Our Polarized Parties Dimly Seen

Harvey C. Mansfield

Winter 2020

By various measures, America's political parties now have a high degree, perhaps an excess, of polarization. Mainstream political science has difficulty understanding this condition because political science now tends to begin from the notion that interests, rather than ideas, are the prime cause of political behavior. This premise blinds many scholars to a great deal of what matters about politics — and it may be especially blinding just now. 

And yet, political science turns out not quite to believe its own premise. To understand "interest," one must consider the contrast between opinion (as the expression of ideas) and interest. Opinion is common and ordinary, distinct from scientific knowledge, yet it is also an attempt at knowledge of what is true and what is good. The contrast between opinion and interest can bring us to understand some subtle, implicit concessions that contemporary political science makes to the power of opinion (or ideas) — particularly through the concept of "preferences," and through a reliance on surveys.

With these concessions, mainstream political science evinces a loss of nerve in its devotion to the scientific project, which it neither fulfils nor forswears. And it opens the door to an even more profound challenge from ambition, a human quality not explicable by interest but which is the outstanding feature of America's party politics. With partisanship and ambition comes the notion of rule, which is a key to understanding politics but is now covered over and lost to sight under layers of scientific pretensions. By digging out that concept, we might come to understand our situation far more clearly.


The political scientist Richard Neustadt used to tell a story about himself that illustrates the difference between opinion and interest. Once, coming out of a polling booth, he was confronted by a young exit pollster with a clipboard, who asked him for whom he had voted. When Neustadt told him, the pollster wanted to know why. Neustadt replied, "Because I'm a Democrat." This seemingly obvious response stymied the pollster. To his dismay (and the professor's amusement) there was no category listed on the clipboard for that answer. Neustadt was expected to give the cause of his vote as his income, his religion, his ethnicity, his social status, or some such factor — in every case something outside of, and antecedent to, the partisan idea that he expressed. If partisanship were to be considered at all, it would be as something like a group identity rather than a commitment to any set of substantive ideas.

In accordance with mainstream political science, the idea had to be an effect of these supposedly determining components, not itself a cause. It has to be mere superstructure, while what really matters is something more like interest.

Interest is calculated and specific to one's own or one's own group's advantage; so it is called "self-interest." But an interest is a contrived concept most at home in economics, and is even more narrow and specific than one's advantage. Interest is confined to this world and is typically commercial. Buyers and sellers have contrary interests in a negotiation that both can understand and predict. As Adam Smith said, the butcher and the baker bring their products to the public out of "their regard to their own interest" rather than public spirit. The concept of interest is theoretical and was conceived to bring harmony out of differences — like those between buyers and sellers, who have opposed interests in a negotiation and yet a common interest in a regular, predictable settlement. Interest is more harmonious and more effectual (so it is supposed) in practice than is opinion. As calculated and specific, self-interest is quantifiable, and so it is countable and indeed is best understood by an observer rather than the person being observed and counted. It is therefore tempting to a political science that wants to be clear, exact, and universal like the rest of science.

Opinions, on the other hand, are broad and hence vague, which means they are not definable as causes of the sort that produce effects. Opinions are what people say, but interests are the motives that prompt them to act. Interests are thus more reliable, because less deceptive, than opinions. But to understand interests one must learn to deprecate or even to ignore opinions, for it is often in your interest to disguise your interest as a high-minded opinion. One could say that the analysis of partisanship by interests necessarily interprets the resort to opinion and ideas as boasting, whether high-minded and self-deceiving like the idealist or angry and willful like the agitator or revolutionist.

The study of ideas and opinions, however, can make a claim to capture obvious facts not seen by students of interest, who must subordinate and re-interpret what they see and hear to hidden causes. This broader and deeper study allows partisans to explain themselves by taking seriously the human desire to rise above one's selfish interest, thus assuming that humans can think and that what they think matters. It pays them the compliment of supposing that as humans they are capable of taking charge of themselves, either individually or together, and of functioning as free agents rather than being determined by imputed causes that deny their freedom.

Perhaps the sensible course would be to say that in politics people are sometimes motivated by an idea or a principle, sometimes by interest — and thus people are, in a sense, both free and slave. This is what James Madison argues in Federalist No. 10. If this is so, perhaps the difference between interest and principle is not as stark as it seems. One can understand high-minded principle as common or mean interest, but one can also see interest as chosen principle that elevates a common good. It is no surprise to see a professor vote Democratic or a stockbroker vote Republican, but should they therefore be accused of a thoughtless pursuit of interest for the sake of declaring political science "scientific"? 

It is hard to avoid the conclusion that one must choose between elevating and demeaning human beings when explaining their behavior. If so, perhaps it is better to elevate, because it is safer to risk disappointment than to miss recognizing what is admirable. Any decision to judge people by interest, and by interest alone, as if interest were an unfailing cause, levels one's conclusions by raising the bad and lowering the good, as well as leveling one's praise and blame to an even, grim satisfaction with whatever happens.

This may seem naïve if you insist that ideas are simply or mostly excuses for interests. But in our era of polarized parties, it may actually be easier than usual to see the influence of ideas. Our partisan ideas become clearer as causes when their clashing goes beyond mild disagreement around the middle to intense disputes at the polar extremes. At that point, partisanship itself becomes open to question, and citizens looking for the cause of such intense feeling readily distinguish it from the calculation recommended and rewarded by self-interest. It becomes clear that voters want to be loyal partisans rather than utility-seekers indifferent to both the passionate and principled claims of partisanship.

It may still be unclear whether passion produces or expresses opinion, but when one believes that interest and opinion are opposed, as the very concept of interest almost necessarily implies, the prevalence of passionate opinion makes it seem a reasonable thing to blame. Political science becomes more open to opinion, rather than interest, as cause when opinion seems reckless rather than self-interested. In that condition, political science may approach the truth that all human action has a reason accompanying it.


Ever since Arthur Bentley's The Process of Government was published in 1908, American political science has relied on the concept of "interest groups" to explain the behavior of parties. The subtitle of Bentley's book was "A Study of Social Pressures," and "pressure," a term borrowed from physics, describes a material cause producing a material effect. Hence it was common for political scientists such as V. O. Key, Jr. — a pioneer in the study of parties in the 1950s and still regarded with great respect — to speak of "pressure groups."

The pressure these groups cause (or transmit) comes from outside those pressured and is felt rather than reasoned to, and so is basically involuntary. One can object that pressure groups of humans may not always produce an effect, for humans have the faculty of resisting pressure and insisting on going their own way. And if several pressures operate at once, there will be a choice to weigh. The use of a term from physics should be understood, therefore, as a rough analogy that is often misleading.

An "interest" in the exact, scientifically viable sense must be all the more involuntary. Yet the most interesting human interests, particularly those that inspire political parties, are too vague in extent and too much in dispute to be obvious. In politics, it seems, interest groups are actually best understood as opinion groups.

Thus one sees that Key used both interests and opinions to develop the undogmatic insights, in some ways still unsurpassed, of his major work Southern Politics, published in 1949. A decade later, a collaborative landmark study appeared, The American Voter, which was more reliant on the concept of interest than Key's work, and thus more strictly behavioral. (After all, imputing interest to humans allows one to judge solely by behavior, all human intent being the same and, apart from interest, imaginary and irrelevant.) In 1957, Anthony Downs published An Economic Theory of Democracy, which carried the concept of interest back to its origin in economics and, by reducing opinions to interests, attempted to invest the study of parties with the greater rigor that attends the dismal science.

Downs's book is still well-known today for its compact analysis of the "median voter." But it has more readers than fans for its insistence on the simplification of politics that results from focusing on interests. In Downs's theory, the median voter is the one between the parties who likes to think himself independent and is in fact probably merely apathetic or ignorant or both. Since neither party can be sure that its particular coalition of interests will gain a majority, it must turn to the center to gain votes, and the median voter becomes the center of attention for both sides.

This logic is "economic" in character even though it applies to parties of principle as well as, and actually more than, those of interest. Downs shows that even zealous partisans of principle have an interest in calculating by interest — which is good because it keeps principled parties moderate. Using Downs's theory, however, one would expect voters in any identified group to divide roughly equally for the Democrats and the Republicans so as to gain the benefits that accrue to the median voter who will be courted by both sides. Yet about 90% of black voters today, for example, are Democratic. Is this their interest or a matter of principle? Perhaps it is not necessary to decide. But then one loses the nobility of principle as well as the exactness of interest, each spoiling the other in the observer's sensibly moderate mix of motives to be assigned. One can neither condemn black voters for irrationality of opinion nor praise them for calculation of interest.

This troubling situation is by no means constrained to any particular race or other group in our politics. We now find political scientists in this general situation of methodological unease across the board, unwilling either to look beneath the surface of political opinion like a cynic or to take it seriously at face value like (what they think is) a fool. Polarization, the outstanding fact of party politics today, means that the pull of interest toward the center, where the median voter resides, has been replaced by the attraction of principle to the steamy extremes where zeal rather than interest prevails. So you would think this would lead political scientists to take opinion more seriously. But they have managed to avoid doing so.

They have avoided it by a variety of methods. Some have chosen a partisan approach to explaining away partisanship. They are inclined to blame polarization on the "ideological" conservative moralism of the right for having initiated a movement that became general when the left was compelled to defend itself and its interests. E. J. Dionne, Norman Ornstein, Thomas Mann, Christopher Achen, and Larry Bartels all fall into this category, for instance.

Others seek recourse in "choice." John Aldrich, author of Why Parties?, the most prominent text in the field, is a follower of the so-called "formal theory" of rational choice. That approach develops Downs's theory of economic interest so as to determine the path of rational pursuit of one's interest, regardless of its content (hence "formal"). Formal theory emphasizes choice (which is easy to assume) rather than utility (which is hard to define); it liberates the political scientist from rank utilitarianism at the cost of the causal determination that arises from knowing for sure what one's true interest is.

Aldrich takes the moderate and sensible position discussed above that some behavior is determined by interest and some by principle. But to do this he combines them promiscuously, apparently forgetting that they are defined by being opposed to each other; an action done on principle is done despite one's interest, and vice versa. To explain the function parties perform, he studies party history in America because Americans had to discover that function; it wasn't so clearly in their interest as to come to sight spontaneously.

His resort to party history implies that a pluralist principle had over time to be accepted, as opposed to merely pursuing one's interest. Yet in the book's title (Why Parties?) and beginning chapters, he remains faithful to the rational-choice theory according to which collective action is the focus of attention. The title's question asks whether collective action is in one's interest as an individual, assuming that the individual and his interest are the foundation from which to begin. The kind or character of party is less important in this view than the very existence of party, which poses the prime question of whether one's interest as an isolated individual is to cooperate with others or take advantage of them. It is noteworthy that the concept of interest, which is intended from the first and above all to avoid the partisan disputation of political opinion, cannot resolve the simplest question of whether political action is in one's interest.

With the polarization of the parties, however, the focus of the rational-choice school on collective action as such seems inapt. Today, we find it all too easy to collectivize as Republicans or Democrats; the problem is rather that our collective action is too sure of itself. We care little for calculating interest and rush into principle, preferring purity of principle ("ideology") to tolerant harmony, and rejecting both the egotistical calculation and the urbane civility required for furthering one's interest.

In this fraught situation, Morris Fiorina appears, able, cool, and genial — wearing the maestro's mantle of V. O. Key and ready to save the relevance and reputation of political science to our politics. He points out that polarization is far from complete: In fact, more people are between the parties than at either extreme. The median voter is the modal voter. Fiorina characterizes recent party history as a period of "sorting," in which liberal Republicans went home to the Democrats and conservative Democrats to the Republicans, together leaving behind non-partisans in the middle. Fiorina rather likes these sensible folks, whom he calls "normal." Democracy is the rule of ordinary, normal people who quite reasonably turn their backs on partisans and refuse their restless agitations against our peaceful and productive comity.

Call these people "apolitical" if you will, but they and not their supercilious critics are the salt of the earth. Their only defect is being unable to defend themselves against over-political moralism. Happily, Fiorina has volunteered to come to their aid. If he wanted to, he could point out that the zealous divisiveness of party was the main reason why party was once thought by America's founders to be the bane, not the salvation, of a free society. It is therefore doubtful and disputable that either interest or principle counsels "collective action" in the form of party.

But the form — there's the rub. Ironically, "formal theory" pays no attention to the particular form that collective action takes, assuming as it does that players in a simple-minded, made-up game called the "prisoner's dilemma" are analogous to a party system teeming with citizens. But Fiorina has the merit of introducing form into the everyday notion of "sorting" by which he describes the process of polarization. Each party has a form of policies and ideas by which its partisans can be sorted out from those of the other party, and this is what Democrats and Republicans have been doing in recent decades.

Policies and ideas have a certain end for which they are adopted, and those who espouse them tend to have a certain character and organization. The phenomenon of sorting cannot be understood in terms of a theory of "collective action" because of the difficulty of explaining any move from individual interest to a collectivity. With sorting, the problem to be explained is not collective action as such, but moving from one sort of collection (competing coalitions that both contain liberals and conservatives) to another (liberals vs. conservatives). The sorting we have seen has two fundamental qualities: the fact that parties sort themselves rather than being sorted by a political scientist outside them, and the fact that parties are sorts opposed to each other.

It is amazing but true that the political science of parties makes little or nothing of the obvious fact that parties are opposed. Surely economic interests are typically opposed, as in buyer vs. seller, when the common reason for being opposed — each wanting more — is obvious and not in dispute. But when the interests say why they oppose each other, as they typically do in politics, dispute can be broad and deep. It is obvious that in all their thoughts and actions the parties address each other; what one likes the other dislikes, what one does the other wants to undo. They do battle every day in every way. They are not merely on the one hand or the other, two separate composites to be sorted by the political scientist. The usual designation of left and right is misleading if it suggests either left and right hands that normally cooperate or left and right turns in opposite directions. Parties neither cooperate nor go off on their own. They form themselves by action that is reaction: the Jeffersonian Republicans against the Federalists, the Lincoln Republicans against the Democrats, the Roosevelt Democrats against the Republicans, the Reagan Republicans against the Democrats.

In fact, we would have to say that parties are antagonistic in argument. What do politicians mainly do? In public and in many private moments, they argue. They debate with friends and opponents, offering their own opinions and in the process refuting those of their adversaries. Even when thinking strategically or tactically, they seek to promote their own opinions and look for weakness, whether in persuasiveness or logic, in those of their opponents.

This way of thinking about politics leans heavily in favor of ideas over interests as its organizing concept. But there is one further avenue that modern political science has pursued to avoid arriving at this conclusion.


Political scientists can continue to ignore ideas when studying parties by describing ideas as "preferences." This is a usage borrowed from economics that is intended to pay respect to the fact/value distinction by refraining from specifying whether the thing "preferred" is worth being so. One can discover behaviorally, it is supposed, that people prefer one thing or taste to another, as for example vanilla to chocolate, even though one cannot say scientifically that either taste is good in itself. One can know better without knowing good.

Economic goods are mostly good in a commonsense way, even if they include (as Adam Smith suggested) trinkets and baubles that give pleasure to superficial minds. In any case, money is useful, and economics is pretty solid when it confines itself to money. But in politics it is misleading to describe principles and policies as mere preferences, for unlike vanilla and chocolate, they oppose each other. It is as if vanilla lovers wanted to do away with chocolate instead of merely preferring it while still liking or tolerating another choice. Neither preference cares about prevailing over the other because in principle they do not exclude each other. "Voters' preferences" are not really preferences; they are contentions not merely in favor of one's own but also, and mainly, against another's views. A preference does not perhaps need an argument, but a partisan contention does need one, indeed consists in one.

A partisan contention wants to prevail over its opposite, but not merely in the sense of winning a game, which is winning once at the end of the game. It wants to rule the whole society, including its opponent, and it wants to do so by imposing its rule, which means its set of rules. To rule is to rule by a principle that rules, by continuing to rule. Thus the pro-life or pro-choice contention today wants not just a single victory but one that establishes (or attempts to establish) its principle as ruling the whole. Neither side is happy with a compromise in which each side, like lovers of vanilla or chocolate, cares merely to practice its own preference.

In this sense, a party is a part of the whole that wants to rule the whole on the basis of its partisan idea of what would be good for the whole. Its contention is a claim to rule, a justification of its rule against that of its opponent or opponents. In democratic America, rule is rule of the majority, and success (say, in a presidential election) is, in Alexis de Tocqueville's words, "not so much to make [a party's] doctrines triumph...as to show...that those doctrines have acquired a majority." Party contention is mainly about the rules and not merely within the rules as of a game. If one listens to political debate rather than dismissing it as of no account, it is immediately obvious that parties are continually attacking and defending ideas, not merely stating them as one does with preferences. Parties claim to rule with ideas that oppose their opponents' ideas and claim to rule the whole despite their opponents' ideas. Of course parties will have to compromise with their opponents and cede to them when they lose, in accordance with the system of party government, yet this is not what they want or claim to be satisfied with.

Another difference between partisan ideas and partisan preferences is that the arguments of the former have to go beyond stating a wish. In economics, preferences represent the demand curve; they state the wishes of consumers or buyers in terms of the goods they want and the prices they're willing to pay for them. But the ultimate price is determined at the intersection of demand with the supply curve, which represents what is available at what price — or we might say reality as opposed to wish. In the survey research of political science, however, voters state their preferences without regard to the feasibility or cost of what they wish for. There is no supply curve, or none as clear as the demand.

Of course, each party in its partisan argument claims to be in touch with reality; it represents its wishes as needs or necessities that would bring disaster if not met, and it proposes means or remedies that are said to be available rather than merely promised. To be sure, it is easy for partisans to overestimate the resources available to finance their party's policies and to underestimate the strength of resistance from the opposing party or parties, thus avoiding reality. Parties are typically desirous of defeating their adversaries rather than winning them over with prudent concessions. They want to rule and to be seen to rule rather than to serve the community anonymously and without calling attention to themselves. One might gain a preference with quiet satisfaction, but for a partisan, nothing beats the joy of a victory celebration. In this, partisanship is indeed like winning (not just playing) a game. Political science thus avoids reality more than partisans do when it covers over the fact of partisanship with the language of preferences.

It is true that the respectability of party under a regime of party government (rather than one-party government) means that the majority party (or coalition) has to admit the respectability of a minority party or parties. This is its reality, and it certainly moderates the character of party rule. A majority party, just by calling itself a party, cannot claim to represent the whole, but it can and will rule the whole. It holds the public authority, and the minority has to obey despite its own wish to rule after the next election — a fact that the majority has to bear in mind.


Given all of this, political scientists cannot truly ignore the existence of opinion. But the language of preferences nonetheless allows them to seek a moving force that isn't quite opinion. Accepting the relevance of partisan views, they have turned to surveys, which are a form of the self-reporting of preferences normally unacceptable to science. This means that they are accepting partisan opinion as a cause of behavior ("because I am a Democrat"), or, since they have no notion of an accompanying cause (once known as a formal cause), they have to be satisfied with "correlation" rather than causation. When dealing with human behavior, choice often intervenes to disrupt the necessity of an effect. With or without a sigh of regret, therefore, political scientists turn to opinion surveys.

Beginning only around 1935 with the work of an election pollster, George Gallup, survey science has grown into the main business of social science, now an industry known for the production and use of "data." In surveys, the primary source for data (things given) is in dicta (things said), despite the pretensions of social science to a realism that rejects what is said for what is done.

The late Sidney Verba, a leader in the construction and promotion of political-science surveys, showed his awareness of the difficulty of this feat in an important address in 1995 summing up what he had accomplished in a long career. Verba presents the citizen as a respondent rather than a voter or ruler. When the survey scientist wants to predict an election, he formulates a sample of the electorate that simulates how he thinks the electorate is divided in its parts and investigates how each part will vote. He makes an artificial electorate that is similar to but not the same as the real one to come. To do this, he avoids going to particular voters to ask how they will vote this time because he cannot afford to ask everybody to anticipate the election. He makes a random selection of respondents within his categories, and so abstracts from what he knows or might guess about popular opinion in order to fashion a scientifically representative sample. When a thousand respondents represent millions of voters, each must be selected with great care yet without foreknowledge of what he will say. One must avoid the "selection bias" of introducing before the survey the things one wants to learn as a result of the survey.

Rather than going to meet the voters to get their opinions as a politician would do, the survey scientist must remain aloof and must observe without becoming involved. He must not argue with voters, helping them to see consequences they might not appreciate on their own, or giving them advice. For him, political science is not practical but theoretical. He constructs a scientific election by means of sampling to predict the real one to come, but also as an alternative to the real one. His scientific election is not influenced by accidents arising from personalities or events and is as little affected as possible by the campaign. He does not arrive at his conclusions by weighing issues, characters, events, and speeches, but rather by ignoring them. The scientific analysis of opinion does not listen to opinions, as a voter would do. On the contrary, it offers an expert critique of volunteered opinion. The scientist "achieving representativeness through science" knows better than the electorate what the true representation of its opinion is, and that truth is not affected by the stated intent of voters.

With fine irony, Verba admits that "real life is dominated by selection bias" — that is, by partisan ideas — but in his view political science provides an unbiased view of the public. This view is superior to that of non-scientists, those who actually vote (or decline to vote). Yet strangely enough, our democracy accepts the untutored vote of non-scientists as decisive and authoritative. The actual as opposed to the simulated electorate is not a random collection, but is arbitrary and not scientifically determined. When a discrepancy occurs between the real and the simulated election, political science defers to the real one and accepts that its prediction was in error. But why? Is not the scientific representation of opinion better, because it is not subject to selection bias and the vagaries of chance in who happens to vote?

Verba himself, out of loyalty to his government rather than to his profession, bows to the authority of the electorate despite its dubious claim to know best what it thinks. But he does permit himself to offer advice. Survey science is more democratic than democracy, he says, because it knows what is not revealed in actual elections. It knows the random electorate, which is the scientific electorate, as opposed to the voting electorate, which dominates and decides. "Democracy" as practiced is not the "better cross-section," the true democracy that survey science can discern. Voters are not representative of the people; they are better informed and more capable on the whole than non-voters. Verba comes to the aid of the non-voters, the "quiescent" electorate that stays home and lets bumptious, noisy people decide. Opinion surveys can bring the quiescent to life and allow them a voice.

For Verba, in a sense, the model voters are the non-voters. Democracy as practiced, as it exists in fact rather than in a model, is characterized by selection bias in favor of the chance multitude that happens to show up and prevail. As an example of political-science democracy, Verba notes the difference between the opinion of welfare recipients, often quiet or subdued, hampered by a modest but dim awareness of ignorance, and the pushy energy of busy Social Security recipients, eager to make their grievances known and their power felt. Elections empower the latter, but political science can help equalize or democratize things.

It is almost but not quite needless to add that this "true democracy" is a caricature, not an analysis, of politics. It has been made by the need of science to find a preceding cause for every effect and thus to deny that government can be its own cause, acting on its own, either prevailing (in Verba's term) or ruling (an older term). Thus too, the people cannot rule in government but can only employ an elite that is forced to keep more or less faithful through elections and especially surveys.


In popular language and in political science when speaking loosely, "interest" can still be heard, because it is still the preferred route from cause to effect. But in survey studies, the term "interest" yields, as we have seen, to the newer and vaguer term "preference," and opinion is restated in terms of preferences. This tenuous middle position between science and humane studies is questionable in several ways.

First, how can opinion be transformed into survey data? One can start from the elementary fact that opinion can be either singular or plural, but data ("things given," a Latin plural) are necessarily plural, as in "what do the data say?" Data are necessarily plural because they are arrived at by measurement, by counting. If one believes that all science requires measurement, social science as part of science must seek and find what is countable. By its scientific ambition, social science does so of necessity; it has an inherent, spontaneous, and forcible inclination to declare that what is measurable is the same as what is relevant and important. The problem then becomes how to decide what commonality is important enough to be counted and how to formulate it. Can this countable commonality be imposed by the political scientist or must it be taken from the self-description of the persons he asks?

James Campbell, a leading scholar of polarization, notes the "important distinction between actual opinions and our measurements of those opinions." Especially for less involved voters, "structured questions" in a survey may elicit a different response than the one the voter would make on his own. Campbell, like Verba, takes the side of the survey subjects but, unlike Verba, doubts random coding and argues for the superiority of surveys based on self-reporting.  Particularly when attempting to identify voters, one should take their word for what they say they are, Democrats or Republicans.

Campbell prefers the self-reporting of partisans to the randomized scanning of scientific professors. His choice illustrates the difficulty, nay, the impossibility, of taking data from opinion. Opinion is assertive rather than descriptive. It has an agenda of things important to whoever offers an opinion, and to whomever it is directed against. It reflects human pride and self-importance, and it is closer to common sense than to science. By taking note of human pride, common sense is able to discount it when appropriate, which is not to discount it in principle as does science. The value of science is taken for granted by political (and social) scientists, and, because science has become so powerful in society, its force is mixed with common sense or partisan opinion, and its distinctiveness is lost to view. Obvious examples of distrust of science can be seen in both our parties: on the right regarding environmental issues, for instance, and on the left regarding differences between the sexes.

Yet a political appeal to opinion is sometimes stronger than scientific objectivity, for too much is lost when human assertiveness is leached out of opinion in order to remake it into data. Human pride sometimes causes people to lie, as when a greater percentage of voters claims to have voted for the winning candidate than actually did so. Through the secret ballot, our democracy allows its citizens to conceal their opinions on the crucial occasion in which they decide and declare who rules. Survey science demands that you naïvely blurt out what you think, but suppose you don't want to? And together with pride is shame, for through shame people may not reveal a disreputable opinion they hold. For a survey, everyone speaks in an unsubtle, humorless monotone; there is no nuance, no irony, no fear. Survey scientists have to judge their results, or interpret them, or spin them. But there is no science for what science cannot reach.

Science admits its "margin of error," as if it knew when it might be wrong, but that margin too is subject to error. In any case, science, whose hallmark is exactness, is not entitled to the margin that only common sense, with its lack of that pretension, deserves. Unfortunately but not surprisingly, experts in survey science tend to believe in it excessively. When they should be admitting (as do Aldrich and Campbell) that surveys need to be supplemented by history and biography, that gradations and individualities matter, and that sudden changes are possible, they instead report what they have found as if this is all that knowledge has to say.  But there is knowledge beyond the limits of modern surveys.

The second major limitation of modern survey science is that it deals in averages. The individual, whether the best or the merely eccentric, is buried in an average figure of some kind, with the consequence that data, once again, refer to the many. The many can be solidified in a mass or a mob, and they can be left uncommunicative, each by itself. The classical alternatives to the many are the one and the few, forms of government signifying the political relevance of the one sovereign or alternatively the one best person, and of the few who defend their distinctive quality against the undistinguished, undistinguishing many. In the featureless multitude of the survey, "the one person who counts" — the top or the best person — is not counted except as among those who do not count especially, and the same goes for the few. The survey mimics the "one person, one vote" principle of the democratic election. One can of course take a survey of a certain group, say doctors, but they would be a randomized selection of all doctors, not the best ones. Randomization in the selection of those surveyed means refusing to take a visible pattern as authoritative, since any apparent pattern is the result of bias (that is, an effect humanly caused and imposed) and open to doubt. A scientific pollster must firmly reject any volunteers who say "ask me, I'll tell you."

The focus on averages induces a search for the average person, the median voter, as we have seen, the one who does not volunteer for politics but still usually has an opinion when asked for it. Yet the survey scientist cannot look too hard among the less interested lest he compromise his result by counting hardened non-voters (in an election survey) who resist all persuasion to serve themselves or the public. The average includes both more and less than the average but not as identifiable minorities, only as non-average. When average becomes the norm, the more and the less are not merely ignored but positively squeezed out. The squeezing-out can be seen in the behavior of the survey scientists themselves. They do not use surveys merely to show averages (which is what surveys can do) while then also turning to history, biography, and political philosophy to round out the picture, bringing out the contributions of remarkable persons in deed and thought, and the consequences of notable unforeseen events. Instead of following a procedure that puts the average in its place, they devote themselves entirely to surveys and form themselves into a coterie of survey scientists speaking mainly to one another and addressing themselves to the public as if everything not data is unserious anecdote.

Lost in the average that survey science insists on is the typical difference between the people and the elite, making for a third difficulty with survey science. The elite are the select, the few chosen from the many. But chosen for what quality? At times in today's usage, the elite are the best, deservedly at the top; at other times, the elite are the arbitrarily privileged, undeserving of rank. If one speaks of an "elite school," this can mean that the school is really the best, but with the suggestion that it produces a snobbish, privileged elite unjustly favored. Is no one deserving of rank? In our democratic age that is often the implication. Is there any regular or natural difference in temperament between those who want to get ahead and those who want a decent competence and no more? In some sense, the elite are better if only in energy, even if their energy gives them greater capacity for mischief. If it is regular and normal that elites exist within the people, so that their appearance in any society can be expected, does this not create a distinction in the people that is relevant for survey science? Some people matter more than others, it seems, and especially in the formation of opinion. If so, surveys should take account of the difference and not merely assume all are equal in the ability to form and promote an opinion.

Fiorina, we have seen, distinguishes between "normal" people who do not want to be bothered with politics and activists, and the political class that wants nothing else. What normal people understand is that the political life is not an end in itself but essentially instrumental to the goods of non-political life. Politics is necessary to ensure the conditions of private happiness, but it also needs to be kept subordinate to the private goals outside politics lest it become an excuse for interfering with those goals. The political problem is that the political class is not satisfied with its modest function. Instead of adopting an attitude of tolerant non-interference — for example in the question of abortion, allowing both right-to-life and right-to-choose to coexist in peace if not harmony — the political class wants to rule one way or the other. "Overreach" is its mode. In this way, the elite, whether liberal or conservative, has an officious, argumentative, meddling, bullying character quite opposed to the innocent goodwill of the multitude. To bring out these elites, hidden often because they are hiding, survey scientists would have to depart from their quantitative method, set aside statistics, and venture into the depths of opinion.

In the face of the phenomenon of opinion as a whole, survey science seems to be preoccupied with the minute, but categorized, inspection of the surface while neglecting the problem of its ambiguity. The very questions survey researchers put to their subjects ignore the questionable character of opinion and turn subjects into "respondents" instead of contributors to political science. To understand someone's opinion, one has to listen to his formulation of it and not substitute one's own. One can question the formulation, but first one must listen to it.

The depth of public opinion is revealed only when one addresses the questions it tries to answer and investigates those it overlooks. One must look not only to its quantities but also, and even more, to its quality. Related to the lack of depth in survey science, to its artificial averageness, is the lack of irony. To ask people their opinions and then solemnly categorize their answers is as naïve in the questioner as is demanded in the answers of those questioned. It is to suppose that the role of "respondent" is robotic.


Partisanship, we have seen, is more than a preference. It is a desire to rule over preferences, directing them to a certain kind of society that prizes those preferences — and, let us not forget, antipathies. Some few in our and in every society desire intensely to rule, holding the spirited desire called "ambition." The desire that partisans hold is the one that the ambitious few want to act on, and so they go into politics, seeking office or the vicinity of office.

Mainstream political scientists have taught us to say that these people seek power, but that is too vague. Power by itself is unspecific potential, but the ambitious seek the exercise of power, and power exercised is always in a certain direction, for a specific purpose. What is that purpose? Here the political scientists tumble into embarrassment, for a desire to rule seems too strong to be a preference and cannot be resolved by a survey. "Do you desire to rule?" This is a question that will receive an ironic answer, because the desire to rule is normally, particularly in a democracy, masked by prudent reluctance to admit it.

Aldrich, one of the few modern political scientists to deal with ambition, reduces it to interest and constructs what he calls an "ambition theory" to explain it. Ambition, he says, is in one's interest as a form of careerism. So, consider Thomas Jefferson or Abraham Lincoln: These fellows had successful careers! But Aldrich does not note that it is in the interest of an ambitious person to deny that he has that interest. Such a one is likely to say that he is ambitious for the public or common good, a response all the more likely since political scientists have taught us to say that ambition is the desire for power, meaning personal power for oneself, not for the community. Yet, with a view to the vexations and dangers of getting and holding public office, the rational-choice theorists have tried to teach us that the more thoughtful choice may be to stifle one's ambition, not enter politics, and enjoy life as a free rider on those who feel an irrational compulsion to do so. For such theorists, "interest" has determinate meaning only as personal interest, not that of a community, a group, or indeed any collectivity (including a family or a friendship). So by their own reckoning one cannot have an ambitious interest, because it would be opposed to a calm, considered calculation of advantage (as an interest must be when it is not obvious). There can be no spirit in politics, any more than elsewhere in life; all is unexcited, unproblematic, uninspired.

But this is just plainly false. There is plenty of spirit in politics, in the antagonism of the two parties always apparent and now particularly intense. It is sure that ambition is for oneself, but it is also for the interest of some larger group as well. The latter "interest" requires some sacrifice of one's usual private comforts, that is, of one's interest as commonly understood. This larger interest, with its element of sacrifice, requires a certain nobility or public spirit, putting one's higher or larger interest over one's lower or usual interest. One might say that it is in one's interest to make this sacrifice, by which one becomes a nobler, more responsible person — in which case it isn't really a "sacrifice." But it's an effort, at least at first, until one becomes accustomed to it.

Some people have ambition spontaneously; some are trained or educated to it; some have it through their own reflection. All this makes the notion of one's "interest" more complicated than in its usual, reductive sense that was originally (that is, in the 18th century) intended to simplify the complicated character of the human soul. This complication is what distinguishes politics from economics, giving political science access to the higher regions of the human soul as opposed to the concern of economics for the conditions of survival and comfort: the good life as opposed to the mere life of the body. The political scientists who want to copy economics go wrong nowhere more obviously than in their inability to deal with ambition.

In Federalist No. 51 appears a sentence that is, and deserves to be, the most famous statement of the essence of American politics: "Ambition must be made to counteract ambition." What does this mean? The ambitious will defend against the ambitious, as they know better what the ambitious are up to, and it is they, not ordinary citizens, who love rivalry with their kind and whose interest (desire) it is to be ambitious. Madison shows republican citizens that they can trust those they distrust, precisely to oppose and not simply deplore ambition, using its ways to defeat its ways. Ambition is dangerous by itself, but it becomes useful by counteracting itself. It is in a sense republican to the extent that it despises fixed, hereditary classes, and although the ambitious constitute a class, it is a class that, when properly managed by means of the separation of powers, helps maintain diversity. Ambitious people distinguish themselves by leading others to do new and different things. Thus the forms of the separated powers create their own social support in the spirits of the ambitious. Unlike the simplistic democratic citizens of our time and their theorists, Madison does not frown upon ambition. In order not to frown upon it, he has to understand that it is neither simply equivalent to interest nor without reference to interest.

The democratic (or republican) principle says that power is safer in the hands of many than in the hands of one or a few. Yet what makes democracy work are the qualifications of that principle, by which the one and the few are transformed from dangers democracy must avoid or combat into contributions democracy needs and can use. Their ambition gives voters candidates they can vote for, without whom they would flounder in an unwieldy, inarticulate mass. If it is to be ruled not by survey results but by leaders, democracy needs ambition.


Democracy cannot afford to reject the few. However much it rails against the elite, it cannot live without one. Railing against the elite is in the character of democracy and hardly peculiar to our time. But democracy needs to be instructed not to succumb to its normal character, and this ought actually to be the job of a genuine political science.

Instead, however, much of our political science is devoted to democratizing American politics even more than American democrats desire, not least through a simplistic, democratized notion of "interest." When interest proves inadequate to explain the behavior of parties, political science turns to artificial opinion in the form of alleged preferences that shape desires into quantities that it wants, with unrestrained confidence, to count. Surveys, with their coded categories, also try to count quantities of artificial opinion with the aim, actually announced as we have seen, of democratizing opinion more than it wants to be democratized. The further aim of this sort of science — the overall aim of democracy in the view of quantitative political science — is to remove from human life all striving, tension, and competition, which is no less than to say all freedom and virtue. The neglect of ambition, or its reduction to interest, signifies the removal of partisanship from parties. Political science itself, one can say, has an ambition: It wants to do away with human ambition.

But appreciating ambition opens up a path to the causal power of formal institutions, above all the Constitution. With institutions come the power of the ideas that conceived the institutions.  One can go further and see more. The American Constitution and its institutions enable the people to rule through representatives rather than directly, through laws rather than by imperious decrees of popular will. There is a certain space between the people and its government, a constitutional space, enabling the government to rule without micromanagement, which in turn enables the people to hold it to account after allowing it to exercise its own judgment.

With this limited but vital freedom of action, the government — and the parties that compete to hold its offices — can be given a chance to govern responsibly. Responsibility is a word used promiscuously in our time for all occasions of democratic praise. But it was invented, or at least brought into English, by James Madison in the Federalist Papers to designate the virtue of republican government, and particularly of the one he helped to found. Responsible government is not merely responsive to popular will but rather takes charge of a situation in order to do what it expects or hopes the people will come, with due deliberation, to accord with. Together with partisanship, our political science needs to rediscover responsibility.

Harvey C. Mansfield is a senior fellow at the Hoover Institution and the William R. Kenan, Jr., Professor of Government at Harvard University.


from the


A weekly newsletter with free essays from past issues of National Affairs and The Public Interest that shed light on the week's pressing issues.


to your National Affairs subscriber account.

Already a subscriber? Activate your account.


Unlimited access to intelligent essays on the nation’s affairs.

Subscribe to National Affairs.