survey research has become more accessible, comparable and transparent, but the availability of data has also dissuaded scholars from seeking alternative materials and has shaped disciplinary knowledge.

In this new environment, the classics of political philosophy lost their secure place in the curriculum (Ricci 1984). In his time, Rogers predictably took a contrarian position: “On undergraduate instruction in political science…Save for political theory, the less of it the better” (in Fessler et al. 1951, 1022). Graduate programs in political science have devoted more time to conveying sophisticated statistical methods and new empirical theories, and these shifts in attention have meant that a number of subjects, including “political theory and philosophy … have all become defensive, peripheral, and secondary subject matters. As a result a large part of the political science tradition is no longer being transmitted effectively to younger generations” (Almond 1990, 50; for an example, see Fried 2002). Rogers's work stood against this self-imposed schism. Behavioralists and pollsters, he believed, overstated their contributions and were far too comfortable with their limitations and jargon.⁷

As “Rogers spoke for the traditional normative concern of political theory” (Gunn 1995, 106),” he was forgotten as political theory has become sequestered from “the wider interdisciplinary field” (Gunnell 1993, 2) of political science.

In looking back at this traditional political scientist, today's political scientists gain a sense of where the discipline once was and the elements they could seek to recover—the abilities to converse across subfields and specialties, to contribute to public activities and debates, and to integrate the normative and empirical in studies of politics and political life. Rogers's critiques of polling and survey research, though still worthwhile, were forgotten because of political and institutional developments in political science, including the role of alliances between commercial pollsters and academic survey researchers in defending, promoting, and improving polling; the successful efforts of a committed group to cast the field in a behavioral direction; and the building of institutions that have provided readily available data to analyze, shaped what it means to do empirical research, and socialized new generations of graduate students and young scholars.