By way of conclusion, building on the chapters in this book, we address some of the key issues that arise as scholars work with concepts.

Hyperfactualism and the Contribution of Concepts

Jorge Luis Borges (Reference Borges and James1962) tells the remarkable story of “Funes the Memorious,” a reclusive young man with a perfect memory. Whatever the benefits of perfect recall, such ability proves to be something of a curse. For example, Funes can (and does) spend an entire day remembering the events and observations from a previous day. That level of granular recounting and disaggregation is simply paralyzing and ultimately yields no insight.Footnote ¹

The rest of us, “blessed” with less capacious grey matter, navigate the world by lumping things, places, people, events, and ideas into categories, by pure necessity. Differences in the degree to which people categorize leads to debates about the relative contributions of lumpers and splitters. Yet once we accept the necessity and virtue of eliding differences by some lumping, a key question arises: how to improve the lumping? That question motivates the thirty-plus years of writing which is the point of departure for this book. This question is as vital today as it ever was, especially since new technological opportunities to work with massive amounts of data might dwarf Funes’ memory of one day – indeed of many days.

In Praise of The Overconscious Thinker

Even if we shudder at the absurdity of Funes’ unfiltered approach to the observable world, the advantages of a serious and systematic approach to concepts may not be obvious to many scholars. That uncertainty is a driving rationale for the research on concept analysis included in the present volume.

Yet we should acknowledge a nagging concern that haunts research on methodology generally, and concept analysis more specifically. That is, the concern that concept analysts are over-conscious thinkers, so consumed by method and terminology that they are blind to the phenomenon of interest, the “substance” that presumably animates social scientists and their audiences.

One powerful response to this concern is that questions framed methodologically are just as much substantive questions, and this may be especially true for concept analysis. For the concept analyst, the substance is always in sight – perhaps even in sharper focus than otherwise. Part of the reason, perhaps counterintuitively, is that conceptualization encourages description and further empirical digging. Setting out to categorize a phenomenon puts one up against the phenomenon itself, in ways that might not otherwise occur. Concepts simply improve our observations and how we present the observations. This connection was apparent in a humbling anecdote that Collier has recounted from his graduate student days.Footnote ² An esteemed but impatient professor interrupted Collier’s elaborate description of housing for the urban poor in Peru with the stinging challenge: “Mr. Collier, can’t you be analytic?” The interjection usefully, if provocatively, led Collier to recognize that he needed to combine his vivid description with more appropriate and compelling concepts. We see this symbiotic relationship between the abstract and the concrete in many domains and phenomena. For example, for Straus to ask whether an episode of violence in Rwanda constitutes “genocide” motivates not only conceptualization, but also examination of the case.Footnote ³

To put it another way, conceptualization begets measurement. Some acts of conceptualization engage the phenomenon immediately. We think of definitional strategies such as typologies, disaggregation, ideal types, prototypes, and ostensive definitions – as in “look, that’s what I mean by a populist” – which may be the most direct example of illuminating the world through concepts. Yet other definitional strategies are also linked to empirical observations, that is, to substance. Thinking through particular cases, and deciding which may fit one category or another, is an important part of the back and forth of building both bright-line (deterministic) definitions and prototypical (probabilistic) ones.

This symbiosis between conceptualization and measurement is evident in the focus on validity assessment, which runs through this volume. Four chapters, which involve Collier, Adcock, and Seawright in various combinations and alone, take up the matter explicitly.Footnote ⁴ Validity, of course, involves the critical question of whether one has accurately represented the concept with one’s observations. One element that is clear in the various methods of validity assessment is the reciprocal nature of evaluating the meaning of the concept and that of the indicators. Concept and measure stand unsteadily on their own, but borrow strength from one another. There is a parallel here to our earlier point about categorization and description. In the same way that cases help in the formulation of definitions, indicators help in formulating and remembering the content of concepts.

All that to say that doing concept analysis does not mean neglecting substance. On the contrary, a discussion of concepts mobilizes empirical examples and examples mobilize concepts – each of which is hard to understand without the other. In the case of concepts and substance, the “medium is the message,” to extend Marshall McLuhan.

Concepts as Knowledge Representation: Explicit Versus Implicit Knowledge

If anything, we would be tempted to go in the other direction and worry about underconscious thinkers—those who are not subjecting concepts to regular scrutiny. Our concern is that those who are not focused on concepts neglect an opportunity to document, internalize, and communicate ideas. Concepts, after all, stand for ideas, and it is hard to remember and recount ideas without them. A set of concepts, we might say, represents knowledge in a particular domain. For example, thumbing through the various “democracies with adjectives” that Collier and Levitsky unearth tells us something about contrasting forms of democracy and provokes an examination of the cases.Footnote ⁵ This sort of conceptual documentation is vital in fields such as political science, in which unlike medicine or biology, concepts are not highly regulated, negotiated, and documented. In a less regulated domain, concepts still serve to represent knowledge – it just demands more of researchers to introduce and circulate the terms judiciously.

Still, part of engaging in concept analysis means navigating between explicit versus implicit knowledge and language, a distinction that may seem unnatural for some. After all, from the time we can crawl we pick up language in an informal, highly implicit, way. It was only in the nineteenth century, for example, that it occurred to someone like Noah Webster to write down a lexicon of American English (Lepore Reference Lepore2006). Being explicit and self-conscious about vocabulary may well run against some of our basic instincts. Language evolves organically and naturally, and something that smacks of language control might strike some as overregulating the market of ideas. One way of thinking about this idea is Adcock and Collier’s notion of background versus systematized concepts. The background concept functions in a loose, “folk” state without the definition, boundaries, and measures, whereas the systematized concept has a defined empirical and definitional basis.Footnote ⁶

Being more explicit about meanings – while not regulating them – can assist the market of ideas in many ways, principally by providing infrastructure for communication. In this sense, we shouldn’t think of concept analysis as regulating the intellectual market so much as providing critical infrastructure for it – that is, a framework for developing and communicating ideas.

Integration, Translation, and Traveling

That brings us to the fundamental advantage of a focus on concepts in formalizing knowledge. A conceptual infrastructure facilitates the integration and translation of ideas among scholars and across time and space. The premise is simple. If scholars can attach names to ideas and use the same names, we can then debate, evaluate, and extend those ideas.

A research program on democracy, for example, would fail unless one could agree on terms and collect data that operationalized the concept. Michael Coppedge’s analysis of the remarkable V-Dem dataset illustrates this point.Footnote ⁷ V-Dem and its codebook are, in effect, a dictionary of democracy and its dimensions. And note once again the substantive effect of the “overconscious” thinker here: The development of various conceptual options encourages empirical work on the phenomenon. The infrastructural effect of V-Dem is evident in how its conceptual mapping has integrated research on democracy and facilitated conversations about democracy. As many of today’s polities struggle with polarized publics, we can speak with substantial precision of the degree to which they lack the “deliberative” virtues of democracy, in part because V-Dem has successfully conceptualized and measured that component. That is, the level of granularity of the data is adequate for carrying out nuanced analysis of this specific aspect of democracy.

This promise of integration is probably most obvious in the idea of conceptual traveling, which runs through much of this volume, and also through Collier’s early interest in concepts.Footnote ⁸ The basic problem is whether we can drag a concept from one context to another without sacrificing (stretching) its original meaning. In a version of the lumper versus splitter debate, scholars sometimes disagree about the degree to which a concept is temporally and spatially bounded. For example, Collier speculated that at the end of the twentieth century, corporatism in Latin America would cease to be a useful concept and an important phenomenon, only for Chartock to show its utility in understanding social movements up to the present day.Footnote ⁹

We are perhaps predisposed to stretch a concept that already has intellectual power and currency. Why else would scholars continue to use the word “institution” for almost anything and everything in politics that has some structure? Sometimes a word also has normative power, which is one of the reasons concepts can become essentially contested, according to Gallie. If democracy is both highly prized and highly multidimensional, we are less likely to reach agreement about which cases are democratic. Gallie’s ideas are evaluated by Collier, Hidalgo, and Miljanic, and are explored further in Gerring and Cojocaru’s large-N analysis, and in the Straus discussion of genocide.Footnote ¹⁰ It is not just concepts with positive valence (such as democracy) that are stretched, and contested. We see similar patterns with Kurt Weyland’s concern with “fascist,” and other epithets.Footnote ¹¹ Weyland favors treating such concepts as well bounded and he wonders, for example, whether we can call Donald Trump a “fascist” without doing violence to the concept that did such incisive analytic work in the mid twentieth century. These kinds of traveling problems are exactly what Collier and Giovanni Sartori sought to solve with their analyses.

Boundedness of Concepts

The traveling of concepts is, in important respects, related to their “boundedness.” Concepts that do not travel well suggest the possibility of a partial classification. Thus, it is the boundaries of the concept (not its setting) that may need to be relaxed if instances in other contexts do not fit neatly into a category (concept). Elkins credits Collier for introducing political scientists to insights from cognitive linguistics, which help to reconcile the boundaries of concepts and contexts. We are thinking here of solutions for problems of partial classification, including ideas of radial subtyping, prototypes, family resemblance, and other ways of thinking about categories more probabilistically. Collier builds on these foundational ideas in chapters with Mahon and Levitsky, using political science examples.Footnote ¹²

The discussion of boundedness takes various forms, including the controversial notion of binaries, which often seeps into discussions in political science regarding democracy, but which seems to have unsettled people for some time in other domains. Coppedge repeats a compelling adage: “there may be said to be two classes of people in the world; those who constantly divide the people of the world into two classes, and those who do not.”Footnote ¹³ Indeed, Weyland and Coppedge offer sharply contrasting dichotomous versus graded views of major political concepts, and Sartori had a strong preference for well-bounded concepts that yielded dichotomies.Footnote ¹⁴ Collier and Adcock address the question head on in the case of democracy, a chapter that tackles many issues in the debate.Footnote ¹⁵

One way to think about concepts that are interpreted as graded is to unpack their dimensions, and it turns out that typologies are visually and cognitively helpful for this exercise. With classic typologies, subtypes “off the diagonal” emerge quite conspicuously. Collier, Laporte, and Seawright describe best practices for those who would use typologies more seriously.Footnote ¹⁶ And others have subjected the method to evaluation in specific domains. In treatments of clientelism and post-communism, respectively, Nichter and Lussier remind us of the power of typologies in describing classes and subclasses.Footnote ¹⁷ It may be that this sort of disaggregation is not as common among scholars as it should be. Kurtz offers a corrective, and emphasizes the deliberate use of disaggregation as a tool of discovery in theorizing “grey-area” cases.Footnote ¹⁸

Future Work on Concepts

What does the future hold for analysts of concepts? It is a risky business to make predictions about information science these days. Classifying anything and everything has become a massive algorithmic process across all sorts of domains. Elkins argues that for many years, digital technologies have been in place that allow researchers to organize concepts more easily, share them, expand them, compare them, and, ultimately, align them across scholars.Footnote ¹⁹ One senses that the kinds of scholarly translation and communication problems that Sartori and others worried about can become easier to address.

It is also true that concept work is evident in research design and theorizing in ways that are not obvious. Dunning, for one, shows that concept formation is vital to the causal inference strategies that have become so important in political science in recent years.Footnote ²⁰ Lessing describes and explains how a number of sharp and incisive concepts such as elasticity and audience costs have emerged from mathematical formulations of theoretical expectations. Formal theory is, in Lessing’s keen reading, well suited to systematic conceptualization. In both domains (causal inference and formal theory), concept analysis is not always an explicit concern of scholars, yet it permeates the work in important ways.

Teaching

It is appropriate that this book concludes on a very pragmatic level, that is, teaching concept analysis, which motivates many authors in this volume. Bussell presents a systematic approach to fundamentals of the craft for students who may not yet be fully committed to concept work.Footnote ²¹ With the core examples of democracy and corruption, she works through the challenges of disentangling folk versus scientific concepts, exploring the semantic field, building typologies, and evaluating measures. Elkins,Footnote ²² on the other hand, focuses not on analyzing published texts, but rather on patterns of concept usage found in the ordinary communication of students and scholars. This includes, for example, discussing a favorite concept or explaining the focus of one’s own research.

Bussell’s and Elkins’ advice provides a crucial reminder. Through teaching concepts, much of what we know (or think we know) comes into sharper focus.