1. What is the puzzle?

Although instructors sometimes suggest that students should begin a research paper with a question, I echo Zinnes (1980) and suggest that it is often more useful for them to begin with a puzzle. A puzzle involves two or more cases that we expect to have similar outcomes but do not, or different outcomes but a similar outcome occurs. Unlike beginning with a general question, such as “what causes war?”, starting with a puzzle requires some knowledge of events or, if it is a philosophical study, some knowledge of different arguments. “This would seem to be the major difference between a puzzle and a simple question. To be truly puzzled, we must already know something. We can be extremely uninformed and yet ask a question” (Zinnes 1980, 338). As Zinnes (1980) observes, focusing on a puzzle is useful for two reasons. First, a puzzle helps anchor the theorizing in concrete cases. No one values a theory that applies to nothing. Second, “puzzles seem to make us think like detectives, and thinking like detectives makes us look for the not necessarily obvious or even visible operating principles” (Zinnes 1980, 318).

2. What are the microfoundations? Who are the actors and what motivates them?

With a puzzle in hand, we then want to look for clues to fit the pieces together, to solve the puzzle. What kind of process might lead to the type of event we are trying to explain? In thinking about the process that produces the outcomes, we are likely to ask some or all of the following questions. Who are the primary actors? What are their goals? What are the constraints affecting their decisions? In answering these questions we will be providing microfoundations for the argument.

To provide microfoundations for an argument is to provide an account of how individuals choose and act. What motivates an agent in this context? Are people in this situation likely to be self-interested or altruistic? Microfoundations, then, are the behavioral characteristics of individual actors. In the social world, events are caused by actors operating under particular constraints, that is, “the mechanisms through which social causation is mediated turn on the structured circumstances of choice of intentional agents, and nothing else” (Little 1998, 203).³

To argue that a complete and coherent theory of social phenomena requires clear microfoundations does not mean structural factors are irrelevant. On the contrary, a focus on individuals naturally leads to examination of institutions because “institutions have effects on individual behavior (incentives, constraints, indoctrination, preference formation), which in turn produce aggregate social outcomes” (Little 1998, 204). It is little surprise that institutionalists tend to emphasize methodological individualism.

After answering these microfoundation questions, we will have identified actors, their goals, constraints, and at least some ways in which the cases are similar and some in which they are different. To more fully develop the argument, more clearly identify the scope of the theory, and make the theory more fruitful, we need to understand the role of assumptions.

3. What assumptions inform the theory?

All theories contain assumptions. In the natural sciences and in philosophy, these building blocks are referred to as axioms and premises. There is no significant difference between an assumption in a theory and a premise in a philosophical argument. Both have to be justified. In explaining social phenomena, we typically make assumptions about the identity of the primary actors, their goals, and constraints. For example, to understand world politics, we might assume that the primary actor is the nation-state, that the goal of the nation-state is the goal of the leader of the state, which is usually to retain office, and that specific domestic or international constraints, such as the balance of power, influence the attainment of this goal.

In making assumptions, we begin to engage in generalization and simplification. With simplification the goal is “to try to find the central tendency among a confusion of tendencies, to single out the propelling principle even though other principles operate, to seek the essential factors where innumerable factors are present” (Waltz 1979, 10). In American politics, for example, we often assume that the primary goal of a politician is to be elected or re-elected. Most politicians also have specific policy goals, but more than anything else what unites the set of politicians is the office imperative, making it a useful assumption and simplification of reality.

Next, I encourage students to write out formally (mathematically) or informally (in prose) the assumptions in their argument. This practice helps students understand where hypotheses come from, that is, a hypothesis links a causal concept with an effect, and the assumptions explain why the hypothesis is logically sound. We typically see a hypothesis written as: if X, then Y. But all hypotheses depend on a set of assumptions, and given the lack of attention paid to assumptions, it may be better to write: given this set of assumptions (A), if X, then Y: “If the specified axioms are true for a given case, then the model [or hypothesis] should apply” (Most and Starr 1984, 398).

4. Is the form of the argument valid? How are the hypotheses connected to the theory?

In contemporary political science, the most common way to ensure our conclusions follow from our assumptions is to use game theory. Game theory places an emphasis on specifying microfoundations and relevant social constraints, states (most) assumptions explicitly, and is formalized, thereby ensuring that the propositions flow from the premises (provided the mathematical operations are correct). Game theory is a formalized system of logic, but it is not the only system of logic, and other systems of logic may even be better at disciplining student thinking and facilitating their ability to make theoretical arguments.

Logic means right reasoning and analyzes whether the form of an argument is valid, that is, whether the stated premises entail the stated conclusion. A basic system of logic is syllogistic logic, which focuses on arguments that contain all, no, or some.⁶

Although limited in scope, understanding syllogistic logic highlights two important points about building a theory, namely the importance of sound premises and valid connections between premises for producing an accurate, or from the scientific perspective, empirically testable, conclusion. For example, consider the following syllogism:

Is the conclusion valid? Does Jane want to be re-elected? In this case, intuition should tell us the answer: Jane wants to be re-elected. Drawing conclusions, however, is not always straightforward. Leblanc (1998, 81) offers the following example.

Is this conclusion valid? We all know that the conclusion is true, some students do indeed fail tests. But, is the conclusion valid? Does it follow from the premises? No, it is not valid. The conclusion does not contain a distributed term (a term that makes a claim about all members of a class). The premises of the argument focus on students who do or do not study, but the conclusion does not say anything about studying. Learning syllogistic logic, then, provides practice in working with assumptions and combining thoughts in a logical manner, thereby improving our ability to think carefully and theoretically. Finally, learning syllogistic logic makes it much easier to learn other systems of logic, especially propositional and quantificational logic, that better address how a decision-maker deals with trade-offs.

Another way to think of the premises in a syllogism is as the mechanism that gives causal capacity to the independent variable, the first concept, in the conclusion statement. Causal capacity means a concept has the “capacity to produce a certain kind of outcome in the presence of appropriate antecedent conditions” (Little 1998, 205). In other words, under conditions A, X, the causal concept, has the capacity to produce the outcome Y.

For example, the informational theory of the democratic peace assumes (1) that military conflict happens because of bargaining failures, the chief cause of which is uncertainty about resolve and escalatory strategies that causes leaders to become locked-in to a course of action, (2) that democratic leaders are more accountable than non-democratic leaders, (3) that accountability increases audience costs, (4) that as audience costs increase uncertainty about resolve decreases, and (5) that as uncertainty about resolve decreases, states adopt less escalatory strategies.⁷

This combination of assumptions leads to the conclusion that pairs of democracies are less likely to experience military conflict than other regime pairs. More importantly, connecting the assumptions of an argument addresses the process behind the hypothesis and generates an answer to a why question. Why are democracies less likely to fight each other than other regime pairs? Because democracies are better able to convey their true preferences and when dealing with other democracies, they are less likely to take steps that heighten tension. In other words, democratic institutions have the causal capacity to reduce uncertainty about resolve under the conditions of accountability and audience costs.

It is common to note that a syllogism is only as sound as its premises, that is, a valid syllogism does not guarantee a sound conclusion. For a conclusion to be sound, the premises need to be true. The need for true premises requires us to justify our premises. For example, for the information theory of the democratic peace to be accurate it is important to justify why democratic leaders can generate higher audience costs. Moreover, to the extent that democratic leaders are not able to generate higher audience costs, the expectations of this theory will not hold. In the social sciences, our premises will never hold all of the time because people have free will and there is enough situational variation that we can only identify central tendencies; as a result, the assumptions of a theory delimit the domain of that theory.

In addition to verifying the logic of an argument, outlining the core premises of an argument produces two other benefits. First, in stating a premise we see the importance of defining our concepts. Poorly defined or ambiguous concepts lead to inappropriate comparisons and contrasts as well as ambiguous conclusions. Second, a useful research strategy is to problematize an assumption in a popular theory. Assumptions, we noted, typically express a central tendency, which naturally leads to the question of when does an assumption hold and when does it not? In turn, an investigation of the assumption may suggest either an alternative concept that encompasses the previous one or an alternative model specification that significantly increases understanding of the topic.

5. Some theory building exercises

As is the case with most skills, the best way to improve is to practice, practice, practice. How can students cultivate theory building skills? Lave and March (1975, 19–20) recommend the following approach.⁸

Start with a fact. Then, create an explanation that accounts for this fact. “Look at the facts as though they were the end result of some unknown process (model). Then speculate about processes that might have produced such a result” (20). They do not say anything about assumptions, but in speculating about the process generating the phenomena to be explained we need to identify a set of connected assumptions to account for the logic producing the hypothesis. (Recall that a hypothesis is the end point of theorizing; it is the conclusion of a chain of reasoning.) With a preliminary hypothesis in hand, ask: If this hypothesis is accurate, what outcomes should we observe and not observe? What type of information or events would falsify the hypothesis? If nothing will falsify the hypothesis, then the premises are internally inconsistent. To add a little complexity to this theory generation exercise, ask students to create a second explanation of the original phenomenon, and then identify a way in which to discriminate between the competing theories.⁹

For building on this basic theory generation exercise, Lave and March recommend specifying additional hypotheses. Presumably, the original argument produces a hypothesis that accounts for the original type of event. To develop confidence in the theory, however, we would like it to explain something else. Two strategies for developing additional testable implications are based on the development of a “logical tree” (Platt 1964). First, extend the logic of a theory by focusing on one branch, a particular assumption, and ask what other observable implications follow from it, in conjunction with other assumptions. For example, if democratic leaders experience higher audience costs than non-democratic leaders, then democratic leaders should be less likely to bluff; as a result, other states should be less likely to respond to challenges made by democracies, all else equal.¹⁰

Second, problematize an assumption by asking why an assumption sometimes holds and sometimes does not. In other words, create a theory that makes this assumption the conclusion. In problematizing an assumption, we extend a research program and come to better understand the robustness of the original theory.

Another exercise for cultivating theory comprehension and construction skills is the argument outline. Assign an article and ask students to outline its argument by listing the key premises and a hypothesis that follows from these premises. For example, we can summarize Huntington's (1993) clash of civilizations argument this way. Civilizational/cultural differences lead to different preferences over a variety of policy issues. Different preferences lead to conflict (policy disagreements). Cultural differences lead to conflict. With this exercise, students come to understand the concept of a theory as the logic behind a conclusion (hypothesis), understand that all arguments contain a set of premises and at least one conclusion (hypothesis), increase their ability to identify the key premises and hypothesis (or hypotheses) in an argument, and start to recognize that arguments rise or fall based on the soundness of their premises, the logical connections between the premises and conclusion, and the empirical accuracy of the conclusion. In Huntington's argument, for instance, the soundness of each premise is questionable, and the empirical record also does not seem to offer much support for the hypothesis that cultural differences lead to conflict. To increase the complexity of the exercise, ask students to include the justification given for each premise and to evaluate the logical validity of the argument.

Sometimes students are paralyzed at the thought of developing a new theory, but everything does not have to be completely new. One of the most common methods of theorizing is to take a theory from one field or on one subject and apply it to another. The creativity is in recognizing the application to the new subject. For this reason, Lave and March (1975) recommend learning some basic models,¹¹

which may be used as a jumping off point for developing other models.