The Craft of Political Research

W. Phillips Shively

• Publisher: Routledge
• Year: 2017
• eBook ISBN: 9781351979412
• Edition: 10
• Topic: Social Sciences
• Subtopic: Social Science Research & Methodology
• Index: Social Sciences

The Craft of Political Research is a non-technical introduction to research design and analysis in political science, emphasizing the choices we make when we design a research project and analyze its results. The book's approach centers on asking an interesting research question, and then designing inquiry into the question so as to eliminate as many alternative explanations as possible.

How do we develop theory, and what constitutes a good research question? How do we develop measures and gather evidence to answer a question? How do we analyze our findings? Students will be introduced to such topics as multidimensional concepts, levels of measurement, validity, reliability, random and non-random measurement error, sampling, case selection, causality, experimental and quasi-experimental design, statistical inference, and regression and correlation analysis. Throughout, the emphasis is on understanding the "back story" of analysis — why do we measure in a particular way, why do we choose one design as against another, why do we conduct our analysis as we do.

Emphasizing the internal logic of research methods and the collaborative nature of the research process, the greatest strength of the book is its clarity and the large range of political science examples it provides. It works at a conceptual level, seeking an understanding of the principles that underlie techniques and the reasons that we choose them.

New to this edition:

• 
  
  

• Updated and international examples from the US, UK, Latin America and China amongst others, and international organizations such as the World Bank and the United Nations.

• New section, "Reading Political Science" reviews sources of published political research, with some broad principles for how to find good sources, and advises students on what to look for in reading a research report

• New section, "Gathering Accurate Information" reviews published sources of data, such as UNESCO, and offers advice about how to use such sources. It advises students on how to gather data in personal interviews and it acquaints them with publicly available data sets for secondary analysis.

• Online material featuring revised learning objectives for each chapter, and a new section offering projects and questions for each chapter.

p.1

1 Doing Research

Scholarly research is exciting and fun to do. Admittedly, some students, caught in the grind of daily and term assignments, may not see it this way. But for people who can carry on research in a more relaxed manner, for professors or students who can involve themselves in a long-term project, research may be a source of fascination and great satisfaction.

Francis’s preoccupation with DNA quickly became full-time. The first afternoon following the discovery that A–T and G–C base pairs had similar shapes, he went back to his thesis measurements, but his effort was ineffectual. Constantly he would pop up from his chair, worriedly look at the cardboard models, fiddle with other combinations, and then, the period of momentary uncertainty over, look satisfied and tell me how important our work was. I enjoyed Francis’s words, even though they lacked the casual sense of understatement known to be the correct way to behave in Cambridge. (Watson, 1968, p. 198)¹

This is the way James D. Watson describes his and Francis Crick’s search for the structure of the DNA molecule. The Double Helix, his account of their work, gives a good picture of the excitement of research. It is more gripping than most mystery novels.

Although research can be exciting in this way, the sad fact is that writing papers for courses is sometimes something of a drag. First of all, course papers are tied to all sorts of rewards and punishments—your future earnings, the approval of others, and so on. All of the anxiety associated with these vulnerabilities comes, indirectly, to lodge on the paper. Yet this is probably a lesser cause for frustration in student research. After all, each of these anxieties may also be present for professional scholars. A more important reason for a student’s ambivalence is the simple fact that a paper is generally regarded, by both teacher and student, as a practice run, going through the motions of scholarship. Usually, not enough time is allowed for the student to think long and seriously about the subject, especially with other papers competing for attention. And even when adequate time is allowed, there usually is a feeling on both sides that this is “just a paper”—that a student will learn even if the research is incomplete. Students must have the chance to learn from their own mistakes, but this attitude toward the research work cheats them of the pleasure and excitement that research can bring, the feeling of creating something that no one ever saw before.

p.2

There is probably no way out of this dilemma. In a book such as this, I cannot give you the drama and excitement of original research. I can only give my own testimony, as one for whom research is very exciting. But I can introduce you to some selected problems you should be aware of if you want to do good research yourself or to evaluate the work of others. I also hope to make you aware of what a challenging game it can be, and of how important inventiveness, originality, and boldness are to good research.

Social Research

Social research is an attempt by social scientists to develop and sharpen theories that give us a handle on the universe. Reality unrefined by theory is too chaotic for us to absorb. Some people vote and others do not; in some elections there are major shifts, in others there are not; some bills are passed by Congress, others are not; economic development programs succeed in some countries, but fail in others; sometimes war comes, sometimes it does not. To have any hope of controlling what happens, we must understand why these things happen. And to have any hope of understanding why they happen, we must simplify our perceptions of reality.

Social scientists carry out this simplification by developing theories. A theory puts a specific phenomenon, such as the fact that the United States has just two main parties, in a broader, general category of causal relationships. It takes a set of similar things that happen—say, the development of the number of parties in democracies—and looks for a common pattern among them that allows us to treat each of these different occurrences as a repeated example of the same thing. Instead of having to think about a large number of disparate happenings, we need only think of a single pattern with some variations.

For example, in a classic book on political parties, Maurice Duverger was concerned with the question of why some countries develop two-party systems and others develop multiparty systems (1963, pp. 206–280). The initial reality was chaotic; scores of countries were involved, with varying numbers and types of parties present at different times in their histories. Duverger devised the theory that (1) if social conflicts overlap, and (2) if the electoral system of the country does not penalize small parties, then the country will develop a multiparty system; otherwise, the country will develop a two-party system.

His idea was that where there is more than one sort of political conflict going on simultaneously in a country, and where the groups of people involved in these conflicts overlap, there will be more than two distinct political positions in the country. For example, a conflict between workers and the middle class might occur at the same time as a conflict between Catholics and non-Catholics. Then, if these groups overlapped so that some of the Catholics were workers and some were middle class, while some of the non-Catholics were workers and some were middle class, there would be four distinct political positions in the country: the Catholic worker position, the non-Catholic worker position, the Catholic middle-class position, and the non-Catholic middle-class position. The appropriate number of parties would then tend to rise, with one party corresponding to each distinct position.

p.3

However, Duverger thought that this tendency could be blocked if the electoral system were set up in such a way as to penalize small parties— by requiring that a candidate have a majority, rather than a plurality, of votes in a district, for instance. This requirement would force some of the distinct groups to compromise their positions and merge into larger parties that would have a better chance of winning elections. Such a process of consolidation logically would culminate in a two-party system. To summarize the theory: A country will develop a two-party system (1) if there are only two distinct political positions in the country, or (2) if despite the presence of more than two distinct political positions, the electoral law forces people of diverse positions to consolidate into two large political parties so as to gain an electoral advantage.

Having formulated this theory, Duverger no longer had to concern himself simultaneously with a great number of idiosyncratic party systems. He needed to think only about a single developmental process, of which all those party systems were examples.

Something is always lost when we simplify reality in this way. By restricting his attention to the number of parties competing in the system, for example, Duverger had to forget about many other potentially interesting things, such as whether any one of the parties was revolutionary, or how many of the parties had any chance of getting a majority of the votes.

Note, too, that Duverger restricted himself in more than just his choice of a theme: He chose deliberately to play down exceptions to his theory, although these exceptions might have provided interesting additional information. Suppose, for instance, that a country for which his theory had predicted a two-party system developed a multiparty system instead. Why was this so? Duverger might have cast around to find an explanation for the exception to his theory, and he could have then incorporated that explanation into the original theory to produce a larger theory. Instead, when faced with exceptions such as these, he chose to accept them as accidents. It was necessary for him to do this in order to keep the theory simple and to the point. Otherwise, it might have grown as complex as the reality that it sought to simplify.

As you can see, there are costs in setting up a theory. Because the theory simplifies reality for us, it also generally requires that we both narrow the range of reality we look at and oversimplify even the portion of reality that falls within that narrowed range. As theorists, we always have to strike a balance between the simplicity of a theory and the number of exceptions we are willing to tolerate. We do not really have any choice. Without theories, we are faced with the unreadable chaos of reality.

p.4

Actually, what social scientists do in developing theories is not different from what we normally do every day in interpreting our environment. Social scientists merely interpret reality in a more systematic and explicit way. Without theories, students of society are trapped. They are reduced to merely observing events, without comment. Imagine a physicist—or a fruit picker for that matter—operating in the absence of theory. All she could do if she saw an apple falling from a tree would be to duck, and she would not even know which way to move.

Social theory, then, is the sum total of all those theories developed by social scientists to explain human behavior. Political theory, a subset of social theory, consists of all theories that have been developed to explain political behavior.

Types of Political Research

The way a particular political scientist conducts research will depend both on the uses that she visualizes for the project and on the way she marshals evidence. Political research may be classified according to these two criteria.

The two main ways by which to distinguish one piece of research from another are as follows:

1 Research may be directed toward providing the answer to a particular problem, or it may be carried on largely for its own sake, to add to our general understanding of politics. This distinction, based on the uses for which research is designed, may be thought of as applied versus basic research.

2 Research may also be intended primarily to discover new facts, or it may be intended to provide new theories to account for old facts. Thus, political research can be characterized by the extent to which it seeks to provide new factual information (empirical versus nonempirical).

Table 1.1 shows us the four types of political research based on different combinations of these two dimensions. Normative theory consists of arguments about what should be in politics. Probably the oldest form of political research, it includes among its practitioners Plato, Karl Marx, Ayn Rand, and others. It is applied research; that is, its goal is problem solving. This means that its main purpose is not so much to devise or amend political theories for their own sake, as to develop political theories that will help us to make good political decisions. It is also nonempirical, in that it does not consist primarily of investigating matters of fact. It typically takes certain political facts as given and combines them with moral arguments to prescribe political action. A good example is John Stuart Mill’s argument in Considerations on Representative Government, in which Mill urges the adoption of democratic representative government because (1) the chief end of government should be to facilitate the development in each citizen of his full potential (moral argument), and (2) democratic government, by giving the people responsibility, does this (factual assumption).

p.5

Table 1.1 Types of Political Research

Like normative theory, engineering research is geared to solving problems. However, its stance is empirical; it is concerned with ascertaining the facts needed to solve political problems. Some examples would be measuring the effects of various reapportionment methods, trying to design a diplomatic strategy to effect disarmament procedures, and designing policies to reduce inflation in health care costs.

These two forms of applied research are important parts of academic political science, but they are also often pursued outside of academia. Political engineering is a thriving industry and many courses relevant to it are taught in political science departments, but research and instruction in it are also often found in special separate institutes like the Kennedy School of Government at Harvard or the Institute for Research on Poverty at the University of Wisconsin. Private corporations such as the Rand Corporation and the Gallup Poll also study engineering problems in politics.

Normative theory is taught extensively in political science departments, but often this means primarily the history of normative theory and its development, not the active formulation of normative theories. Much work in normative theory is done in political science departments, of course, but much is also done in law schools, or outside academia by writers in philosophy, or in publications like the Weekly Standard or the Nation.

At the other end of the continuum from applied research is what I will call recreational research. It is usually called “pure” or “basic” research, but this carries the unpleasant implication that applied research is either impure or of limited value. This type of research is really not as flippant as the choice of the term recreational might make it seem, for this is research carried on for its own sake, to improve political theory. Political scientists pursue this type of research for the twin pleasures of exercising their minds and increasing their understanding of things. In a high sense of the word, it is “recreation.” Theory-oriented research is recreational, empirical research.

Positive theory, largely a post-World War II phenomenon, is the most recently introduced form of political research. Like normative theorists, positive theorists posit certain facts about politics; but in contrast to normative theorists, they posit facts as empirical conditions rather than as the foundation for moral arguments. And they distinctively operate by deriving further implications of the posited conditions by precise logical and mathematical operations. Their concern is to take the posited facts, or assumptions, and derive theories from them. Their end goal is to develop reasonably broad and general theories based on a small number of agreed-upon assumptions.

p.6

A good example of positive theory—indeed, a work by which many would date the emergence of positive theory as a distinct field in political science—is Anthony Downs’ An Economic Theory of Democracy (1957). Downs builds a wide-ranging theory from a set of assumptions such as (1) voters and parties behave rationally, (2) political conflict occurs on only one issue at a time, and (3) political events are not perfectly predictable. Some of the predictions generated from his theory include: (1) in a two-party system, parties will tend to agree very closely on issues, whereas in a multiparty system, they will not; (2) it may be rational for the voter to remain uninformed; and (3) democratic governments tend to redistribute income. (Of course, one must recognize that excerpts such as these do even more than the usual violence to a rich net of theories.) It is important to emphasize that this sort of work is almost solely an exercise in deduction. All of the conclusions derive logically from a limited set of explicit assumptions. Downs’ purpose in this was simply to see where the assumptions he s...

Table of contents

• Cover Page
• The Craft of Political Research
• Title
• Copyright
• Contents
• Detailed Contents
• Foreword
• Preface
• Acknowledgments
• 1 Doing Research
• 2 Political Theories and Research Topics
• 3 Importance of Dimensional Thinking
• 4 Problems of Measurement: Accuracy
• 5 Problems of Measurement: Precision
• 6 Causal Thinking and Design of Research
• 7 Selection of Observations for Study
• 8 Introduction to Statistics: Measuring Relationships for Interval Data
• 9 Introduction to Statistics: Further Topics on Measurement of Relationships
• 10 Introduction to Statistics: Inference, or How to Gamble on Your Research
• Glossary
• Selected Bibliography
• Index