“I remember all these mathematical formulas,” Robert Caro recalls of his coursework as a Neiman Fellow at Harvard University. “Equations. Equations. They kept getting longer and longer. All of a sudden I looked at my notebook and said, ‘This is all wrong. This isn’t how highways get built!”

Caro is explaining how he came to write The Power Broker, his penetrating study of Robert Moses, the former New York City highway czar who drew much of the map for that metropolis.

Previously a reporter at Newsday, Caro knew how highways got built; it was when Robert Moses wanted them built. His decision to tell the story resulted in a prize-winning account of the interaction between an individual and institutional power: how a wily and single-minded administrator, never elected to public office, came to tower over his supposed superiors who were elected. Caro says he never intended to be a biographer. But, “the more textbooks I read on political science and urban and regional planning, the more I felt they didn’t have any relationship whatsoever to the realities of how political decisions are made.”

Caro is not the first to conclude that much “political science,” in its obsession with computer analyses and theoretical “models,” is out of touch with the realities of government and power.

About a decade ago, a study called the “Political Science Utilization Directory” attempted to evaluate the discipline by asking political scientists working in government whether their education was helping them in their jobs. The assessment read like a Dean Martin roast. Political science, one Commerce Department official tartly suggested, should start to “deal with issues, theoretical or other, which are related in some way to reality, past or present” Perhaps the most pungent comment came from a grant officer at the Education Department who was asked about the employment prospects for political scientists at his agency. “Present Hiring: Only Two or Three of 2,000. Future Hiring: Less?”

What prompts such responses? For a clue, one might turn to the American Political Science Review, a good barometer of the profession’s academic concerns. One recent entry, “The Political Reliability of Italian Governments: An Exponential Survival Model,” aimed to discover “a well-defined, systematic pattern in the downfall process of Italian government . . . from a probabilistic and dynamic perspective?” There followed 19 pages of graphs, equations, and scholarly citations informing breathless readers that the “half-life” of an Italian government is “approximately 33 weeks” and that “the average government falls when it has accumulated exactly one unit of political stress.”

An extreme example, of course. But it is neither exceptional, nor as funny as it might at first appear. Unfortunately, the effects of these scientistic obsessions extend well beyond the academy. The people who design computer models of the Italian government also teach successive generations of American college students what politics and government are all about. They write many of the books that inform us as citizens and voters. If these scholars were giving us readable accounts of how the institutions of our democracy really work, we could have a more informed political debate. We might even have better programs and policies. A recent remark by Sar Levitan, an economist at George Washington University, applied as much to political science as to the social sciences generally. “I’m not interested in running another regression analysis, another statistical model, ” Levitan told The New York Times. “I’m not interested in writing in an obscure way, and being understood only by an elitist group.”

Cheerleaders versus scientists

The notion of a political “science” is a fairly recent development in the west. Traditionally, political thought grew from first-hand observation and experience. Hobbes, Machiavelli, Locke, Mill, and others were all involved in the politics of their day, as were architects of American democracy, such as Jefferson and Madison. The Federalist Papers are lacking in multilinear equations. But they achieve a salty sense of political reality.

A semi-official history of the political science profession, The Development of Political Science, does not begin with the Federalist Papers. The starting point is almost a hundred years later. “We deliberately limited our attention to those aspects of the past which bear directly on the present ‘state of the discipline,” the authors explain. But if Madison and Jefferson didn’t count, the members of this profession weren’t sure in the early years who did.

The political science of the first decades of this century divided roughly into two schools. One was what Professor Thomas Reed called “education for democratic citizenship .” Reed launched a crusade in the 1930s to bring political science to the people, producing a national radio program called “You and Your Government .” “There is much too little being done by American colleges and universities in preparing young men and and women for actual participation in politics,” he declared.

Yet for all the high-mindedness, the actual product of this approach tended towards conceptual flatulence and simplistic cheerleading for democracy. Civics texts would expound on city charters and the principle of separation of powers without exploring how these documents and formulations actually worked. They exhorted students to vote for the candidate of their choice. John W. Burgess, founder of the first American graduate school of political science at Columbia University late in the nineteenth century, underscored the need to lure American youth away from the German universities, which were infecting them with state socialism. “This evil is in the process of remedy by American schools of political science,” Burgess said, just possibly to the reassurance of potential donors.

Alongside the stalwarts of civic virtue was a second school, which desired to establish the profession in the prestigious ranks of science. If Darwin could uncover the laws of evolution, and Freud the workings of the psyche, could not political scientists lay bare fixed principles of political behavior? This was the vision of Professor Charles Merriam of the University of Chicago, who wrote in 1925 that “jungle politics and laboratory science are incompatible.” The directors of the Laura Spelman Rockefeller Fund agreed and put up millions of dollars for Merriam’s Social Science Research Council, which in turn attracted promising new scholars to what became known as the “behavioral” approach.

Nevertheless, many political scientists remained close to the world of political action. The best example was Woodrow Wilson, who, two years before he took the president’s oath of office, addressed the American Political Science Association as its president. “There is a statesmanship of thought and there is a statesmanship of action,” Wilson said, urging scholars and politicians to “draw near to one another and feel they are engaged in a common enterprise.” In 1911, the APSA tried to survey the political activities of its members but had to abandon the project after it was deluged with information. One 1914 survey showed that 34 of 57 political scientists were politically involved in “more than a casual manner.” Even Charles Merriam was a member of the Chicago City Council. More than half of the articles published in the Political Science Quarterly in five sample years in the early twentieth century dealt directly with current problems and events.

Academic participation in government reached a crest with the New Deal and, later, during World War II. Political scientists poured into Washington along with scholars of other stripes. Columbia alone sent not just the famous Brain Trust of Raymond Moley, Adolph Berle, and Rexford Tugwell, but lesser lights to such agencies as the Interstate Commission on Unemployment Insurance, the National Recovery Administration, and the Federal Emergency Board of Public Works.

If World War II brought yet more political scientists into government, it also brought a rude awakening. “The confrontation of theory and reality provoked, in most of the men who performed their stint in Washington or elsewhere, a strong sense of the inadequacies of the conventional approaches,” wrote Robert Dahl of Yale, a leading exponent of scientistic methods.

This realization might have moved the profession to see that neither the behavioralist nor the civic booster approach was adequate and that political scientists needed to study more closely the agencies and bureaucracies in which they worked during the thirties and forties. But something else was happening around this time, namely, the rise of science into the innermost circles of American public policy. The explosion of the atomic bomb had elevated nuclear physicists to a position of prominence in the military as the designers of democracy’s arsenal. At the same time, the writings of John Maynard Keynes had given birth to “economic policy,” the idea that governments, by manipulating such variables as taxes, spending, and the supply of money, could keep the entire economy on even keel. Paul Samuelson and other leading economists seemed to know something tangible and real that would keep us out of another depression. They could calculate the effect of a given tax cut on jobs and growth with the precision—so it was thought—of a physicist calculating the destructive capacity of a new bomb.

But political scientists—what did they know? Bearing what might well have appeared to be a “Dear John” letter to the profession, a special Committee on Wartime Services reported to the APSA in 1944 that “few departments indicate ‘political science’ as a necessary or even a highly desired background of knowledge for any of their research positions.” It wasn’t just a question of access to government jobs; it seemed to portend an exclusion from the national policy debate. With this crisis of status, the urge to emulate the natural sciences really took hold in political science, as in other branches of social study. Retreat from the world of personal observation and experience was not universal, but the mainstream of the profession clearly leaned that way. A poll of political scientists in 1965 found that seven of the ten “great men” of their field were behavioralists.

The key concepts of the new scientistic political science were modeling and quantification. It wasn’t enough to explain how an agency or party caucus worked; the information had to be cast in terms of a model. Political scientists seemed to think they were assembling, print-out by printout, a universal theory governing voters with the certainty of the principles governing atoms. Many turned methodological somersaults to demonstrate the obvious. So scholars produced lengthy equations that proved Italian governments tend to fall frequently and multivariant analyses of the determinants of congressional voting “behavior.” (Conclusion: interest groups do exert influence, but further research is needed.) Hans Morgenthau, perhaps the most respected critic of behavioralism, commented on one book of that genre that it conveyed information “which Aristotle would have taken for granted.”

In the eyes of the new political scientists, something was not known until it could be expressed as a number. Words were soft and indefinite; numbers were hard and conveyed authority. Hence these academics gravitated towards subjects that could be quantified and factored into computer models. Adding to the gravitational pull was the Ford Foundation, which supported the post-war behavioralism movement with millions of dollars for research. Through concerted lobbying, the APSA got the National Science Foundation to recognize political science as a bona fide “behavioral science,” and funding and prestige began to flow in from this source as well.

The new orientation was reflected in the professional journals. From 1958 to 1968, for example, the American Political Science Review published only three articles on the urban crisis, four on racial conflicts, one on poverty, two on civil disobedience, and two on violence. However, readers did get such entries as “A Multi-Bloc Model of the International System,” “Spatial Models of Party Competition,” and “Dimensions of Political Systems: Factor Analysis of a Cross-Polity Survey.”

But the people who labored over such esoterica. were, unsurprisingly, much less involved in government and public affairs than the earlier generation had been. The Utilization Directory estimated in 1973 that the percentage of APSA members in federal, state, and local government was “substantially below one percent .” PS, the journal of the APSA, could identify only 30 political scientists in the 50 state legislatures in 1979. The distance from government was both a symptom of the new emphasis on “science” and a cause of the increasingly sterile tone of political science writings. The preferred role was now that of expert consultant, which enabled academics to dispense learned counsel without leaving the universities or compromising their “objectivity!’ Where Charles Merriam, a pioneer of the earlier era, was a member of the Chicago City Council, a leader of the post-war behavioralists named David Easton was a consultant to Canada’s Royal Commission on Bilingualism.

Modeling for Tenure

The cult of scientistic political science did not go unquestioned. During the 1960s, in particular, the bitter unrest in the streets spilled over into the discipline that was supposed to be able to explain these events. Liberals attacked the status-quo bias of the “equilibrium” models favored by the mainstream, while conservatives like Leo Strauss denounced with equal fervor a bias towards liberalism. The scientistic fetish came in for renewed attack. One C.W. Harrington complained in a letter to the American Political Science Review, for example, that “there is but one article [in the last issue] that does not read as though written with the aid of a computer, or in some cases, actually written by a computer itself!’ Harrington took the Review to task for “recurrent cant” — “stochastic,” “probabilistic,” etc.—and suggested that the journal be published in two editions, one for political science and the other called the “American Computer Fondlers Review.”

The Vietnam war in particular jolted the profession; by 1970 even leading behavioralists such as Karl Deutsch and Robert Dahl were willing to compromise their objectivity by signing a letter to President Nixon opposing the war. The convulsive events of the era just didn’t fit the neat predictive models. “NAACP can be handled in the equilibrium power, but Black Power cannot,” Charles McCoy and John Playford wrote in a 1967 essay.

Today, by most accounts, the behavioralists and their critics have reached a truce. There is still room in the discipline for a Wilson Carey McWilliams at Rutgers writing about the idea of fraternity in American history, or a Byron Shafer of Florida State University, who has chronicled the struggle for control of the Democratic party. But the mainstream is still in the quantitative, model-oriented mode. They “do dominate,” Ross Baker of Rutgers observes—an assessment reinforced by the American Political Science Review, which can still resemble a National Lampoon parody of an academic journal. While what gets in can be confusing, what’s left out can be even more so. From 1973 to 1975, for example, there were no articles on Watergate, the CIA, or Vietnam.

The quantifiers and modelers tend to perpetuate themselves by shaping career paths to advanced degrees and tenure. Playing it safe, many graduate students choose thesis topics that lend themselves to quantitative research. Picture the young political scientist, struggling to gain a toehold in a viciously competitive field. An assistant professor is more likely to impress the department chairman by constructing a diagram of input-output interactive feedbacks than of studying how the chairman of the Federal Maritime Commission uses or misuses power. Seventy-three percent of the political scientists polled in 1963 agreed that “much research . . . is undertaken simply because the projects lend themselves to research by a fashionable tool or because financial support can be readily secured,” and there is little evidence of change since then.

The widely reported case of Paul Starr at Harvard University illustrates the prestige the quantifiers still enjoy in the social sciences generally. Starr, a sociologist, won a Pulitzer, among other prizes, for his book, The Social Transformation of American Medicine. Harvard President Derek Bok vetoed the decision of his Sociology Department to grant Starr tenure on the advice of an advisory committee that urged Bok to lean toward quantitative types more attuned to computer-modeling. Certainly the input-output diagrams will carry more weight than actual experience in the field. “Those with superior academic credentials are definitely preferred to those with policy experience,” says Professor Richard Falk of Princeton. “All things being equal, the person who doesn’t participate is in a better position.”

Originally, there was good reason for seeking a more empirical approach to political science. But in striving to be hardheaded, the profession went off the deep end. Scholars could have used quantitative research to inform judgment. They could have analyzed congressional budget votes, for example, as part of a study of how the inner dynamics of Congress affect such votes. They could have done their computer correlations of voting patterns in combination with gritty, firsthand reporting of how the parties actually enlist those loyalties at the local level. The tendency instead has been to use numbers as a substitute for judgment. “We don’t want to float back to the armchair period,” says James David Barber of Duke University. “[But] where it gets idiotic is the modeling—the incredible hubris that social life can be reduced with such precision.”

Learning from Royko

This hubris comes across to students. The vision of politics that they encounter in “political science” is cold and aloof, conveying a sense of superiority towards democracy itself. “The training tends to distance you from [politics],” William McCallister, a political science professor at Barnard College, recalls of his education at the University of Chicago, still a behavioralist stronghold. “It channeled that interest in a way that made me more alienated from the political world.”

When I took courses in political science a few years ago, I was struck by the contrast between the books on our reading lists by journalists and historians on the one hand and the political science texts on the other. With few exceptions, the former were far more valuable and interesting. For example, in my urban politics course, Mike Royko’s Boss—a book about the late Mayor Richard Daley and the Chicago political machine—taught me more than did the political science models of municipal service delivery. Royko didn’t have to set up a feedback model to grasp that neighborhoods that support the mayor get the streets cleaned faster. He was able to convey the drama of the political arena and still draw broad theoretical conclusions that made the book more than just a good read.

What we need is the best of both the pre- and post-war political science: scholars who know the literature and the theory, but who also seek direct exposure to the people and institutions they study. First-hand experience can give scholars a feel for the process side of government that they might otherwise miss: How do programs actually. work? How do bureaucratic concerns—infighting and the like—prevent an agency from doing its job well? Watching government like a journalist— or, better yet, as a participant—can also reveal where policies fall short.

A recent investigation by the Los Angeles Herald Examiner provides a good example. An Examiner reporter spent weeks observing the operations of southern California unemployment compensation offices. The process seemed to run smoothly, as each day thousands of people lined up to receive their checks. But by interviewing scores of the recipients, the reporter discovered that many of them were wealthy film industry employees who qualified for support because they worked sporadically and were often technically “unemployed .” A political scientist analyzing California’s unemployment program could easily overlook such waste unless he saw the lines and talked to the recipients himself.

One way to bring political science back to reality would be to require those who teach the “science” of politics to gain practical experience during their graduate training—working in an unemployment office, managing a political campaign, or even running for city council. The APSA’s Congressional Fellows program should be expanded to encompass work throughout the federal bureaucracy and in state and local governments as well.

People who have been there know the questions to ask and which numbers are important. Howard Wriggins, a Columbia professor of international affairs who was President Carter’s ambassador to Sri Lanka and a staff member of the National Security Council, says he gained new “skepticism about elaborate methodologies” on the job. These “highly elaborate models are a distraction,” he says, because they do not consider adequately the “marvelous variety of human beings, human forces.” Nor, one might add, the ways institutions shape and channel those human forces.

Even in the realm of economics, it has become evident in recent years that the “elaborate models” are inadequate and that we need to pay more attention to “human forces,” such as management and entrepreneurship, which do not lend themselves to statistical analysis. Now that we have plenty of books on why corporations aren’t working, it’s time for the political scientists to start telling us why government isn’t working either. Just because Woodrow Wilson talked about the New Freedom instead of the New Systems Output, doesn’t mean he was entirely on the wrong track.

Our ideas can save democracy... But we need your help! Donate Now!

Steven Waldman

Steven Waldman is chair of the Rebuild Local News Coalition, cofounder of Report for America, and a contributing editor at the Washington Monthly.