What Is Comparative Politics?

The Comparative Method

If comparison is an important way to test our assumptions and shape our ideals, how we compare cases is important. If there is no set of criteria or guide by which we gather information or draw conclusions, our studies become little more than collections of details. Researchers thus often seek out puzzles—questions about politics with no obvious answer—as a way to guide their research. From there, they rely on some comparative method—a way to compare cases and draw conclusions. By comparing countries or subsets within them, scholars seek out conclusions and generalizations that could be valid in other cases.

To return to our earlier question, let us say that we are interested in why democracy has failed to develop in some countries. We might approach the puzzle of democracy by looking at North Korea. Why has the North Korean government remained communist and highly repressive even as similar regimes around the world have collapsed?

A convincing answer to this puzzle could tell scholars and policy makers a great deal and even guide our tense relations with North Korea in the future. Examining one country closely may lead us to form hypotheses about why a country operates as it does. We call this approach inductive reasoning—the means by which we go from studying a case to generating a hypothesis. But while a study of one country can generate interesting hypotheses, it does not provide enough evidence to test them. Thus we might study North Korea and conclude that the use of nationalism by those in power has been central to the persistence of nondemocratic rule. In so concluding, we might then suggest that future studies look at the relationship between nationalism and authoritarianism in other countries. Inductive reasoning can therefore be a foundation on which we build greater theories in comparative politics.

Comparative politics can also rely on deductive reasoning—starting with a puzzle and from there generating some hypothesis about cause and effect to test against a number of cases. Whereas inductive reasoning starts with the evidence as a way to uncover a hypothesis, deductive reasoning starts with the hypothesis and then seeks out the evidence. In our example of inductive reasoning, we started with a case study of North Korea and ended with a testable generalization about nationalism; using deductive reasoning, we would start with our hypothesis about nationalism and then test that hypothesis by looking at a number of countries. By carrying out such studies, we may find a correlation, or apparent association, between certain factors or variables. If we were particularly ambitious, we might claim to have found cause and effect, or a causal relationship.¹ Inductive and deductive reasoning can help us better understand and explain political outcomes and, ideally, could help us predict them.

Unfortunately, inductive and deductive reasoning is not easy, nor is finding correlation and causation. Comparativists face seven major challenges in trying to examine political features across countries. Let’s move through each of these challenges and show how they complicate the comparative method and comparative politics in general. First, political scientists have difficulty controlling the variables in the cases they study. In other words, in our search for correlations or causal relationships, we are unable to make true comparisons because each of our cases is different. By way of illustration, suppose a researcher wants to determine whether increased exercise by college students leads to higher grades. In studying the students who are her subjects, the researcher can control for a number of variables that might also affect grades, such as the students’ diet, the amount of sleep they get, or any factor that might influence the results. By controlling for these differences and making certain that many of these variables are the same across the subjects, with the exception of exercise, the researcher can carry out her study with greater confidence.

But political science offers few opportunities to control the variables because the variables are a function of real-world politics. As will become clear, economies, cultures, geography, resources, and political structures are amazingly diverse, and it is difficult to control for these differences. Even in a single-case study, variables change over time. At best, we can control as much as possible for variables that might otherwise distort our conclusions. If, for example, we want to understand why gun ownership laws are so much less restrictive in the United States than they are in most other industrialized countries, we are well served to compare the United States with countries that have similar historical, economic, political, and social backgrounds, such as Canada and Australia, rather than Japan or South Africa. This approach allows us to control our variables more effectively, but it still leaves many variables uncontrolled and unaccounted for.

A second, related problem concerns interactions among the variables themselves. Even if we can control our variables in making our comparisons, there is the problem that many of these variables are interconnected and interact. In other words, many variables interact to produce particular outcomes, in what is known as multicausality. A single variable, such as a country’s electoral system or the strength of its judicial system, is unlikely to explain the variation in countries’ gun control laws. The problem of multicausality also reminds us that in the real world there are often no single, easy answers to political problems.

A third problem involves the limits to our information and information gathering. Although the cases we study have many uncontrolled and interconnected variables, we often have too few cases to work with. In the natural sciences, researchers often conduct studies with a huge number of cases—hundreds of stars or thousands of individuals, often studied across time. This breadth allows researchers to select their cases in such a way as to control their variables, and the large number of cases prevents any single unusual case from distorting the findings. But in comparative politics, we are typically limited by the number of countries in the world—fewer than 200 at present, most of which did not exist a few centuries ago. Even if we study some subset of comparative politics (like political parties or examples of revolution), our total number of cases will remain relatively small. And if we attempt to control for differences by trying to find a number of similar cases (for example, countries with a similar level of economic development), our total body of cases will shrink even further.

A fourth problem in comparative politics concerns how we access the few cases we do have. Research is often further hindered by the very factors that make countries interesting to study. Much of the information that political scientists seek is not easy to acquire, necessitating work in the field—that is, conducting interviews or studying government archives abroad. International travel requires time and money, and researchers may spend months or even years in the field. Interviewees may be unwilling to speak on sensitive issues or may distort information. Libraries and archives may be incomplete, or access to them restricted. Governments may bar research on politically sensitive questions. Confronting these obstacles in more than one country is even more challenging. A researcher may be able to read Russian and travel to Russia, but if they want to compare authoritarianism in Russia and China, it would be ideal to be able to read Chinese and conduct research in China as well. Few comparativists have the language skills, time, or resources to conduct field research in many countries. There are few comparativists in North America or Europe who know both Russian and Chinese. As a result, comparativists often master knowledge of a single country or language and rely on inductive reasoning. Single-case study can be extremely valuable—it gives the researcher a great deal of case depth and the ability to tease out novel observations that may come only from close observation. However, such narrow focus can also make it unclear to researchers whether the politics they see in their case study has important similarities to the politics in other cases. In the worst-case scenario, scholars come to believe that the country they study is somehow unique and fail to recognize its similarities to other cases, or they draw sweeping conclusions from the single case that don’t easily apply to other countries.

Fifth, even where comparativists do widen their range of cases, their focus tends to be limited to a single geographic region. The specialist on communist Cuba is more likely to study other Latin American countries than to consider China or North Korea, and the specialist on China is more likely to study South Korea than Russia. This isn’t necessarily a concern, given our earlier discussion of the need to control variables—it may make more sense to study parts of the world where similar variables are clustered rather than to compare countries from different parts of the world. This regional focus, however—often referred to as area studies—is distributed unevenly around the world. For decades, the largest share of research tended to focus on Western Europe, despite the increasing role of Asia in the international system.² Why? As mentioned earlier, some of this is a function of language; many scholars in the West are exposed to European languages in primary or secondary school, and in many European countries the use of English is widespread, thus facilitating research. But English is also widespread in southern Asia; in spite of this, scholarship lagged behind. For example, we find that for much of the postwar era the top journals in comparative politics published as many articles on Sweden as on India. To be fair, much of this is changing thanks to a younger, more diverse generation of scholars, many of whom come from or work in a much wider array of countries around the world. Yet overall, comparative politics remains bounded by particular kinds of questions and areas of focus.

Sixth, the problem of bias makes it even harder to control for variables and to select the right cases. This is a question not of political bias, although that can sometimes be a problem, but of how we select our cases. In the natural sciences, investigators randomize case selection as much as possible to avoid choosing cases that support one hypothesis or another. But for the reasons mentioned earlier, such randomization is not possible in political science. Single-case studies are already influenced by the fact that comparativists study a country because they know its language or find it interesting. Yet even if we rely instead on deductive reasoning—beginning with a hypothesis and then seeking out our cases—we can easily fall into the trap of selection bias.

For example, say we want to understand revolutions, and we hypothesize that their main cause is a rapid growth in inequality. Revolution is what we would call our dependent variable—the variable that is dependent on, or affected by, another variable. Rapid growth in inequality would be our independent variable—the variable that doesn’t depend on changes in other variables and is the presumed cause. How should we select our cases? Most of us would respond by saying that we should find as many cases of revolution as possible and then see whether a rapid growth in inequality preceded those revolutions. But this seemingly logical approach is a mistake, as it leads to what is known as bias on the dependent variable—in other words, a bias in sampling on the effect, rather than the cause. Why is this a problem? By looking only at cases of revolution (the dependent variable, or effect), we miss all the cases with rapid growth in inequality (the independent variable, or cause) where revolution has not taken place. Indeed, even if every revolution is preceded by changes in inequality, there may still be many more cases without revolution than with it, undermining our hypothesis. So, we would do better to start with what we think is the cause (growth in inequality) rather than working backward from the effect (revolution). While this may seem the obvious choice, it is a frequent mistake among scholars who are naturally drawn to particular outcomes and so start there.

A seventh and final concern deals with the heart of political science—the search for cause and effect. Let us for the sake of argument assume that the half-dozen problems we have laid out can be overcome through careful case selection, information gathering, and control of variables. Let us further imagine that with these problems in hand, research finds, for example, that countries with a low rate of female literacy are less likely to be democratic than countries where female literacy is high. Even if we are confident enough to claim that there is a causal relationship between female literacy and democracy—a bold statement indeed—a final and perhaps intractable problem looms. Which variable is cause and which is effect? Do low rates of female literacy limit public participation, empowering nondemocratic actors, or do authoritarian leaders (largely men) reinforce gender inequalities? Or is there yet another variable that is shaping both literacy and authoritarianism that we are not taking into account? This problem of identifying and distinguishing cause and effect, known as endogeneity, is a major obstacle in any comparative research. Even if we are confident that we have found cause and effect, we can’t easily ascertain which is which. On reflection, this is to be expected; one political scientist has called endogeneity “the motor of history,” for causes and effects tend to evolve together, each transforming the other over time. Thus early forms of democracy, literacy, women’s rights (as well as other factors) may well have gone hand in hand, each reinforcing and changing the others. In short, many things matter, and these many things affect each other. This makes an elegant and sequential claim about cause and effect problematic, to say the least.³

Can We Make a Science of Comparative Politics?

We have so far elaborated many of the ways in which comparative politics—and political science in general—makes for difficult study. Variables are hard to control and can be interconnected, while actual cases may be few. Getting access to information may be difficult, and comparisons may be limited by regional knowledge and interests. What questions are asked may be affected by selection bias and endogeneity. All these concerns make it difficult to generate any kind of political science theory, which we can define as an integrated set of hypotheses, assumptions, and facts. At this point, you may well have concluded that a science of politics is hopeless. But it is precisely these kinds of concerns that have driven political science, and comparative politics within it, toward a more scientific approach. Whether this has yielded or will yield significant benefits, and at what cost, is something we will consider next.

Political science and comparative politics have a long pedigree. In almost every major society, there have been masterworks of politics that prescribe rules or, less often, analyze political behavior. In the West, the work of the philosopher Aristotle (384–322 BCE) departed from the traditional emphasis on political ideals to conduct comparative research on existing political systems (what we will call regimes), eventually gathering and analyzing the constitutions of 158 Greek city-states. Aristotle’s objective was to delineate between what he took to be “proper” and “deviant,” or despotic, political regimes. He also framed this discussion in terms of a puzzle—why were some regimes despotic and others not? With this approach, Aristotle conceived of an empirical (that is, observable and verifiable) science of politics with a practical purpose: statecraft, or how to govern. Aristotle was perhaps the first Westerner to separate the study of politics from that of philosophy.⁴

Aristotle’s early approach did not immediately lead to any systematic study of politics. For the next 1,800 years, discussions of politics remained embedded in the realm of philosophy, with the emphasis placed on how politics should be rather than on how politics was actually conducted. Ideals, rather than conclusions drawn from evidence, were the norm. Only with the works of the Italian Niccolò Machiavelli (1469–1527) did a comparative approach to politics truly emerge. Like Aristotle, he sought to analyze different political systems—those that existed around him as well as those that had preceded him, such as the Roman Empire—and even tried to make generalizations about success and failure. These findings, he believed, could then be applied by statesmen to avoid their predecessors’ mistakes. Machiavelli’s work reflects this pragmatism, dealing with the mechanics of government, diplomacy, military strategy, and power.⁵

Because of his emphasis on statecraft and empirical knowledge, Machiavelli is often cited as the first modern political scientist, paving the way for other scholars. His writings came at a time when the medieval order was giving way to the Renaissance, with its emphasis on science, rationalism, secularism, and real-world knowledge over abstract ideals. The resulting work over the next four centuries reinforced the idea that politics, like any other area of knowledge, could be developed as a logical, rigorous, and predictable science.

During those centuries, a number of major thinkers took up the comparative approach to the study of politics, which slowly retreated from moral, philosophical, or religious foundations. In the seventeenth century, authors like Thomas Hobbes and John Locke followed in Machiavelli’s footsteps, advocating particular political systems on the basis of empirical observation and analysis. They were followed in the eighteenth century by such scholars as Jean-Jacques Rousseau and Baron de Montesquieu, whose studies of the separation of power and civil liberties would directly influence the writing of the U.S. Constitution and other constitutions to follow. The work of Karl Marx and Max Weber in the nineteenth and early twentieth centuries, which analyzed the nature of political and economic organization and power, would further add to political science. All these developments reflected widespread changes in scholarly inquiry and often blended political ideals with analytical concepts and some attempt at a systematic method of study.

Thus, by the turn of the twentieth century, political science formally existed as a field of study, but it still looked much different from the way it does now. The study of comparative politics, while less focused on ideals or philosophy, resembled a kind of political journalism: largely descriptive, atheoretical, and concentrated on Europe, which still dominated world politics through its empires. Little of this work was based on the comparative method.

The two world wars and the rise of the Cold War would mark a turning point in political science and comparative politics, particularly in the United States. There were several reasons for this. First, a growing movement surfaced among universities toward applying more rigorous methods to the study of human behavior, whether in sociology, economics, or politics. Second, the world wars raised serious questions about the ability of scholars to meaningfully contribute to an understanding of world affairs. The creation of new countries, the rise of fascism, and the failure of democracy throughout much of interwar Europe were vital concerns, but political scholarship did not seem to shed enough light on these issues and what they meant for international stability. Third, the Cold War with a rival Soviet Union, armed with nuclear weapons and revolutionary ideology, made understanding comparative politics seem a matter of survival. Finally, the postwar period ushered in a wave of technological innovation, such as early computers. This development generated a widespread belief that, through technological innovation, many social problems could be recast as technical concerns, finally to be resolved through science. The fear of another war was thus married with a belief that science was an unmitigated good that had the answers to almost all problems. The question was how to make the science work.

Although these changes dramatically transformed the study of politics, the field itself remained a largely conservative discipline, taking capitalism and democracy as the ideal. In comparative politics, these views were codified in what was known as modernization theory, which held that as societies developed, they would become capitalist democracies, converging around a set of shared values and characteristics. The United States and other Western countries were furthest ahead on this path, and the theory assumed that all countries would eventually catch up unless “diverted” by alternative systems such as communism (as fascism had done in the past).

During the 1950s and 1960s, comparativists influenced by modernization theory expanded their research to include more cases. Field research, supported by government and private grants, became the normal means by which political scientists gathered data. New computer technologies combined with statistical methods were also applied to this expanding wealth of data. Finally, the subject of investigation shifted away from political institutions (such as legislatures and constitutions) and toward individual political behavior. This trend came to be known as the behavioral revolution. Behavioralism hoped to generate theories and generalizations that could help explain and even predict political activity. Ideally, this work would eventually lead to a “grand theory” of political behavior and modernization that would be valid across countries.

Behavioralism and modernization theory were two different things: modernization theory was a set of hypotheses about how countries develop, and behavioralism was a set of methods with which to approach politics. However, both were attempts to study politics more scientifically to achieve certain policy outcomes.⁶ Behavioralism also promoted deductive, large-scale research over the single-case study common in inductive reasoning. It seemed clear to many that political science, and comparative politics within it, would soon be a “real” science.

By the late 1970s, however, this enthusiasm began to meet with resistance. New theories and sophisticated methods of analysis increased scholars’ knowledge about politics around the world, but this knowledge in itself did not lead to the expected breakthroughs. The theories that had been developed, such as modernization theory, increasingly failed to match politics on the ground; instead of becoming more capitalist and more democratic, many newly independent countries faced violent conflict, authoritarianism, and limited economic development. This did not match Western expectations or ideals. What had gone wrong?

Some critics charged that the behavioral revolution’s obsession with appearing scientific had led the discipline astray by emphasizing methodology over deep knowledge of the countries under consideration. Others criticized the field for its ideological bias, arguing that comparativists were interested not in understanding the world on its own terms but in prescribing the Western model of modernization. At worst, comparativists’ work could be viewed as simply serving the foreign policy interests of the United States. Since that time, comparative politics, like all of political science, has grown increasingly fragmented—or, if you prefer, more diverse. While few still believe in the old descriptive approach that dominated the earlier part of the century, there is no consensus about a direction for scholarship and what research methods or analytical concepts are most fruitful. This lack of consensus has led to several main divisions and lines of conflict.