We all use concepts in our research, but as with learning to walk, our acquisition of these skills is largely incidental and subconscious – that is, tacit knowledge. In this chapter, I build on the topics covered in this volume to develop a basic structure for teaching the formation and analysis of concepts more explicitly.Footnote ¹

This structure can be incorporated both into courses on research methods, and also into substantively oriented courses. A common constraint on teaching concepts is that the subject is rarely allocated more than a session or two in a course. The recommendations of this chapter are intended to lay the foundation for conceptual thinking in students’ future work and should be applicable to either a single session or an entire course.

Two pedagogical tools, example and practice (exercises), are core to much teaching, but I argue that they are especially important in teaching things that we think we know but in fact may not know very well. Concepts are complicated beasts in that they are identified by “folk” terms with which we are often very familiar, such as democracy or corruption,Footnote ² and yet for which a broadly agreed upon definition can be elusive. Thus, graduate students may assume that the meaning of a concept is widely shared, when in fact even their peers in the seminar room may have quite a different understanding of it, let alone those in the broader discipline (or outside it). Sartori’s (Sartori, Riggs, and Teune Reference Sartori, Riggs and Teune1975) concern about a “tower of Babel” problem in political science – the counterproductive proliferation of terms and meanings – is certainly salient today. Hence, training students to work effectively with concepts can maximize clarity and understanding and, not incidentally, contribute to students’ future effectiveness as researchers and teachers.

Motivating the Focus on Concepts and Concept Analysis

Concepts are ideas that we use for ordering and organizing discussions about the world. To do this successfully, we must define and measure concepts clearly. The resulting clarity allows us to talk about ideas in a coherent manner and cumulate knowledge about specific topics. I suggest that careful work with concepts in the context of a research project includes four basic steps:

1. (1) Formulation and definition of the concept.

2. (2) Discussion of its place in the broader conceptual field – that is, the connection with related concepts.

3. (3) Operationalization – to define the descriptive indicators that can be used to determine whether a concept is present empirically. This iterative, multistep process involves researchers’ evaluating potential indicators and assessing their matchup with the concept being considered. This potentially can involve teasing out multidimensionality, and even modifications in the concept itself.

4. (4) Measurement – the application of specific empirical indicators to evaluate the components of a concept and generate data.

If indicators appropriate to a given study have previously been established, step 3 will probably be skipped. Yet for teaching, all four steps should be emphasized.

Each of these tasks can be introduced briefly to begin a discussion of concepts. This will help students see the close relationship between developing the idea and characteristics of a concept while also tackling the practical aspects of identifying indicators and techniques for measurement. Rather than a straight path from concept to measurement, these tasks often involve an iterative process wherein a researcher discovers assumptions they have made or gaps in their definitions that emerge during efforts to operationalize the concept. This process then leads to a reevaluation of the concept itself, as well as possible techniques for measuring it. Introducing concepts in this holistic, interactive manner should lead to both richer conceptual discussions and increased validity in measurement.

Once these basic tasks of concept analysis are laid out, examples of concepts serve as the first pedagogical tool to explore. Here, it can be useful to begin with students’ own interests, rather than jumping to examples of established concepts in the discipline. This allows students to grapple with the intricacies of concept analysis on their own, before engaging in a discussion of topics in other work. Their own experiences then provide a foundation for better examining and understanding the issues faced in conceptual analysis more generally. The very act of prodding students to articulate the concepts that run through their topics of research is enlightening for them and provokes conceptual thinking. And, of course, to name examples pushes one to understand what concepts are and how they vary.

A useful assignment for this type of practice is a short, written conceptual analysis on something relevant to a student’s research interests. This should include attention to all four of the basic steps in concept analysis and can be motivated with these and related questions:

• What is the “folk” use of the concept in daily (nonacademic) life?

• How, if at all, has it been used in social science research to date?

• Are the folk and scientific uses similar? If not, how do they differ?

• What limitations, if any, exist in past work and why? These could be related to definition, operationalization, and/or measurement.

• How do you plan to use the concept?

• With this usage, how does the concept relate to other associated concepts in the field?

• How does the concept stand up to Gerring’s (Reference Rose-Ackerman1999) criteria for “what makes a concept good”?

• What are the different attributes that make up the concept?

• What are some possible indicators – existing or proposed – that you could use to measure these attributes?

Building Blocks: Concept Formation and Typologies

Contending with the complexities of their own research helps to open students’ minds to the issues raised in conceptual work more generally, but it may introduce only certain types of concerns, depending on the particular concept. This is where examples can play a powerful role in broadening their understanding. This section shows how concept formation can be advanced through the construction of informal typologies, which can help with understanding the multidimensionality of many concepts.

Examples of concepts already in use can be highly instructive for illustrating the four basic steps in concept analysis, as well as for evaluating the quality of a concept as a whole. Here, I highlight these steps with two important concepts in political science: democracy and corruption. Intriguingly, the meaning of democracy is often contested, whereas that of corruption is not – and it can be productive to press students to figure out why.Footnote 3

In considering specific examples, students should be encouraged to evaluate both the definition of the concept and its relationship to other concepts. That is, mapping the “semantic field” (Sartori Reference Sartori1984). The three fundamental elements of a conceptual definition are: (1) the events or phenomena to be defined, (2) the properties or attributes used to define, and (3) the label for the concept (Sartori Reference Sartori1984; Gerring Reference Gerring1999). Once defined, the concept can be compared with other relevant concepts in the field, which is typically achieved through use of conceptual hierarchies. Two types of hierarchies are frequently used in political science: kind hierarchies, as described by Sartori (Reference Sartori1984), and part–whole hierarchies, as discussed by Collier and Levitsky.Footnote ⁴ With kind hierarchies, concepts at a lower level of the hierarchy retain all the characteristics of the root concept but are differentiated by the addition of more specific characteristics. For example, presidential and parliamentary democracies both fit all of the “procedural minimum” characteristics of the root concept “democracy” (as discussed later) while exhibiting additional features that differentiate their executive/legislative institutions. In part–whole hierarchies, by contrast, lower-level concepts possess some but not all the characteristics of the root concept, resulting in “diminished” subtypes.Footnote ⁵ A diminished subtype of democracy might be missing competitive elections, while another lacks universal suffrage. This attention to vertical hierarchies and horizontal types will help students establish where their concept of interest fits in the broader conceptual field.

Democracy

These techniques can then be applied to prominent examples in the field. For example, what is “democracy”? It may be sufficient in some contexts to define democracy as “rule by the people,” but for empirical social science, a further elaboration is necessary for what “rule” means and who “the people” are. Within political science, many analysts have settled on conceptualizations that are procedural and in general “minimalist,” as in “procedural minimum definition,” emphasizing the multidimensional character of this concept and the minimal number of procedures that must exist for a case to be considered a democracy.Footnote ⁶ Significant procedures that may be included are universal suffrage, free and fair elections, electoral competition, and protection of civil liberties. Students can debate whether specific definitions of a concept seem complete and how the definition establishes the relationship to other, parallel concepts.

Yet even with a minimal definition, analysts may argue that (1) additional features are necessary for fully representing the concept of democracy, or that (2) different mixes of the “minimum” procedures can be appropriate. This suggests that typologies may also be relevant to understanding the concept of democracy, in that they can be used to map out the relationship between different dimensions and different types. An important example in conceptualizations of democracy is the two-by-two typology produced by Dahl’s (Reference Dahl1971) focus on contestation and participation.

Corruption

Another concept useful for instruction is “corruption.” As with democracy, it is often defined in minimalist terms as “the abuse of public office for private gain.” But in contrast to democracy, this definition is hardly debated at all, to the detriment of our understanding of corruption (Bussell Reference Bussell, Rose Ackerman and Lagunes2015). I have argued that the frequent failure to interrogate the characteristics falling within this definition “makes it largely, if not entirely, impossible to determine whether the corruption considered in one analysis ‘should’ be the same or different from that evaluated elsewhere” (Bussell Reference Bussell, Rose Ackerman and Lagunes2015: 24).

Typologies can again play a useful role here, given their contribution to mapping dimensions and types. They allow the researcher to build detail into the definition of the concept in a way that is appropriate to the specific research question. Past work has differentiated corruption on multiple dimensions, for example, petty versus grand corruption (Heywood Reference Heywood1997; Rose-Ackerman Reference Rose-Ackerman1999; Basu Reference Basu2011), retail versus wholesale, harassment versus nonharassment, and local versus national. Heywood (Reference Heywood1997) identifies still other dimensions upon which typologies can be developed, but I posit that lack of attention to when a particular typology should be used has impeded the accumulation of knowledge on corruption in general (Bussell Reference Bussell, Rose Ackerman and Lagunes2015: 23–24).

The solution I propose is to develop typologies of corruption tailored to specific research projects. Where possible, they should build on existing work, yet also draw on the specific characteristics of corruption relevant to the analytical question at hand. For example, in work concerned with identifying constraints on anti-corruption reforms, I focus on control over resources as the key dimension of corruption to consider and differentiate between types of corruption based on who has control (direct or indirect) over what type of government resource (Bussell Reference Bussell, Rose Ackerman and Lagunes2015: 37–41). This allows for a more specific consideration of which actors within government might resist reforms to corruption related to the distribution of which resources. While this distinction may not be useful for all analyses of corruption, establishing whether it is useful becomes more straightforward with a clear delineation of the assumptions and logic behind the types. Table 23.1 summarizes a set of examples for how “corruption” has been typologized, depending on the goals of the researcher. Instructors can use these examples to encourage students to explore the conditions under which different variants of a concept would be useful for addressing a specific research question.

This presentation of existing corruption typologies also shows that much work in this area has focused on only a single dimension or characteristic, which might result in an underrepresentation of the larger concept. While this kind of simplification can be useful for analysis, it may also force important substantive differences between types of behaviors into a single type, thereby limiting what we can infer from the results of an analysis. Thus, attempting to distinguish between multiple features within a definition could help to clarify our thinking on what differences within a concept may be theoretically relevant while also improving the match of measurement techniques to overall concept.

To illustrate with one hypothetical example, I explore in Table 23.2 how we might combine existing distinctions between petty and grand corruption with the differentiation between whether the bribe-giver is eligible for the service/good or not (otherwise known as harassment versus nonharassment bribe). The latter distinction, eligibility, might help us to evaluate, for example, the implications of bribes paid by multiple viable contractors to get a government contract, versus bribes paid by contractors who are actually unable to fulfill the contractual demands in a satisfactory way. The interaction of these dimensions of scale and eligibility yield interesting categories that lead to a richer description of the phenomenon. Thus, considering dimensions jointly, as typologies, can be powerful.

Having reviewed these different models of definition and typologies, one may also return to Gerring’s (Reference Gerring1999) criteria for establishing “what makes a concept good,” by applying his framework to different variations on the concepts of democracy and corruption.

Untangling Concepts

Understanding when an instance or a case no longer fits the concept under consideration is also a key issue.Footnote ⁷ Examples again provide an important tool for teaching by elucidating strategies we can use to untangle concepts. Two related issues often raised in discussions of corruption are relevant here. First, how do we think about something that is considered “corruption” in one place, but is legalized in another context? And second, can legal behavior that “corrupts” a system be considered corruption?

In the first case, consider a typical act of corruption: A bureaucrat accepts a bribe from an individual to provide a service more quickly than they otherwise would. This instance of “speed money” is common in many parts of the Global South, but largely unheard of in the Global North. Yet legalized speed money, in the form of extra payments to receive government services, such as a passport, more quickly is common in the Global North. The difference is these funds go into the official receipts of the government, rather than the pocket of the individual bureaucrat. When this act is legal, there is no longer abuse of office by a single actor, but there remains variation in who can access services quickly.

This question leads us to the second case, in which legal acts or influences are seen to corrupt the functioning of an institution. This “institutional corruption” is seen to exist where there is a “systemic and strategic influence which is legal, or even currently ethical, that nonetheless undermines the institution’s effectiveness,” its ability to achieve its purpose, and its trustworthiness (Lessig Reference Lessig2013b). Where, for example, politicians are dependent on donations for their political success, this can conflict with their institutional role as representatives in a democracy (Lessig Reference Lessig2013a).

In both cases, the acts that we have traditionally considered “corruption” are those that are explicitly illegal forms of private gain, even though the question of legality is not made explicit in the standard definition of corruption. Perhaps defining corruption as “the abuse of public office to engage in illegal activity for private gain” would add clarity to our analyses. In this sense, to untangle different, but similar, forms of behavior, it is necessary to add specificity to our definition. This also allows us to distinguish between those acts that, in the case of corruption, are likely to have different political and institutional ramifications.

Students can be asked to consider which practices or behaviors an existing conceptual definition would “allow for.” Do these other cases fit with what we generally assume to be instances of the concept? Does including or excluding a case change what we see as the fundamental characteristics that must be present in the definition?

Operationalization and Measurement Validity

Once defined, how do we operationalize and measure a concept? Operationalization is a process of identifying the indicators that one will use to evaluate the degree to which the concept is relevant in a given context. For the procedural minimum definition of democracy discussed here, this would mean identifying which indicators of those procedures are sufficient for establishing the presence or absence of democracy. This could include suffrage laws and their implementation; election rules and regulations; the presence of multiple viable candidates and/or political parties; and legal protections and their enforcement for freedom of association, speech, and assembly.

In a classroom setting, students can be asked to consider what is necessary and sufficient in mapping the definition of a concept to specific indicators. For example, is it necessary to include indicators of each attribute of a multifaceted concept? Are certain attributes more fundamental than others to the definition?

It is then important to develop a strategy for measuring the relevant attributes. Measurement involves several considerations fundamental in research design, including practical concerns related to accessing useful data. A measurement plan must include a description of the proposed measure(s) and sources, such as administrative data, a survey, or news reports. For each indicator, there should also be consideration of whether a single measure or a composite index will be used.

This raises the important topic of measurement validity, or the degree to which our operationalization and measurement of a concept adequately reflect the content of the concept in question.Footnote ⁸ Consider an extreme minimum definition of democracy as universal adult suffrage. In this case, the operationalization includes only one indicator: inclusive suffrage. However, as Paxton (Reference Paxton2000) points out, most empirical studies of democracy prior to 2000 used a measure of adult male suffrage. As a result, many countries were counted as democracies long before they achieved the key characteristic identified in even a narrow operationalization, thereby bringing into question any analysis that drew conclusions based on this invalid measure. Instructors should encourage students to revisit research in their area of interest to evaluate the match between concepts, specific defining attributes, and measures.

Similar issues arise in the measurement of corruption. While a minimum definition seems useful for comparing across studies, in practice it has resulted in defining corruption in similar ways, while implementing dramatically different operationalization and measurement strategies. Consider these very different corruption measures, extracted from recent work:

• External “expert” perceptions of corruption as measured by Transparency International (Glaeser and Saks Reference Glaeser and Saks2006)

• The accumulation of unpaid parking tickets by diplomats in Manhattan (Golden and Picci Reference Golden and Picci2005)

• Estimated changes in elected politicians’ wealth (Fisman and Miguel Reference Fisman and Miguel2007)

• The difference between built infrastructure and funds allocated to building (Bose et al. Reference Bose, Capasso and Panini Murshid2008)

• The number of officials convicted for corrupt practices (Bhavnani Reference Bhavnani2009)

Each of these measures is intended to evaluate whether public actors are abusing their offices for private gain. Yet they are so diverse as to be largely incomparable – and one might in fact find low correlations among indicators (Bussell Reference Bussell, Rose Ackerman and Lagunes2015). While each measure may indeed be an indicator of corruption, the sharp contrasts among them highlight both issues in the translation from concept to measurement and difficulties in comparing findings across research ostensibly studying the same concept.

Pulling It Together

A final exercise for motivating students is to return to their original conceptual analyses and revisit the initial conclusions in light of subsequent discussions. Are students satisfied with their initial analysis? Are there nuances that they missed in the first pass, but which would be important for their work? How might they apply each topic covered here to the evaluation and measurement of their concept?

Engaging in teaching concept analysis also allows us to demystify the idea of concepts, while grounding it in specific guidance for rigorous analytic work. Thus, in addition to the tools discussed earlier, we can encourage a conceptual mindset that values the quality of an analysis. This enables researchers to recognize that:

• It is not necessary to generate a “new” concept to do meaningful work. Rather, it is necessary to define the concept in the clearest and most parsimonious way possible, and to consider how each student’s usage relates to previous uses in the literature.

• If students are using a typology, ensure that it clearly reflects the conceptual foundation they have developed and is substantively appropriate to the research question (typological completeness) (Bussell Reference Bussell, Rose Ackerman and Lagunes2015).

• The indicators for a concept must incorporate the full range of characteristics identified as important for the specific research question (concept representation).

• “Good” data sources do not equal good measurement. Measures are only good (i.e., useful for analyzing a question) if they reflect the underlying assumptions and characteristics formulated in the broader process of operationalization (measurement validity).

Concept analysis can be fun, frustrating, inspiring, and difficult all at once, much like the rest of our research endeavors. But it is the core of what we do and of our ability to cumulate knowledge about the world around us. Empowering our students with these tools is one of our most fundamental tasks as advisors, and it will pay dividends for many years in the exciting research it enables.