This chapter explores the application of large language models (LLMs) in empirical legal studies, with a focus on their potential to advance research on EU law at scale. The chapter provides a non-technical introduction to LLMs and the role they can play in legal information retrieval, including the classification of case characteristics and outcomes, which constitutes one of the most common research tasks in legal scholarship. The chapter stresses the importance of validation – researchers cannot treat the output of LLMs as automatically correct and instead must demonstrate the relevance and reliability of measures and results obtained through the use of LLMs in the context of their research topic. While LLMs are capable of significantly reducing the cost of doing legal research, their use will place growing demands on scholars to ensure the integrity of their findings. The chapter also reflects on the distinction between closed- and open-source models and how ethical and replicability imperatives might influence model choices in an increasingly crowded field.

Are people already at increased risk for disease more likely to be exposed to the risk factor of interest? Does closer observation of people with a disease lead to a false association? In retrospective studies, do people with a disease recall prior exposures more (or less) that healthier people? Are research interviewers a source of biased data collection? Confounding is an existential threat in biomedical research; here a second factor, which is associated with both the disease and the risk factor being studied, is an actual cause of the disease. If studies cannot fully control for the effect of the second risk factor, residual confounding will bias the risk estimate. Who participates and doesn’t participate in research is another source of bias. How diseases and risk factors are classified and categorized may introduce bias, and changing defined categories is yet another source of bias.

Chapter 2 tells the story of how ethnicity came to be known in Kenya through territory, providing an overview of the history of ethnic territorial boundary drawing from its inception with the first colonial administration, to today. The principal motivation for the earliest hard boundaries between purportedly homogenous ethnic groups was to free up land for white settlement and capital accumulation. After independence, the administrative boundaries of provinces and districts were deliberately retained, and ethnic patterns of land settlement were engineered. With multi-party elections in the 1990s, these established ‘ethnic territories’ motivated electoral gerrymandering, the most significant postcolonial driver of ethnic territorialisation. All these practices cemented a profound connection between land, boundaries, identity, rights, power, and security. I show how the 2010 constitution worked within this paradigm, too, but in novel ways that moved toward vagueness to manage the inflammatory, grievance-based politics tethered to boundary drawing in Kenya. In doing so, I show how ethnic territorial population concentration today is less certain than commonly imagined.

This chapter presents a set of practical, classroom-tested exercises for teaching concept analysis, emphasizing how deliberate engagement with concepts improves research and communication. It outlines several strategies, including reconceptualizing familiar terms by identifying defining and elective attributes, and situating them within semantic fields. It highlights the heuristic power of Collier’s question, “What is that a case of?”, which prompts students to move from empirical examples to abstract categories. Taxonomy construction is another key tool, helping students systematize ideas across domains – from constitutions to cuisine – and understand how classification affects knowledge. Binary sorting (“There are two kinds of people…”) and genre-mapping (“What do you work on?”) also serve to stimulate reflection on research categories. The chapter argues for the pedagogical value of testing, suggesting that students benefit from identifying, defining, and illustrating core concepts as a way to internalize intellectual terrain. Field exams, concept glossaries, and vocabulary tests help solidify these connections. The chapter concludes with a case for “conceptualism” as a core scholarly orientation: Concepts allow generalization while grounding knowledge in empirical cases. Working with concepts is cognitively satisfying and essential for memory, communication, and cumulative learning – what more could a good course (or concept) hope to achieve?

This concluding chapter offers some final reflections on the nature of knowledge about ethnicity in Kenya. I argue that if the nature of this knowledge is purposefully vague and makes ethnic categories polyvalent, then the best way to protect against problematic uses of ethnic knowledge is vigilance. This is far less satisfying and reassuring than law or rights as a framework for governing the risks of diversity, but it is far more appropriate, and I briefly consider what this might look like. Finally, I look forward to the digitisation of Kenya’s population register and aspirations to establish a population knowledge architecture so sophisticated that it could render numerous registers interoperable and ultimately replace even the census. I reflect on the nature of ethnic classification in such an architecture and argue that it would lose all the qualities that have made it amenable to solidaristic and pluralistic purposes thus far, while amplifying all its dangers.

This chapter concludes the volume by reflecting on the ongoing value of concept analysis in the social sciences. It revisits the tension between hyperfactualism – obsessive attention to granular detail – and the necessary abstraction that enables generalization. Conceptualization, the authors argue, helps scholars not only communicate more clearly but also observe and describe phenomena more effectively. Far from being a distraction, conceptual work sharpens empirical inquiry. The chapter highlights the interplay between conceptualization and measurement, especially in validity assessment, and underscores how concepts represent and structure knowledge. Attention to concepts also facilitates integration and translation across time, space, and disciplines, as seen in such examples as the V-Dem project. Issues of conceptual boundedness, typologies, and traveling are revisited, drawing on contributions from cognitive linguistics and classic debates between lumpers and splitters. The authors also reflect on how digital tools and formal modeling offer new avenues for concept innovation. Finally, they affirm the importance of teaching concept analysis as a way to clarify students’ thinking, research design, and disciplinary communication. In sum, the chapter defends the overconscious scholar: one who sees in concepts not distraction, but a path toward cumulative, communicable, and intellectually satisfying scholarship.

This chapter introduces the supposed problem of ethnicity: that it undermines national cohesion, or is a colonial hangover with no appropriate place in political life. In contrast, I argue that ethnicity is neither inherently desirable nor undesirable; its political effects depend on how it is known and used, and our understanding of how it is known remains underdeveloped. I establish that there is no definitive list of Kenya’s ethnic groups, and we must stop taking for granted what we think we know about ethnicity. I offer the concept of cultivated vagueness – a widespread aversion to resolving the ambiguity of lists of Kenya’s ethnic groups – to understand how ethnic knowledge works and to contrast it with legibility and governmentality. Cultivated vagueness is the response from bureaucrats, civil society, citizens and the state to the conundrum that ethnic knowledge is both common sense and impossible to settle. It also explains how ethnic classifications serve both projects of division and of pluralism. I suggest that attention to the benefits of cultivated vagueness may facilitate the advancement of the latter over the former. The chapter outlines the book’s methodology and chapters.

Meteorites are classified using a hierarchical scheme based on the degree of relatedness of samples. Chondrite groups are typically from a single parent body; clans and classes are clusters of related groups that accreted in similar regions of the solar nebula. Classification of a new meteorite requires visual observation of macroscopic characteristics, microscopic examination of textures, and analyses of minerals. Isotopic or bulk compositional data may also be acquired.

A comprehensive classification system for the ultrasound diagnosis of early pregnancy is essential. Such a classification system must provide clear decriptions and diagnostic criteria for all possible pregnancy locations. The use of uniform terminology will also aid to reduce the risk of misdiagnosis and inappropriate treatment.

Reparations are a key mechanism for delivering justice to victims and survivors of armed conflicts. The first generation of victim engagement was marked by demands for reparations from state authorities, making them a core element of post-war justice. This chapter examines how the nature of a past conflict shapes the conditions for victim engagement in reparations. It is shown that social classifications of victim groups that arose during or prior to conflict act as a moderating factor, influencing who is deemed eligible for compensation. However, these classifications are not fixed; victims and survivors can actively reshape them through transitional justice processes. This chapter examines how social classifications shape reparation policies by analysing three case studies – Guatemala, Timor-Leste, and Northern Ireland – each representing a distinct type of conflict. It explores the opportunities and constraints victims face in articulating and securing compensation claims, highlighting how these are influenced by evolving social classifications.

In this chapter, I argue that the first book of the Parts of Animals (PA) expresses a form of realism about animal species. While the claim that Aristotle was a realist about species may seem obvious to those coming to the PA from the Metaphysics, the current view among specialists is that Aristotle’s zoology was not working with a concept of species. Some have even gone so far as to avoid translating eidos as “species” throughout his zoological writings. In contrast to this, I argue: first, that indivisible species constitute the ousiai of Aristotle’s zoology; and, second, that the aim of Aristotelian zoological division is to identify and organize the features specified in the definition of those species. The latter (epistemological) claim is explicit in the discussion of division in PA I 2–3, while the former (ontological) claim is advanced in PA I 4.