In 1978, James Scott began two years of participant observation fieldwork in a small rice farming village in the Kedah region of northwest Malaysia. Scott first went to the village to conduct research on class relations, observing the interactions between the village’s landowning class and peasant farmers. As he later recounted in an interview, “I wanted to study class relations. Ironically, instead of ending up in a place where a revolution was in the making, which obviously would have made me quite happy, I ended up in a village where basically nothing was happening” (Scott Reference Scott, Munck and Snyder2007, 363). Although the dynamics and outcomes Scott was hoping to find were not present, he did notice more “subterranean forms” of class conflict. These hidden acts of resistance by the peasant class, including foot dragging and evasion, were not indicators of a revolution. Nonetheless, their existence was at odds with some of the leading conceptualizations of class relations at the time.
The popular understandings of class relations during that period often entailed an ideological domination of the subaltern class because “everybody and their brother was reading Gramsci and Althusser” (Scott Reference Scott, Munck and Snyder2007, 363). Yet Scott noticed that these Western scholars’ assumptions of ideological domination could not account for the type of symbolic struggles that the peasant farmers very consciously engaged in on a regular basis. As he later argued, class relations also involved everyday forms of resistance, which in his 1985 book he called “weapons of the weak.”
In this example from a now well-known text, the author goes to the field to figure out how a key concept in the literature, class relations, works in practice. On finding that the dominant understandings of the concept are inconsistent with their observations in the field, the author proposes a different conceptualization, one that incorporates their findings. Had Scott followed mainstream methodological advice, he might have completed a study on the reasons for a lack of peasant revolution, thrown out this early data and found a different case, or accepted Malaysian class relations as a deviant case. Instead, Scott engaged in theory-based research that explicitly acknowledged and challenged the assumptions underlying the prevailing literature. The very first sentence of the book’s preface presents his approach: “The limitations of any field of study are most strikingly revealed in its shared definitions of what counts as relevant” (Scott Reference Scott1985, xv). Why, Scott asks, have peasants only been studied when they are being organized by outsiders into revolution and otherwise been viewed as anonymous contributors? Perhaps, he suggests, “social scientists in the West” look for the types of formal organizing they most expect to find based on their own experiences and understandings. In doing so, these scholars overlook the perspectives of subordinate groups, which often do not have the power to engage in open, organized politics (xv).
As social scientists, we hold assumptions and experiences that shape how we view the world. These biases inevitably affect theorizing in our discipline. As Scott’s experience shows, outdated or unexamined assumptions lead to both political and methodological problems. Politically, they can create inequities within theory, for example, in the relationship between Western social scientists and research subjects in the global south. Methodologically, they can lead to empirically deficient explanations and concepts.
Generalized concepts in the literature rarely align exactly with observations in the field. Yet, as we argue, for researchers who come across significant discrepancies between theory and practice, the existing options for challenging and revising concepts are insufficient. Even in Scott’s case, he did not initially set out to upend the dominant conception of class relations, which had assumed that the subaltern were subject to ideological domination in their day-to-day lives. Yet his observations in the field reflected significant differences from this widespread assumption about how class relations worked.
In this article, we argue that this type of theory-based research that Scott and other scholars have used is distinct from either theory building or theory testing. It is neither about the construction of a new theory or concept nor about falsifying an existing theory. Instead, this approach is specifically based around existing concepts. It focuses on the “reconstruction” of existing concepts based on empirical observations for the purposes of challenging the literature. We therefore describe this approach to concepts as theory reconstruction (Meade and Walton Reference Meade and Walton2024).
Theory reconstruction is an abductive, or “puzzle-based,” approach to research that challenges the use or understanding of a key concept in a given context. Instead of understanding concepts as uniform, fixed phenomena, we draw on Wittgenstein’s “family resemblance” approach to concepts, which we see as an important alternative to the more mainstream advice on conceptual innovation. In this article, we identify four strategies of theory reconstruction used in existing scholarship: revising features or mechanisms of a concept (revision), applying an existing concept in a new or unstudied context (extending), dividing a concept into distinct subcategories (narrowing), and unsettling a spectrum or common dichotomy of concepts (disrupting). We outline our approach in detail while examining the limits of the conventional wisdom on conceptual innovation and the implications for the discipline.
Rethinking Conceptual Innovation
Today’s mainstream advice often discourages students from using their data and observations to revise concepts. We experienced this in our own research when, as graduate students, we were faced with limited guidance after we found that our observations in the field contradicted key assumptions in the literature. Take, for instance, the research design text by King, Keohane, and Verba (Reference King, Keohane and Verba1994). The authors acknowledge that it is uncommon for data at the end of the process to perfectly match the theories hypothesized at the beginning. Despite this, they caution students against revising theories and concepts based on their data, warning that such adjustments should be done “rarely and with considerable discipline” (21). Similarly, Giovanni Sartori (Reference Sartori1970) famously warns against “conceptual stretching,” arguing that observations of new cases should not mean a revision of the original categories. Gerring and Seawright (Reference Gerring and Seawright2022, 43) do offer ways of innovating an existing concept, including by redefining it or reclassifying its subcategories. However, they also include a caveat that conceptual innovation should be used sparingly, so as not to “steal attributes from neighboring concepts” or, as Sartori (Reference Sartori1970, 42) once put it, to avoid “playing musical chairs with words.”
Following this advice, students are taught that if their empirical observations do not fit the literature’s understanding of a specific concept, they should define it as an alternative phenomenon or “deviant case” and avoid amending the general definition of the concept unless doing so is absolutely necessary. This commonly used approach understands concepts as being strictly defined phenomena with rigid boundaries to distinguish among them. From this perspective, conceptual innovation is about removing ambiguity and sharpening the borders that define what does and what does not constitute an example of a concept.
However, this formulaic advice stands in tension with real-world practices in the discipline. For example, Kapiszewski, MacLean, and Read (Reference Kapiszewski, MacLean and Read2022, 2) find in a survey of APSA members that most political scientists who do fieldwork engage in an iterative process between data collection and analysis that “encourages original conceptualization.” Yet these same scholars later presented their research in a more linear way because of the disciplinary norm against making such changes. Furthermore, the emphasis on limiting conceptual innovation, as well as the use of descriptive inquiry more broadly, can discourage researchers from contesting findings and, as a result, reproduces biases in the discipline (Holmes et al. Reference Holmes, Guliford, Mendoza-Davé and Jurkovich2024).
This conventional approach to conceptual innovation often encloses key concepts and seeks to avoid “unsettling the semantic field” (Collier and Levitsky Reference Collier and Gerring2009, 445), based on the assumption that the insights of (largely) Western social scientists are of general applicability. Yet as scholars like Becker (Reference Becker1998) and Schaffer (Reference Schaffer2015) argue, concepts presuppose specific experiences or perspectives. Put another way, although the field of political science is global, many of our key concepts, as well as their underlying theories and assumptions, are parochial. This critique of the discipline has long been articulated by scholars. For instance, Hanna Pitkin (Reference Pitkin1967) captures this issue in her seminal work on political representation. She asks us to imagine a complicated, three-dimensional object sitting in a dark room, with different theorists photographing it from different angles and then arguing with one another because they each mistake their partial view for the whole object. Instead of adding one more competing definition, she argues that the literature should examine each definition according to the “context for which it is correct and [by] exploring the assumptions and implications imposed by that context” (11).
More recently, scholars of Indigenous politics have criticized American political science in similar ways, arguing that the field’s understanding of many key concepts centers some “partial views” at the expense of others. For example, Bruyneel (Reference Bruyneel, Lucero, Turner and VanCott2013, 1) asserts that the discipline’s consistent failures to “place indigenous politics in its field of vision” is motivated in part by its emphasis on rigidly bounded categories. In particular, the author highlights how political scientists often locate the concept of sovereignty within states and territories, even though Indigenous nations and tribes understand sovereignty as an inherent self-governing power of the people. As a result, forms of Indigenous governance, decision making, or politics that do not fit the “statism” of the discipline often go unacknowledged or misunderstood. Along the same lines, Grande (Reference Grande2015, 66) points out that dominant conceptions of democracy and its connections to citizenship are rooted in the colonial history of property and title, which conflicts with Indigenous communities’ connections to land. As Bruyneel (Reference Bruyneel, Lucero, Turner and VanCott2013, 7) notes, these are not only political problems but epistemological ones as well. For instance, extending the concept of sovereignty beyond the state to better encapsulate Indigenous politics means going against the discipline’s focus on “fixed political categories.” Other Indigenous scholars like Wilkins (Reference Wilkins2016, 1049) have made similar criticisms of American political science, arguing that “most work is confined within such rigid parameters; it becomes overwrought, self-referential, and devoid of real-world application.” Although some might argue that rigid definitions can clarify, these scholars suggest that uniform conceptualizations also confine the use of key concepts and, in effect, reproduce the marginalization of communities and perspectives that do not fit cleanly within the dominant conceptual framework.
Within the context of American politics, Michener, SoRelle, and Thurston (Reference Michener, SoRelle and Thurston2022) have made related observations, in their case about the concept of the welfare state. The authors point out the tendency of political scientists to study the welfare state “from the vantage point of policy makers and elite institutions” while failing to value the lived experiences of the ordinary people who rely on welfare programs and policies (155). Building on this point, they highlight two dimensions of the welfare state that political scientists tend to overlook—civil legal assistance and consumer credit—both of which are critical for understanding how low-income Americans secure access to basic needs, such as housing. The authors do not seek to re-create a new definition of the welfare state but instead call for a bottom-up experiential view that reimagines how we understand the concept, particularly from the perspective of marginalized groups navigating the welfare system.
These and other examples suggest the need to rethink the dominant approach to conceptual innovation in the discipline and to encourage scholars to challenge key concepts through new understandings and perspectives. Luckily, there is an alternative approach to concepts that encourages this type of innovation. Instead of viewing concepts as abstract terms with rigid definitions that social scientists impose on their observations, one can also understand concepts as part of everyday language used by real people in specific contexts (Schaffer Reference Schaffer2015). The main implication of this approach is that key concepts such as “democracy,” “citizenship,” or “states” are not understood uniformly across different societies and are therefore subject to interpretation and social construction. Working with concepts in this way means accepting that what political scientists often refer to as a single, fixed concept is really a collection of related social phenomena whose meaning is informed by context and perspective.
This approach to concepts has a basis in what Wittgenstein (Reference Wittgenstein1953) calls “family resemblances” and has long been used in political science (e.g., Collier and Mahon Reference Collier and Mahon1993; Schaffer Reference Schaffer1998; Wedeen Reference Wedeen2002). Instead of concepts being defined by necessary and essential features, understanding concepts as “family resemblances” means that different uses of the term are connected by a “network of partially overlapping similarities” (Pitkin Reference Pitkin1985, 64). The emphasis here is on understanding the contexts in which concepts are used and the various overlapping ways they operate in practice. Theory reconstruction builds on this idea of family resemblances and understands concepts not just as terms that need to be strictly defined but also as social phenomena existing within contexts that can and should be revisited. The purpose of theory reconstruction that we describe here is not about rewriting conceptual definitions by sharpening their existing boundaries but about opening concepts up to be challenged from new perspectives. With this rethinking of conceptual innovation as the underpinning of our approach, we now turn to the specifics of theory reconstruction, which we present as one way in which scholars might engage with key concepts differently in their research.
What Is Theory Reconstruction?
“Theory” can, of course, mean different things. For instance, one might be defining the relationship between variables, explaining why a specific phenomenon occurred, or showing how a defined phenomenon is experienced and understood in a specific context (Abend Reference Abend2008). In this article, we focus on the third meaning. Our interest is in theory not as a general proposition or a causal explanation but instead as an account of the processes or the inner workings of a distinct social science phenomenon. In practice, social scientists engaging in this type of theorizing often capture their observations using specific concepts. For example, Scott was observing interactions between the peasantry and landowning class but captured these observations more broadly under the concept of class relations. Likewise, other scholars we mention in this article package their observations as a challenge to specific key concepts in the relevant literature.
Although concepts are the building blocks of social science, they also rely on theorizing. As Gerring and Seawright (Reference Gerring and Seawright2022, 42) point out, all concepts consist of a term, a definition, and a “set of phenomena identified by the definition.” Thus, when we use concepts, we are often observing and describing their underlying phenomena. For instance, regardless of how one defines “democracy,” studying it in practice means identifying certain processes and actions by connecting one’s observations to a corresponding term, such as participation, contestation, or the rule of law. In other words, translating empirical observations into more abstract concepts involves a critical form of theorizing. We think of concepts not as something static; instead, agreeing with Goertz (Reference Goertz2006, 5), we understand that “concepts are…theories about the fundamental constitutive elements of a phenomenon.” In this way, concepts are multidimensional, meaning they often have theories, hypotheses, and causal arguments embedded within them.
The scholarship we identify as engaging in theory reconstruction focuses on the tension between key concepts as understood in the literature and how they appear in the field from an unorthodox or understudied perspective. Building on this, we refer to theory reconstruction as an “abductive” approach to research. By abductive, we mean research that “begins with a puzzle, a surprise, or a tension” that one encounters between theory and empirics, where the researcher seeks to resolve this immediate tension by identifying “the conditions that would make that puzzle less perplexing” (Shwartz-Shea and Yanow Reference Schwartz-Shea and Yanow2012, 27). Unlike both deductive and inductive approaches to research, abduction tends to proceed in a nonlinear fashion, involving an iterative back-and-forth between theory and empirics (Shwartz-Shea and Yanow Reference Schwartz-Shea and Yanow2012). As Timmermans and Tavory (Reference Timmermans and Tavory2012) argue, by beginning with a puzzle based on existing literature and empirical observations, an abductive approach can lead to the development of new concepts as part of a process of “theory construction.” But what happens when that puzzle is about an existing concept in the literature? Here, what may be necessary is neither a construction nor falsifying of a concept but instead a reconstruction.
For instance, in the Scott example, his research was ultimately about resolving a tension between a common understanding of class relations in the literature (which relied on Gramscian notions of hegemony) and the subversive actions of the peasantry he was observing in the field. He was not arguing that all theories of class relations should be redefined. Instead, he asserted that a revised conception was necessary for understanding how class relations worked in that context, from the perspective of the peasantry. For the most part, the authors we draw from here are not primarily interested in modifying the core definitions of concepts or establishing general principles. They are instead trying to adapt key concepts to an immediate, “situated” puzzle, often from an unorthodox perspective (Shwartz-Shea and Yanow Reference Schwartz-Shea and Yanow2012). This does not mean that these authors believe their observations are only applicable to that specific case, however. Reconstructions can result in revised theories and new research agendas. For example, Scott’s identification of “hidden acts of resistance” has been used, more generally, to rethink power from the subaltern perspective. However, the external validity of the authors’ findings is not a given and can vary case by case.
Lastly, given the focus on concepts, it is worth clarifying why we call our approach “theory reconstruction.” First, we adapt the term from Burawoy’s (Reference Burawoy1998) work on the “extended case method.” Burawoy introduces this term to describe his research in Zambia, which he uses to challenge the concept of underdevelopment. At the time, scholars had attributed the underdevelopment of some countries to cultural backwardness, state weakness, or the influence of neocolonial powers. Burawoy chose to study the topic “from below,” using participant observation as a worker within a Zambian mining company shortly after the country gained its independence. As part of what he calls “theory reconstruction,” he shows the discrepancy between his empirics and the literature’s explanations for underdevelopment. In particular, he highlights the way domestic elite alliances between white management and the new Black political elite undermined the mining industry’s impact on development. Importantly, Burawoy (Reference Burawoy1998, 8) emphasizes that he was only able to engage in this form of theory reconstruction because he observed the industry “from the standpoint of the vast majority of unskilled and semiskilled workers.” Although for Burawoy theory reconstruction was just one part of a broader framework for reflexive social science, in this article we adapt the term to describe a specific approach to research.
Moreover, we use the term “theory reconstruction” to avoid confusion with an alternative term sometimes referred to as “reconstructing a concept” (Collier and Gerring Reference Collier and Gerring2009; Gerring and Seawright Reference Gerring and Seawright2022; Schaffer Reference Schaffer2015). This typically involves sharpening or changing an existing concept’s definition based on representative studies and carries with it an emphasis on precise conceptual boundaries and distinctions. Again, this type of conceptual work is important and is necessary for certain studies; yet, it is squarely within the same mainstream approach to fixed concepts that we are trying to rethink. Thus, given that our approach is more similar to what Burawoy (Reference Burawoy1998) describes, and to avoid any confusion with alternative uses of “reconstruction,” we embrace the term “theory reconstruction.”
In this article we cite several authors (of both recent and long-standing scholarship) whom we identify as having used different versions of a theory reconstruction approach but without specifically naming it as such. In each example, the scholars use empirical insights to challenge the meaning or inner workings of key concepts in their respective fields. At its core, the literature that we highlight as having used this approach shares several elements in common. First, the authors identify a “thick” (Coppedge Reference Coppedge1999) or “multidimensional” (Goertz Reference Goertz2006) concept that is important to understanding their field observations and is central to the relevant literature. As they work with their concept, scholars are attentive to how disciplinary norms and power dynamics have structured its historical development and contemporary usages.
Second, the authors observe that the concept works differently in practice than how it is conceptualized in the literature. In the examples we highlight, the authors observe evidence and outcomes that are in direct contradiction to how a specific concept was characterized in previous scholarship. This discrepancy can be at the level of mechanisms, the amount of variation in different instances of the concept, or how the concept is applied in a specific literature. The author’s observation is not a definitive account that discredits or falsifies other uses of the concept, but it should at least challenge important assumptions in the literature’s conceptualization. Of course, there are instances where a concept, and its underlying theory, simply just does not work. An example of this is Menchik’s (Reference Menchik2017) discussion of secularization in which, over time, an outdated key assumption about the concept was shown to be incorrect. Additionally, in cases of less prominent theories, a concept with little empirical backing and little influence in the field is likely not worthy of theory reconstruction.
Lastly, having identified the concept and the discrepancy, the researcher resolves this tension by reimagining or “reconstructing” the concept as it exists in the literature in a way that is more consistent with their observations. This can take different forms: revising mechanisms and components of the concept, expanding the scope of the existing concept to incorporate new contexts or phenomena, narrowing an overly broad concept into different subtypes, or disrupting an existing dichotomy or spectrum of concepts. In practice, because of the continual navigation between theory and empirics, theory reconstruction often follows nonlinear, iterative research paths (Kapiszewski, MacLean, and Read Reference Kapiszewski, MacLean and Read2022). Nevertheless, for the sake of simplicity, we outline theory reconstruction in three steps:
-
1. Establish the discrepancy. The researcher closely examines sources of data in an area of interest and compares these observations to the major literatures in that area, identifying any significant discrepancies between how a key concept is understood in the literature (how it works, what it looks like, when it is applied, etc.) and how that concept appears in the researcher’s observations. This discrepancy between the literature and the empirics forms the “puzzle” of the research question (e.g., If the literature assumes a concept to work a certain way in theory, then why might we observe something different in practice?)
-
2. Identify the source and revise the plan. Taking a reflective approach, the researcher pinpoints where the dominant conceptualization breaks down. For example, where do the concepts’ mechanisms no longer match the empirics? What is missing from the concept that would make it more useful from the point of view one is observing? Researchers then make adjustments to their methods based on this analysis; for example, by gathering new data or incorporating new cases.
-
3. Reconstruct the concept. The author then uses the new data to develop a conceptualization that would reconcile this discrepancy with either revised mechanisms, a more expansive application, new subtypes, or an unsettling of a common distinction between terms, all of which challenge the literature’s current understanding of the concept.
Our approach is far from the first that has thought about changing the way social scientists engage with concepts. In the following, we outline seminal and newer works that demonstrate the problems of traditional approaches to conceptual development, as well as similar research that we build on.
Reexamining Conceptual Foundations of the Discipline
As we described, the standard advice on conceptual innovation in American political science discourages scholars from revising concepts based on their data. It would be simple enough to attribute this advice to a desire to unify the literature’s vocabulary by sticking to conceptualizations held in common. Yet there are reasons to believe that, historically, this advice has been used to shield the parochial nature of the discipline. For instance, if we revisit Sartori’s (Reference Sartori1970, 1034, 1050) famous advice against conceptual stretching, he argues that key political science concepts are part of “the vocabulary of politics which has been developed over millennia” from a “distinctive Western experience.” Yet instead of this being a justification for challenging the use and understanding of these key concepts, he argues that scholars should avoid “straining” concepts by applying them to “primitive, diffuse polities” (1034). In Sartori’s view, these “diffuse polities of the Third World” are fundamentally distinct from Western ones, being typified by “devious, overly sedimented, ‘non-rational’ structural patterns” (1052). Using these cases to further revise the original concepts would risk feeding “primitivism and formlessness into non-primitive settings” (1052). Sartori’s justification dovetails with modernization theory, a once-dominant research paradigm that has been critiqued for its Eurocentrism and the lack of empirical support for some of its key hypotheses (Geddes Reference Geddes2003). The point is not that Sartori’s concerns about conceptual stretching should be wholly disregarded. But like concepts themselves, our discipline’s approaches to concepts and innovation should be open to reexamination from different perspectives.
As scholars have pointed out, normative biases of social scientists get embedded into the concepts we use and influence who and where we study, who can contribute to our conceptualizations, and what constitutes evidence (Becker Reference Becker1998; Holmes et al. Reference Holmes, Guliford, Mendoza-Davé and Jurkovich2024; Thomas Reference Thomas2022). American political science, particularly comparative politics, was founded on ideas and arguments largely developed in Europe and the United States during the early twentieth century that were modified by scholars from similar Western institutions in subsequent decades. Along similar lines, some scholars have recently called attention to the role of race and race science in the founding of American political science (Blatt Reference Blatt2018). Others have noted the connections between political science and the Cold War, as well as US imperialism (Shilliam Reference Shilliam2021). These critiques raise concerns not only about the types of questions or assumptions scholars use in the literature but also where our conceptions come from and who can contribute to knowledge production.
Whether we realize it, many of the categorizations and concepts we use today have roots in unexamined assumptions, such as the idea that the idea that the poor, societies in the global south, younger democracies, or nondemocracies, are less rational or less organized or exhibit different forms of political behavior. For example, Menchik (Reference Menchik2017, 573) argues that “enduring academic Eurocentrism” is one of the reasons why social scientists erroneously believed that religion would decline in salience as economic development increased. Instead of a uniform conception of religion, he argues for a constructivist approach to studying religion from socially and historically specific perspectives. This conceptual pluralism, with its emphasis on “heterogeneity, context, and local conditions, means authors may eschew universal generalizations in favor of middle range theory” (578).
Similarly, Patricia Hill Collins (Reference Collins2000) has argued that the standard conception of political economy relies on an explanation of labor composition that overlooks how Black women specifically are forced to work certain low-wage jobs. Drawing on the experience of Black women, Collins calls for centering understudied perspectives that can uncover new data and knowledge while providing a distinctly different “angle of vision” for interpreting and understanding the “basic concepts used to describe that experience” (43). This approach has been used broadly by scholars using standpoint epistemology (Collins Reference Collins2009; Reference Collins2000; Harding Reference Harding2004).
Finally, one of our authors personally encountered how embedded assumptions can disincentivize certain kinds of research and comparisons. Meade’s dissertation research in the early 2010s comparing populism in the United States and Argentina was initially discouraged by well-meaning scholars because they believed that the two cases were categorically different or would produce an “apples and oranges” comparison. This advice stemmed from a taken-for-granted assumption that politics in the global north was rationally motivated (Downs Reference Downs1957; Freeze and Kitschelt Reference Freeze and Kitschelt2010), whereas in the global south, it was more personalistic and emotionally motivated (Mouzelis Reference Mouzelis1985; Weyland Reference Weyland2020). Thus, according to this logic, populist support in Argentina was an intrinsic feature of the political system, whereas in the United States, by contrast, populist support is a more deviant, dangerous rebuke of the political system.
Ultimately, Meade (Reference Meade, Ron and Nadesan2020) pursued the comparison on the grounds that drawing these cases together could unsettle unproven assumptions and improve the explanatory power of populism. However, her experience reflects some of the obstacles that students and emerging scholars experience in trying to challenge key concepts and deeply ingrained assumptions in the literature. Several of the examples of theory reconstruction we present next involve researchers grappling with similar legacies. We believe theory reconstruction can encourage the development and revision of theory in the field to be more empirically useful while also providing an avenue for marginalized perspectives to revise and shape theories.
Building on Similar Approaches
Our argument builds on both long-standing and recent approaches to rethinking concepts. We see theory reconstruction as compatible with various works that have called for political scientists to improve theory and interrupt bias by using descriptive or interpretive approaches to case selection and fieldwork. Like ours, these interventions set themselves in contrast to the standard methodological advice.
For example, Schaffer (Reference Schaffer2015) calls for political scientists to explore the everyday “experience-near” meanings of concepts in a process he calls “elucidation.” Schaffer contrasts his approach to Sartori’s “positivist reconstruction” (10), which he describes as an attempt to reshape everyday concepts into disciplined, bounded categories. He suggests three techniques for elucidation: grounding concepts by examining how everyday actors understand them, locating concepts in particular times and places, and exposing how concepts are situated in broader social structures. Our general approach to concepts aligns with Schaffer’s. However, where elucidation lends itself to an inductive, grounded approach to research, theory reconstruction is more abductive, grappling with the discrepancy between the literature’s notion of existing key concepts and the way those concepts operate in a specific context.
We also build on recent calls to action such as Holmes et al.’s (Reference Holmes, Guliford, Mendoza-Davé and Jurkovich2024) argument about how description can enrich theory and explanations in political science, as well as Michener, SoRelle, and Thurston’s (Reference Michener, SoRelle and Thurston2022) agenda of “bottom-up” research that can improve theory using missing perspectives. Lastly, although we mostly discuss qualitative work in this article, we do view quantitative work as compatible with theory reconstruction. A recent example is Munger and Phillips’s (Reference Munger and Phillips2022) study of social media that uses descriptive data from YouTube’s API to reconstruct theories of radicalization. We return to these works in the Practical Applications section.
Finally, we build on the agenda set by Simmons and Rush Smith’s (Reference Simmons and Smith2021) edited volume on creative, or noncontrolled comparisons, which they argue can help political scientists return to asking big research questions. Their volume includes contributions from some of the same authors we highlight here, illustrating the overlap in our approaches. Like Simmons and Rush Smith, we aim to provide guidance that would give early-stage scholars the language and tools to justify and carry out this type of work. Our contribution here is also distinct however, in that we emphasize a specific, streamlined approach that focuses on concepts.
Varieties of Theory Reconstruction
There are various potential contradictions between concepts in the literature and the empirics that researchers may encounter. These include issues with the mechanisms of a concept, a concept containing too many contradictory features, or observing a setting where a key concept is rarely applied in the literature. Given these possibilities, we outline four distinct strategies that have been used in existing scholarship, each of which we argue falls under the umbrella of theory reconstruction: revising, narrowing, extending, and disrupting. The first, revising, is the broadest type of approach and most closely resembles Scott (Reference Scott1985). The other three groupings (extending, narrowing, and disrupting) represent more niche approaches that are also common. We summarize these strategies in table 1.
Table 1 Variations of Theory Reconstruction
Although theory reconstruction focuses on challenging existing concepts, one could create a new concept as part of this process. For instance, in both the cases of “narrowing” that we cite, the authors create a seemingly new concept as part of breaking up an existing one that was overly broad or contained too many contradictions. Regardless of whether one creates a new concept, the critical distinction is that reconstruction changes the way we understand or view an existing concept from a different perspective.
Although these strategies of theory reconstruction are inspired by existing social science scholarship, they are not arbitrary. In their work on concepts in comparative research, Collier and Levitsky (Reference Collier and Levitsky1997) argue there are three levels of strategies for conceptual innovation, which they capture using Sartori’s metaphor of a ladder. They begin in the middle of the ladder, at the level of the “root concept” —the core concept being studied; in this case, let’s use the example of “democracy.” Here, innovation involves refining the definition of the concept and, by extension, its application to different cases (what the authors refer to as “precising”). Alternatively, innovation may occur further “down” the ladder at the level of subtypes. Conceptual innovation at this lower level is about differentiating between the diverse forms of the root concept through the development of subtypes (using the same example, liberal democracy versus electoral democracy, etc.). Lastly, innovation may instead occur “up” the ladder at the level of “overarching” concepts. Here, the question is about the relationship between the root concept and a supercategory or typology that it falls under (e.g., for democracy, political regime).
As we have made clear, our own argument for reconstruction differs in terms of the types of innovations, as well as the commitment to uniform conceptions advocated by Sartori. Nonetheless, this ladder structure does offer a useful framework for distinguishing between the four strategies we describe in detail next. For our purposes, the ladder provides a basic guide for thinking about the relationship between an existing concept and what exactly is being reconstructed; for example, subcategories, mechanisms, or challenging a common dichotomy.
We argue that each of the four strategies we describe corresponds with one of these levels. Specifically, “revising” and “extending” correspond with the idea of clarifying the mechanisms and application of the root concept, whereas “narrowing” is about rethinking subtypes of the concept. At the highest level of the ladder, the mainstream approach suggests that scholars can shift to a different overarching concept as a strategy to avoid concerns about conceptual stretching or to increase differentiation (Collier and Levitsky Reference Collier and Levitsky1997). We see this strategy as useful but as potentially sidestepping important concerns about the relationships between existing concepts and overarching typologies. Therefore, we propose “disrupting” as an alternative way to challenge the existing overarching spectrum associated with the root concept.
Although we do not claim that these strategies of theory reconstruction are exhaustive, we believe that any additional strategies should fit within one of these three levels. To be clear, we do not view a concept’s place on the ladder as predefined (democracy, for instance, does not have to be a root concept). Instead, it is based on the research argument and the strategy of reconstruction it is using. Figure 1 shows what this ladder framework looks like and how it corresponds with the different strategies.
Figure 1 Organizing the Strategies of Theory Reconstruction
In the following sections we have included at least one detailed example of existing work (largely in political science or political theory, although, in some cases, sociology as well) for each of the four strategies. Of course, authors can and do make multiple theoretical arguments in the same work. This can make it challenging to associate one piece of scholarship with a particular argument about a single concept. Nonetheless, we limit our summaries to what we see as the main puzzle the author is raising with the empirical evidence.
Revising
Revising is a strategy of theory reconstruction in which one adjusts or alters the inner workings of a concept to account for empirical observations. Many of the works we include here highlight the way a concept operates in practice to clarify previously obscured features or mechanisms. In these cases, the authors are asking specifically how a concept works, as opposed to what the literature assumes. We describe these approaches as revising, in that they are focused on applying an existing concept and clarifying how it works in that context. This strategy may involve describing scope conditions, expanding on some of the mechanisms, or introducing a new concept to clarify how the process actually works based on close observation, such as Scott’s “weapons of the weak” that he uses to describe a form of class relations where the subaltern engages in practices to resist or subvert the dominant class.
In addition to Scott, another well-known work of political ethnography that follows the revising approach is Wedeen (Reference Wedeen1999). Wedeen uses the case of Hafiz al-Asad’s nearly 30-year-old autocratic rule in Syria to challenge Weber’s theory of legitimacy and charismatic authority. On the face of it, political scientists at the time assumed that Asad’s legitimacy derived from popular belief in his cult of personality. Wedeen observed the settings where one would presumably find this charismatic form of legitimacy, yet instead found that, in private, Asad’s cult of personality was regularly subverted, treated with ambivalence, or satirized. This led her to ask why the regime devoted resources to spectacles and symbols of Asad that seemingly failed to work. She argues that, in addition to framing a coherent image of the state’s ideology, the rituals and spectacles are themselves forms of discipline. Wedeen reenvisions charismatic authority as a form of legitimacy that can exist without popular belief. She characterizes participation in state-sponsored rituals, despite a lack of belief, as an instantiation of the regime’s power.
Other theory-based works that follow a similar, revising approach include McDonnell’s (Reference McDonnell2017) study of bureaucracy, which explores the contrast between Weberian conceptions of successful bureaucratic systems and the reality of state bureaucratic offices in Ghana. Instead of a uniform institutional culture within a bureaucratic system, McDonnell finds high-performing “pockets of efficiency” coexisting alongside inefficient departments that operate according to patronage, even within the same ministry. This “patchworked” model challenges the idea that state bureaucracies of the global south are largely non-Weberian and dysfunctional, while also providing a revised conception of how bureaucracies can work in institutionally diverse, developing states.
Although revising is about bringing a concept more in line with observed phenomena, one might be interested in doing the opposite by bringing “new” phenomena into conversation with an existing concept. A good example of this is what we call “extending.”
Extending
Extending is an instance of theory reconstruction where one expands a concept into a new or unfamiliar context. In this case, the researcher finds that an understudied setting or phenomenon is best described using a key concept that is not typically associated with it in the literature. To resolve this, they “extend” the existing concept into the new context, using it to help capture their observations and conversely using the case to expand the literature’s understanding of the concept. One example mentioned earlier that fits this description is that of Michener, SoRelle, and Thurston’s work (Reference Michener, SoRelle and Thurston2022) on the concept of the welfare state. Their research, as well as the other examples we list next, extends key concepts to populations that are subaltern, underrepresented, or were otherwise ignored by the mainstream of the discipline. This demonstrates another way that theory reconstruction can address power inequities in our field by challenging whose voices are incorporated into our studies.
Another example of extending is Soss’s (Reference Soss2018) study of political participation in the United States. Contrary to Sartori, Soss advocates for intentionally “stretching” concepts and applying them to new or formerly neglected populations and contexts, in a process he refers to as “casing.” He suggests this allows researchers to critique, contest, or reformulate theories and to view currently accepted concepts and casings as “a product of earlier intellectual and political activity” (Soss Reference Soss, Simmons and Smith2021, 90). Soss describes how, based on his observations of welfare offices, the behaviors of welfare recipients seemed to match criteria commonly attributed to political participation—despite the fact that the concept’s use in the literature tended to exclude such behaviors and settings. The author suggests that this exclusion can be explained by an unstated normative assumption shared by many academics: a liberal-democratic ideal that suggests that good citizens engage by contacting representatives or voting, not by applying for social services. Motivated by his frustration that the discipline seemed to engage with poor people’s politics in terms of what they lacked, Soss makes the argument for “extending” political participation into welfare-claims making.
Another recent work using a similar strategy of extending is Ngwane (Reference Ngwane2021), which focuses on democracy “at the margins” in South Africa. In this instance, Ngwane expands the literature’s understanding of where democratic processes occur by looking at communal organizations in South African townships and informal settlements. Just as Soss critiques the exclusion of the poor from conceptions of political participation in the United States, Ngwane critiques the assumption that slum dwellers living in South Africa’s urban periphery lack political agency or meaningful participation in the democratic system. Proponents of this assumption either discount the participation of slum dwellers in working-class struggles for democracy or underestimate the extent of political organization and grassroots decision-making processes within these communities. Through participant observation and archival research, Ngwane shows that slum communities are often well organized and engage in grassroots democratic processes of decision making. His study reconstructs democracy as a communal process practiced in South Africa’s urban slums and informal settlements. Ngwane’s work expands our conception of where grassroots democracy operates by shifting our understanding of the concept to a setting the literature often excludes.
Narrowing
Narrowing is either the division of an existing concept into “subtypes” or a division between an existing key concept and a separate (often new) concept. Narrowing begins with the author noticing that a concept that is treated monolithically in the literature exhibits significant variation or contradictory aspects in practice. In other words, scholars who use narrowing are identifying a problem similar to conceptual stretching. In this case, however, the proposed solution is different from Sartori’s. Narrowing reconstructs the root concept by breaking it down into differentiated categories or by clearly distinguishing it from a new concept. In this section we include examples of both strategies.
Soss and Weaver’s (Reference Soss and Weaver2017) intervention into the concept of the state is a type of narrowing. Drawing on the case of police repression of poor black citizens in Ferguson, Missouri, they argue that electoral and liberal democratic models of political science are ill equipped to account for the experiences of state practices in “race-class subjugated” (RCS) communities. They specifically target conceptualizations of the state that depict it as a resource that citizens gain power from through proximity. In this view, the “closer” one is to the state, or the stronger the state–society relations, the more power one has. However, this assumption fails to account for the fact that, in race-class subjugated communities, proximity to certain repressive forms of the state, such as the police or the prison system, is the norm. In this case, instead of empowering citizens, proximity to the state results in surveillance, repression, and predation.
The authors use this insight to argue that the state, as a concept, can be contradictory in terms of how it is viewed by citizens in race-class subjugated communities. They propose a division of the state into two “faces.” Although the first, “liberal democratic” face deals with issues of electoral representation, they argue that more focus should be placed on the predatory, “second face” of the state and its manifestation in poor and working-class communities of color. They claim that this hidden dimension has implications for the broader concept of the state and that scholars “must work to build a less distorted account of American politics that reflects—as more than an unfortunate anomaly—the political lives of RCS communities” (Soss and Weaver Reference Soss and Weaver2017, 584).
Another well-known text that follows a similar theoretical approach is Chatterjee’s book, The Politics of the Governed (Reference Chatterjee2004), which narrows the concept of civil society. Chatterjee argues that the idea of civil society as consisting of individuals with rights, accountable representation, and liberties is distinctly a product of the “modern West” (3). This conception of civil society is at odds with some of the features of state–society relations, clientelism, and practices of population management inherited from colonialism in democracies such as India. Instead of a single concept, Chatterjee argues that, in “most of the world,” the domain of politics between the household and the state can be broken up into two categories: civil society (for the middle class and elites) and what he calls “political society” for the masses. Although in a way that differs slightly from the subdividing approach of Soss and Weaver, Chatterjee too identifies a discrepancy between the theory and empirics of a concept, choosing to narrow the overly broad root concept by giving it a more specific meaning and using what is left to establish a new concept.
In both cases, these works enhance the applicability of concepts in the discipline by highlighting how “the state” or “civil society” is experienced differently by citizens of the global south, as well as by “race-class subjugated” communities in the United States. Notably, all three authors take a more theoretical approach than some of the other cases we discuss. Soss and Weaver present their narrowed concepts of the state through a literature review, supplemented by observations from specific cases like Ferguson, Missouri. Meanwhile, Chatterjee’s conceptualization combines political theory with an analysis of Indian politics.
Although narrowing is generally about creating a new division or distinction between concepts, what happens when researchers find that an existing division or range of concepts is inadequate for understanding their observations? In this instance, the researcher may seek to unsettle the division of an “overarching” concept, which we call “disrupting.”
Disrupting
In the case of disrupting, the concept being reconstructed typically consists of a spectrum or dichotomy. Disrupting starts when the researcher finds that a commonly used distinction between concepts or categories in the literature fails to hold up in practice. This could be because a dichotomy is too limiting, or there may be significant overlap between categories (e.g., political regimes or varieties of political economy, global north/south, etc.). The researcher would then disrupt the spectrum by challenging its limits; for example, by adding a new category to expand its boundaries. Or conversely, scholars could question the usefulness of the conceptual distinction by highlighting instances in which different categories overlap considerably. We envision disrupting as distinct from narrowing, in that narrowing only implies specifying the meaning of a concept, whereas disrupting is about either challenging or obscuring the limits of a range or spectrum of concepts.
One example we associate with disrupting is Linz’s (Reference Linz, Allardt and Yrjö1964) famous argument about political regimes. Linz asserts that political scientists in the postwar period relied on a rigid dichotomy of regime types: democratic versus totalitarian regimes. Democracies were understood to be regimes that provided regular constitutional opportunities for peaceful competition of power and upholding basic liberal values. Totalitarian regimes, in contrast, were marked by total domination not limited by the law, a mass party, and official ideology. Linz (336) argued that this dichotomy of political regimes was a poor heuristic that either obscured or ignored political systems that could be described as “hybrids,” including Spain under Franco and Turkey under Atatürk. These “hybrids’ exhibited some qualities that, at face value, might seem similar to democracies, such as the presence of some civil rights or an autonomous judiciary. Yet they also had qualities similar to those of totalitarian regimes, such as the ill-defined power of a single leader or junta. Linz chose to disrupt this common dichotomy of political regimes, arguing that instead of being democratic or totalitarian, these hybrid regimes were best classified as “authoritarian.”
Moreover, it is important to note that Linz based part of his analysis not on “top-down” institutional features of the state but on the perspectives of citizens living in these regimes, including his own; he grew up in Spain under Franco. Years later in an interview, Linz (Reference Linz, Munck and Snyder2007, 163) agreed that his distinction between autocratic and totalitarian regimes rested on different “ways of thinking” (ideologies vs. mentalities) that were difficult to operationalize. To substantiate this distinction, he highlights the fact that newlywed couples in Nazi Germany would receive copies of Mein Kampf from the city council or that, under Stalin, schoolchildren in the Soviet Union were exposed to the text History of the Communist Party. Yet conversely, citizens in authoritarian Spain would find it difficult to obtain a copy of Franco’s speeches because few people read his writings (163). Although this example is effective in making his point about ideology, it also suggests that, to an extent, Linz’s work disrupting the spectrum of political regimes is based on insights about the state from the standpoint of average citizens.
Hanchard (Reference Hanchard2018) engages in a similar reconstruction of political regimes. The author argues that, despite the distinction in comparative politics between democratic versus authoritarian or totalitarian regimes, this assumption overlooks “democracy’s reliance on regimes of exclusion” (71). Hanchard criticizes the field of comparative politics for failing to recognize the coexistence of racial and social inequality that have long been part of democracies, as far back as classical Athens where democracy existed alongside slavery and antidemocratic institutions. Even today, democracy’s coexistence with highly unequal racial regimes reflects what Hanchard describes as a form of “polyarchy” in which diverse political institutions and forms of governance—some democratic and others antidemocratic—can exist within the same regime. By pointing out that these features of a “dualist regime” (84) coexist within democracies, Hanchard disrupts the dichotomy between democratic and nondemocratic regimes, while challenging comparative politics to rethink the role of race and its historical relationship to democratic institutions.
Disruption can also be used to challenge the dichotomies typically prevalent in case selection by comparative scholars. Key concepts like regime type, state capacity, or types of political economy, for instance, might determine which cases scholars decide to compare and which ones instead are seen as “apples and oranges.” Thus, disruption can refer to scholars using an unorthodox comparison between cases to challenge an established dichotomy or common categorization in the literature. An example would be Meade’s (Reference Meade and Walton2024) aforementioned research that compared politics in the United States to that in Argentina, challenging the “emotional” global south versus the “rational” global north division in the populism literature. Her findings disrupt this division by highlighting common political behavior across populist supporters in both countries.
Although differing in approach, in each of the four groupings (revising, extending, narrowing, and disrupting) the researcher identifies a discrepancy between a key concept in the literature and in practice and then adjusts or reevaluates the concept based on their observations. In each case, hidden dimensions of concepts are revealed, resulting in revised or new concepts that enrich social science theory.
Practical Applications
As highlighted in some of these examples, theory reconstruction does not necessarily need to be preplanned but can also emerge as a viable option during fieldwork. However, for those scholars seeking to adopt a theory reconstruction approach, we draw on the works described in this article to identify a set of methodological tools and practical steps that they can take.
An important first step is to pay attention to the historical context in which concepts have been developed and used. As Munck and Snyder (Reference Munck and Snyder2007) show in their collection of interviews with comparative political scientists, scholars often draw from their everyday experiences, such as Linz’s experience growing up in Spain under Franco. However, they need to be aware of the embedded assumptions they bring into the field. These techniques have long been practiced by scholars in disciplines such as critical Indigenous studies, feminist studies, and critical race theory (Bruyneel Reference Bruyneel, Lucero, Turner and VanCott2013; Collins Reference Collins2009, Reference Collins2000; Grande Reference Grande2015; Harding Reference Harding2004). For instance, building on the works of Black feminist scholarship, Thomas (Reference Thomas2022) points out that scholars are themselves instruments in the research process, bringing into the field their subjectivities and the various ways they have been “trained to see” the world.
Furthermore, engagement with these assumptions can allow for greater reflexivity, as well as a rethinking of key concepts in the discipline. Indigenous scholars Mowatt, Wildcat, and Starblanket (Reference Mowatt, Wildcat and Starblanket2024, 306) argue in the recent Annual Review of Political Science, “Critical inquiry into the interaction of Indigenous and settler sovereignties prompts political science to revisit its understanding of political space and power.” Soss’s (Reference Soss2018) study, which “extends” the concept of participation to include the practices of welfare recipients, provides an example. He notes that the liberal-democratic norms shared by academics, including himself, were shaping contemporary understandings of American democracy while marginalizing perspectives from the poor and reproducing the “powerful commitments scholars often have toward conventional categories in their field” (100).
Second, to build abductive research puzzles that revise or create new concepts, scholars should engage in the important and underacknowledged work of description and data collection in political science (Holmes et al. Reference Holmes, Guliford, Mendoza-Davé and Jurkovich2024). Many of the scholars we discuss arrived at their core insights through ethnographic observations of social behavior, cultural media, and other sources gleaned from conducting substantial fieldwork. Although there is a particular affinity between the theory reconstruction approach and ethnographic methods, quantitative descriptive tools can also be used to revise theory. For example, Holmes et al. (Reference Holmes, Guliford, Mendoza-Davé and Jurkovich2024) note that description is a useful tool for identifying which cases and understandings have been included or excluded in large datasets. Similarly, the recently founded Journal of Quantitative Description: Digital Media publishes descriptive papers that use methods like machine learning, text as data, and network analysis to build theoretical knowledge in the emerging arena of digital media (Munger, Guess, and Hargittai Reference Munger, Guess and Hargittai2021).
Scholars seeking to use data and description for theory reconstruction might consider drawing on neglected perspectives. Following Wittgenstein’s family resemblances approach, these new sources of data can improve the descriptive power of concepts by offering understudied perspectives, or what Ragin and Amoroso (Reference Ragin and Lisa Amoroso2018) describe as “giving voice.” Here we also build on Michener, SoRelle, and Thurston’s (Reference Michener, SoRelle and Thurston2022) call for political scientists to study neglected subjects and communities using grounded ethnographic methods. For example, Ngwane “extends” democracy by examining informal settlements in South Africa, whereas Soss and Weaver “narrow” the state by bringing in the experiences of race-class subjugated communities.
Finally, when presenting their findings, scholars using theory reconstruction should be transparent about the nonstandard aspects of their research process. For example, they should explain how observations from fieldwork, interviews, or real-world experiences led them to initially build research puzzles based on challenges to an existing theory or concept or to revise their methods or frameworks during the research process based on unexpected findings (Kapiszewski, MacLean, and Read Reference Kapiszewski, MacLean and Read2022). Lastly, they can contribute to conceptual development by discussing why existing concepts failed to explain their cases. For example, Michener, SoRelle, and Thurston (Reference Michener, SoRelle and Thurston2022) argue that the US welfare system has missed key research sites and policy tools because elite policy makers and politicians had defined the categories and understandings used by academics.
Conclusion
The examples in this article show how theory reconstruction can contribute to the development of concepts in social science. In each of the works we discuss, an author starts with a discrepancy between their observations and a key concept in the literature. Then, using one of the different strategies described earlier—revising, extending, narrowing, or disrupting—they resolve this tension by reconstructing the concept, in the process revealing hidden dimensions, subtypes, or new relationships. By establishing theory reconstruction as an accessible approach to research, we seek to encourage more theory-based research as one means of moving the discipline forward. Just as our methods of data collection, our sources of data, the kinds of questions we ask, and so on, are all subject to changes and updates over time, so too must our concepts, the building blocks of theory, be subject to reconsideration in the face of new observations.
More broadly, much of our discipline’s ideas and scholarship (even on global topics) remains largely insulated from contributions outside Western academies. This insulation of the American political science canon discourages valuable contributions from scholars and students who may not rely on the same foundational thinkers, theories, and methodologies. Theory reconstruction on its own is not a solution to these deeply entrenched problems. However, in comparison to the status quo of different regional varieties of political science, each siloed off from each other, it does provide a viable alternative. It is a tool that can empower scholars to challenge the defining concepts of their various disciplines and embrace cases, theories, and comparisons where their empirical observations diverge from major assumptions.
Theory reconstruction can also create new avenues of research. Scott’s identification of “hidden acts of resistance” has been used to rethink power from the subaltern perspective. Linz’s identification of authoritarianism radically reshaped the discipline’s understanding of regime type. The recent increase in popularly elected, illiberal regimes has led to a resurgence of hybrid-type conceptions of democracy that still build on Linz’s work, such as “competitive authoritarianism” (Levitsky and Way Reference Levitsky and Way2010) and “autocratic legalism” (Scheppele Reference Scheppele2018). Despite the impact of these and other works of theory reconstruction in political science, this mode of research has not been encouraged in the discipline.
Drawing from the case studies in this article, we believe that theory reconstruction is likely to be especially useful for the study of new and emerging phenomena, for research involving marginalized or understudied groups or contexts, and for work that involves key concepts with long-standing, taken-for-granted assumptions. Societies across the globe are currently confronted with unprecedented crises and challenges, including the impact of social media and AI on the information environment, the rise of global populism, the erosion of democratic institutions, and climate change. In such a landscape, revising outdated paradigms (Geddes Reference Geddes2003) and concepts is an urgent task for our field that will allow us to offer insights on the challenges that will shape our politics for years to come. Theory reconstruction has long been used to build theoretical knowledge and revise outdated concepts. We hope to acknowledge and encourage more of this important work by providing a clearer methodological approach for future research.
Acknowledgments
This article received support and considerable feedback from the Emerging Methodologists Workshop sponsored by the Qualitative and Multi-Method Research (QMMR) section of the American Political Science Association. Our great thanks to Ryan Saylor, Cathie Jo Martin, Jason Seawright, Rich Snyder, Spencer Piston, and our thoughtful reviewers at Perspectives on Politics, all of whom helped make this article much stronger. Lastly, we dedicate this article to the inspirational legacy of Michael Burawoy.
Open access
