A Typology for the Case Study in Social Science Following a Review of Definition, Discourse, and Structure

The author proposes a typology for the case study following a definition wherein various layers of classificatory principle are disaggregated. First, a clear distinction is drawn between two parts: (1) the subject of the study, which is the case itself, and (2) the object, which is the analytical frame or theory through which the subject is viewed and which the subject explicates. Beyond this distinction the case study is presented as classifiable by its purposes and the approaches adopted— principally with a distinction drawn between theory-centered and illustrative study. Beyond this, there are distinctions to be drawn among various operational structures that concern comparative versus noncomparative versions of the form and the ways that the study may employ time. The typology reveals that there are numerous valid permutations of these dimensions and many trajectories, therefore, open to the case inquirer.


The Need for a Typology
Case study research is one of the principal means by which inquiry is conducted in the social sciences. Reviewing work in economics and political science (Acemoglu, Johnson, & Robinson, 2003;Bates, Greif, Levi, Rosenthal, & Weingast, 1998;Rodrik, 2003), Gerring (2004, p. 341) concludes that the use of the case study is "solidly ensconced and, perhaps, even thriving." Bennett, Barth, and Rutherford (2003) showed that, in 14 journals focusing on 2 areas of research in social science, the proportion of articles in which a case study was employed remained broadly stable at about 20% over the period between 1975 and 2000.
Despite the popularity of the case study design frame, there is little in the way of organizational structure to guide the intending case inquirer. Gerring (2004, p. 341) continues his review with this comment: "Practitioners continue to ply their trade but have difficulty articulating what it is that they are doing, methodologically speaking. The case study survives in a curious methodological limbo." de Vaus (2001, p. 219) agrees, in discussing the way that the case study is explained: "Most research methods texts either ignore case studies or confuse them with other types of social research." If "methodological limbo" exists it is not for lack of methodological discussion. Indeed, this has been extensive over the past 40 years across the social sciences-see, for example, Simons (2009), Yin (2009), Mitchell (2006, Flyvbjerg (2006), George and Bennett (2005), Stake (2005), Hammersley and Gomm (2000), Bassey (1999), Ragin and Becker (1992), Merriam (1988), Eckstein (1975), and Lijphart (1971). The problem is perhaps that methodological discussion of case study has tended to focus on its epistemological status, its generalizing "power," or on various aspects of study construction. Less conspicuous, though, has been any synthesis of the discussion that might offer classificatory schemata for intending researchers: There have been only limited attempts to offer intending inquirers a Gestalt, mapping out the terrain and potential routes to travel. By way of response to this state of affairs, I overview some of the ways in which case study is discussed and defined in order to propose a framing structure and typology for case study. In doing this I attempt to disentangle the threads and layers of classificatory principle that have become interwoven in dialogue about the place and use of the case study.

Definition
Differing themes and priorities characterize attempts at definition of the case study. This is to some extent explicable by the diversity of epistemological starting points from which practitioners and analysts of the case study arrive. While those from sociology, education, and psychology have tended to see the case study in an interpretivist frame, those from business, politics, and other areas may espouse the interpretivist holism of case study but address this through what George and Bennett (2005, p. 5) have called "neopositivist" means via the identification of variables to be studied-see, for example, the discussions of Luker (2008) and Yin (2009). By contrast, those in medicine and the law have tended to see the case study principally as a vehicle for exemplifying or illustrating novel or archetypal phenomena.
Notwithstanding these differences, strong commonalities exist across disciplinary margins. Reviewing a number of definitions of case study, Simons (2009) concludes that what unites them is a commitment to studying the complexity that is involved in real situations and to defining case study other than by the methods of data collection that it employs. On the basis of these commonalities she offers this definition: Case study is an in-depth exploration from multiple perspectives of the complexity and uniqueness of a particular project, policy, institution, program or system in a "real life" context. (p. 21) To Simons's definition, I should emphasize the point she makes in her preliminary discussion that case study should not be seen as a method in and of itself. Rather, it is a design frame that may incorporate a number of methods. Stake (2005, p. 443) puts it this way: Case study is not a methodological choice but a choice of what is to be studied. . . . By whatever methods we choose to study the case. We could study it analytically or holistically, entirely by repeated measures or hermeneutically, organically or culturally, and by mixed methods-but we concentrate, at least for the time being, on the case.
Choice of method, then, does not define case study: analytical eclecticism is the key. Ragin (1992, p. 5) centers his definition on the case study's contrast with what we might call "variable-led" research. Rather than looking at few variables in a large number of cases, the case inquirer looks at the complex interaction of many factors in few cases: The "extensiveness" of the former is discarded for the "intensiveness" the latter offers. There is a "trade-off," as Hammersley and Gomm (2000, p. 2) put it, between the strength of a rich, in-depth explanatory narrative emerging from a very restricted number of cases and the capacity for generalization that a larger sample of a wider population can offer. Ragin puts it thus: The . . . case-oriented approach places cases, not variables, center stage. But what is a case? Comparative social science has a ready-made, conventionalized answer to this question: Boundaries around places and time periods define cases (e.g., Italy after World War II). [p. 5] Ragin's definitional discussion serves two purposes: It contrasts the emphasis on cases with an emphasis on variables in other kinds of research. In discussing the boundary, it also stresses the particularity noted by Simons and Stake-wherein the parameters of particularity are set by spatial, temporal, personal, organizational, or other factors.
One important feature of cases only alluded to in the definitions discussed thus far is any emphasis on the significance of an analytical frame in the constitution of the study. For example, while the Korean War as a subject of study might satisfy conditions of singularity, boundedness, and complexity it would not be a case study-or at least not the kind of case study that would be of interest to social scientists-unless it could be said to be a case of something, and that "of" would constitute the study's analytical frame. The Korean War's status as a case of something has to be established. Is it a case of a war? If so, can it be said that it is a case of an especially remarkable or unusual kind of war? Perhaps, by contrast, it may be a case of a border dispute or of U.S. resistance to the perceived threat of communist expansion. George and Bennett (2005, p. 69) put it this way: "The investigator should clearly identify the universe-that is, the 'class' or 'subclass' of events-of which a single case or a group of cases to be studied are instances." The subject of the study is thus an instance of some phenomenon, and the latter-the phenomenoncomprises the analytical frame.
Although writing earlier than George and Bennett, Wieviorka (1992, p. 159) showed that the case is not simply an instance of a class. He unpacked in more detail the distinctions between the case and the class by noting that when we talk about a case we are in fact talking about two elements: First, there is what he calls a "practical, historical unity." We might call this the subject-in my example, above, it would be the Korean War. Second, there is what he calls the "theoretical, scientific basis" of the case (such as U.S. resistance to putative communist expansion, in the same example). This latter forms the analytical or theoretical frame, and we might call this the object of the study. The notion of a "theoretical, scientific basis" thus delimits the "class": The class has to offer more than a set of similar instances. So, the class cannot comprise merely similar phenomena (e.g., wars) but those phenomena as instances of an analytical focus (e.g., resistance to perceived communist expansion). Wieviorka continues: For a "case" to exist, we must be able to identify a characteristic unit. . . . This unit must be observed, but it has no meaning in itself. It is significant only if an observer . . . can refer it to an analytical category or theory. It does not suffice to observe a social phenomenon, historical event, or set of behaviors in order to declare them to be "cases." If you want to talk about a "case," you also need the means of interpreting it or placing it in a context. (Wieviorka, 1992, p. 160) It is important to stress the significance of the separateness of the subject and the object in case study since the distinction between the one and the other is characteristic of all social inquiry, yet relatively neglected in discussion of the case study. It is defined variously in different kinds of research. In his classic work on sociological theory, Wallace (1969, p. 3) pinpointed the significance of the distinction between (a) the thing to be explained and (b) the explanation in a piece of research by calling the thing to be explained the explanandum and the thing doing the explaining the explanans. Some time earlier, Hempel and Oppenheim (1948) had drawn attention to the need for such a differentiation to account for the ability of science to answer "why" rather than simply "what" questions. In social sciencewhere we also want to answer "why" rather than "what" questions-one of the more straightforward means of making this distinction is by differentiation between dependent and independent variables, yet this of course is not the only way of doing it, and case inquirers need to be aware of this.
Case inquirers need to be aware of it because Wallace went further, aligning the explanandum with the dependent variable, and the explanans with the independent variable, and Eckstein (1975) refers to the analytical frame in a case study as a single measure on a variable. But one surely needs to be guarded in the use of terms associated with variable-led research when thinking about idiographic research. The extension of explanandum and explanans to variables by Wallace (and many after him) is, perhaps, metaphorical. But if this is so-if it is indeed a metaphor-it is a dangerous one if extended to all kinds of research, including the idiographic. In fact, here it becomes more like catachresis than metaphor: Let us take the example again of the World War II (subject) as a case study of a "just war" (object). Here, the notion of justness is the explanandum (the thing to be explained), and the thing doing the explaining-the explanans-is World War II. This is a quite valid use of the explanandum/explanans distinction as promoted by Hempel and Oppenheim and Wallace. But the idea that the explanans can be seen as an independent variable so grossly violates our expectations of an independent variable (e.g., singular rather than complex; manipulable experimentally) that it ceases to be tenable.
Likewise, use of the similar term "unit" is confusing, being adopted by Wieviorka to refer to the case (the subject), while VanWynsberghe and Khan (2007, p. 87), for example, use it to mean the object. (They say, "The interplay between the unit of analysis and the case is a constitutive element of case study research" [italics added].) These confusions have arisen, I think, partly because of the neopositivist discourse surrounding so much methodological discussion concerning case study, of which the "variableled" discussion of explanans and explanandum is an example. As another example, look at Lijphart's (1971, p. 684) distinction between experimental, statistical, and comparative method in social science, in which he asserts that in areas such as political history comparative method has to be "resorted to" because of the small number of potential cases and the invalidity of "credible controls." I discuss Lijphart's analysis in more detail later.
The ostensible looseness of the case study as a form of inquiry and the conspicuous primacy given to the case (the subject) is perhaps a reason for inexperienced social inquirers, especially students, to neglect to establish any kind of object (literally and technically) for their inquiries. Identifying only a subject, they fail to seek to explain anything, providing instead, therefore, a simple description in place of a piece of research. For the study to constitute research, there has to be something to be explained (an object) and something potentially to offer explanation (the analysis of the circumstances of a subject).
In brief, as a conclusion to this discussion, I suggest that a case study must comprise two elements: 1. A "practical, historical unity," which I shall call the subject of the case study, and 2. An analytical or theoretical frame, which I shall call the object of the study.
Taking account of this, the definition of case study that I shall adopt for the typology I develop here is as follows: Case studies are analyses of persons, events, decisions, periods, projects, policies, institutions, or other systems that are studied holistically by one or more methods. The case that is the subject of the inquiry will be an instance of a class of phenomena that provides an analytical frame-an object-within which the study is conducted and which the case illuminates and explicates.
I give below an explication of some of the elements and dimensions just discussed, noting points about the kinds of selection and decision likely to be necessary during the case study. This is done to provide a rationale for the typology ultimately summarized in Figure 1.

The Subject and Object Subject
In making a central distinction between subject and object of study, the definition at which I have arrived leads us first into questions about how the subject is identified-whether that subject is a Glasgow gang (Patrick, 1973), the Head Start education program (Zigler & Muenchow, 1992), or an international coffee organization (Bates, 1998). The subject is in no sense a sample, representative of a wider population (and I discuss this further below). Rather, the subject will be selected because it is an interesting or unusual or revealing example through which the lineaments of the object can be refracted. In this, its scope is not restricted: As White (1992) points out, the subject may be as broad as Lenin's analysis of peasant social formations, or as narrow as one of Goffman's smiles. There are three potential routes for selection of the subject.
The first route in its selection may be followed because of the researcher's familiarity with it-a local knowledge case-and this will be relevant particularly for the practitioner or student researcher. In one's own place of work, one's placement, or even one's home, there will be intimate knowledge and ample opportunity for informed, in-depth analysis-ample opportunity for identification and discussion, in the words of Bates et al. (1998, pp. 13-14), of "the actors, the decision points they faced, the choices they made, the paths taken and shunned, and the manner in which their choices generated events and outcomes." The local knowledge case is eminently amenable to the "soak and poke" of Fenno (1986Fenno ( , 1990) since the inquirer is already soaked, and in a good position, one hopes, to poke.
Second, the subject may come into focus because of the inherent interest of the case-it may be a key case of a phenomenon or, third, may illuminate the object by virtue of its difference, its outlier status. The latter is what Lijphart (1971, p. 692) refers to as the deviant case. The essence in both types is in gaining what I call elsewhere (Thomas, 2010) "exemplary knowledge": The "key-ness" or "outlierness" of the case is manifested in its capacity to exemplify the analytical object of the inquiry. This ability to exemplify draws its legitimacy from the phronesis of the case inquirer (together with that of the reader of the case inquiry), and I have argued that the exemplary knowledge thus drawn is distinct from the generalizable knowledge associated with induction. While opinions on the significance of generalization in case study differ (compare de Vaus, 2001;Flyvbjerg, 2006;Gomm, Hammersley, & Foster, 2000;Yin, 2009)-with discussion of varieties of generalization spanning naturalistic generalization (Stake, 1995) to holographic generalization (Lincoln & Guba, 1985) to fuzzy generalization (Bassey, 2001)-I have argued that the validity of the case study cannot derive from its representativeness since it can never legitimately be claimed to form a representative sample from a larger set. The essence of selection must rest in the dynamic of the relation between subject and object. It cannot rest in typicality.
In this choice of the subject, then, I disagree with Yin (2009, p. 48) when he suggests that a case may be selected because it is "representative or typical." Even if we know that a case is typical following some empirical work to show that it is typical-a typical Chicago street, say, in terms of the ethnicity and age distribution of its inhabitants-we cannot draw anything meaningful from this typicality in a case study, for the typicality will begin and end with the dimensions by which typicality is framed. We cannot say from having studied this street that its circumstances will have in any way contributed by their typicality to the particular situation in which it finds itself (whatever that situation, that "object," is). We could study the street and be informed about its problems, its tensions, its intrigues, hostilities, and kindnesses, and while these may in some way be of interest by virtue of the analytical object of the study, they would not be of interest by virtue of the street's typicality since the next typical street would, in terms of such dynamics, in all probability, be very different. I explore these distinctions in more detail elsewhere (Thomas, 2010;2011;2012): In short, the notion of typicality may give an unwarranted impression to any reader that the significance of the analysis rests in the representativeness of the subject. It does not.
I take the subject to be identified, then, in one of three principal ways-as a • local knowledge case, or a • key case, or an • outlier case

Object
The object is less straightforwardly identified, and, as Ragin (1992) notes, it need not be defined at the outset but, rather, may emerge as an inquiry progresses. Whether it is set at the outset or is emergent, it will be this analytical focus that crystallizes, thickens, or develops as the study proceeds: It is the way that this "object" develops that is at the heart of the study.
Whichever-"emergent" or set at the outset-it is important to have some notion of a potential object in mind when the study begins and not to confuse it with the subject. As Wieviorka (1992) puts it: If you want to talk about a "case," you also need the means of interpreting it or placing it in a context. . . . Regardless of the practical approach for studying it, a case is an opportunity of relating facts and concepts, reality and hypotheses. But do not make the mistake of thinking that it is, in itself, a concept. (p. 160) The object constitutes, then, the analytical frame within which the case is viewed and which the case exemplifies. For example, in Ball's (1981) case study of a school, Beachside Comprehensive, the school itself is the subject that exemplifies the analytical frame, the object, which was the process by which change was effected in schools in the movement to comprehensive education in the United Kingdom. Beachside Comprehensive-the school-was the prism through which "facts and concepts, reality and hypotheses" about this change were refracted, viewed and studied. Becker (in Ragin, 1992) shows how important it is to see the process of employing the object as a dynamic one: As a study proceeds the inquirer should be asking the question, "What is this a case of" over and over as evidence accumulates around potential explanations or "theories." As I have noted elsewhere (Thomas, 1997;2007) theory is thus forged-it is malleable, rather in the way that Bourdieu talked about theory (in Wacquant, 1989, cited in Jenkins, 1992, p. 67) being a "thinking tool." As Bourdieu put it, "[theory is] a set of thinking tools visible through the results they yield, but it is not built as such. . . . It is a temporary construct which takes shape for and by empirical work." Eckstein (1975, p. 133) makes the same point, noting that the theoretical enterprise of case study is not about testing probabilistically stated theories. Rather, it is about discovering or testing tools of explanation.
The focus on the development of theory in case study is closely linked with the explication of the analytical object. Bourdieu's emphasis on the theory as tool, therefore, reminds us that the elaboration of theory is a means to an end, with that end being explanation. It is not an end in itself. The development of theory, whether this be in "theory-testing" or "theory-seeking," is central to the dynamic of the relation between subject and object in case study, and I explore it further in the next section.

Beyond Subject and Object: Purpose, Approach, and Process
Methodological discourse stresses a number of themes on the direction and organization of case studies-their design-and I summarize some of the better-known analyses in Table 1. Constraint of space prohibits full discussion of all of these, but I shall outline in a little more detail one of the most recent-the analysis from George and Bennett (2005, pp. 75-76)-for the purposes of explicating the general themes raised by Table 1. Theirs is an especially useful analysis, drawing heavily as it does on the widely used typologies of Lijphart (1971) and, principally, Eckstein (1975). George and Bennett emerge with six types of case study. These are: 1. Atheoretical/configurative idiographic case studiesthat is to say, illustrative studies that do not contribute to theory; 2. Disciplined configurative case studies, where established theories are used to explain a case; 3. Heuristic case studies wherein new causal paths are identified. Outlier cases may be especially valuable here; 4. Theory testing case studies, assessing "the validity and scope conditions of single or competing theories"; 5. Plausibility probes-preliminary studies to determine whether further study is warranted; and 6. "Building Block" studies of particular types or subtypes "of a phenomenon identify common patterns or serve a particular kind of heuristic purpose." In the establishment of these six types a core distinction is being drawn between theoretical and nontheoretical studies, and this is a feature of several of the other classifications in Table 1. Beyond this, the classification draws attention to illustrative and exploratory studies of one kind or another, as do the other classifications in Table 1. Unlike Yin and de Vaus, George and Bennett do not expand their discussion to a further layer of organization: the operationalization of the study-for example, into "parallel," "longitudinal," or "embedded" studies.
Notwithstanding these commonalities and differences, the principal feature to emerge from a listing of this kind is that there is a mixture of criteria for classification. My aim in developing a typology is to synthesize by drawing out strands of commonality while also integrating, where appropriate, classificatory layers and themes-and noting, hopefully to understand, differences. Within and between the commentaries I have selected, purposes are mixed with methods that are mixed with kinds of subject that are in turn mixed with what might be called different operational "shapes" of case study. I shall look at these layers of analysis in turn.

Purpose
There is first a layer of criteria that is about purpose. For example, the terms "intrinsic" and "instrumental" used by Stake, and the term "evaluative" as used by Merriam and Bassey, point to a reason for doing the study: its purpose. Likewise, the term "plausibility probes" used by Mitchell and George and Bennett points to a purpose-of exploration. And Eckstein uses the term "heuristic" to refer to exploration; he suggests that heuristic studies can be about arriving at notions of problems to solve. The purpose is intimately connected with the object of the study: The understanding that is required-the explanation that is needed-will be related to the reason for doing the study, that is to say, the purpose.

Approach and Methods
Next, there is the approach that is adopted. It is in this "layer" that there are the clearest distinctions between kinds of study, reflecting the broad nature of the object and the purpose of the study. Even though those differences exist, a centrality is given in the commentaries to the significance of theory in the conduct of the study, wherein studies that are not in some way theoretical are specifically labeled as such. Thus, in the categorizations of George and Bennett and that of Mitchell (both borrowing from Eckstein) case studies that have no theoretical element are termed "atheoretical/configurative-idiographic," in part to highlight the illustrative nature of the work in hand. As Lijphart (1971, p. 691) puts it, these nontheoretical studies "are entirely descriptive and move in a theoretical vacuum." Likewise, Bassey makes a distinction between, on the one side, two kinds of theoretical case study (theory seeking and theory testing), and on the other those he labels "picture drawing" and "story telling." Thus, one might say that the object of a study may be, essentially (a) theoretical, or (b) illustrative. As far as the former is concerned-the theoretical study-the distinction Bassey draws between theory testing and theory seeking highlights the different kinds of stance that may be taken about the object: It may be set clearly at the outset (theory testing) or developed throughout the study (theory seeking).
After a decision about approach, there are choices to be made about the methods to be adopted. Will the study be entirely interpretative in orientation: Will it be an ethnography? Will it use a combination of methods, possibly incorporating experimental (e.g., using "repeated measures" as in Stake's example), survey, or cross-sectional elements? Will it involve documentary analysis? Given the methodological pluralism noted earlier the choices here are abundant. They will, in turn, lead to questions about the operational process of the study-the means by which it is constructed and the means by which the object is understood and refracted through the subject. It is this operational process to which I now turn.

Process
In this classificatory layer, case inquirers are making decisions about the operational processes of their studies. For this, they need first to return to their subjects (as distinct from the object) and to the boundary decisions made at the outset. There has to be an examination of the nature of the choices that were made at that time about the parameters that delimit the subject of the study. These may fall around a number of loci: The case may be defined by one or more of a range of boundary considerations: person, time period, place, event, institution, or any of a range of singular phenomena that can be studied in their complexity. The first consideration, though, concerns an important distinction that has been raised by Stake (2005, p. 445) that will determine the process of the case study, and this is about whether there is to be a comparative element to the study: Should it be single or multiple? This single/multiple distinction is at the base of much discussion-and confusion-about the case study, emerging principally from Lijphart's exclusion of what he called (and is still sometimes called) "comparative research" from the case study family. I shall discuss the latter in a moment, after considering the central distinction.
Single or multiple. The case study, while it is of the singular, may contain more than one element in its subject and if this is so-that is, if there are two or several cases-each individual case is less important in itself than the comparison that each offers with the others. For example, a study might be conducted of two schools' different capacities for making effective use of a visiting education support service. By contrasting the schools' "biographies"-their histories, catchments, staff relationships, and other characteristicslight would be thrown on the relative dynamics affecting the reception and use of the support service. The key focus would not be on the nature and shape of relationships per se in one school but rather on the nature of the difference between the one and the other and what this might tell us about the dynamics that were significant in this difference. This comparative element is why Schwandt (2001) calls this kind of case study cross-case analysis.
But one now needs to raise the methodological issue to which I alluded a moment ago concerning the firm distinction that Lijphart (1971) posited between comparative study and case study. It is a differentiation that has been troubling for subsequent discourse about the nature and "shape" of case study. Lijphart's influential typology-his six types of case study, distinguished, as they are, from comparative study-presents to us, if we do not read them in the context of four subsequent decades of methodological discussion, some profound misunderstandings. Not many would now agree with Lijphart, for example, that "the analytical power of the comparative method increases the closer it approximates the statistical and experimental methods" (p. 693). Lijphart's epistemological stance, disclosed by comments such as this throughout his seminal article perhaps betrays the methodological tensions existing at the time he was writing (see Alvesson & Sköldberg, 2000). Whatever the reason for his stance, the legacy of his disaggregation of the comparative study from the case study has been confusion about the nature of the case. Suffice it to say that the comparative study is more straightforwardly seen as part of the case study family if one puts the emphasis on the subjectwhich can be singular or plural-rather than the case.
The boundary and the shape. The choice about single or multiple studies determines what follows in the shape of the case study. Single studies, containing no element of comparison, will take essentially three forms, wherein personal or systemic features of the subject are bounded by time in some shape or form. The case inquirer notices change as it happens and seeks its antecedents and its consequences. We have to find the "sequence of steps" as Becker puts it (1992, p. 209) and understand cause in relation to time, with "each step understood as preceding in time the one that follows it." In doing this we conjecture not only about how one thing is related to another but also about how cause and effect change with time as other elements of a situation also change.
I suggest (drawing on other commentators) that the varieties of time-use lead to three kinds of study: retrospective, snapshot, and diachronic. The retrospective study is the simplest, involving the collection of data relating to a past phenomenon of any kind. The researcher is looking back on a phenomenon, situation, person, or event or studying it in its historical integrity. With the snapshot the case is being examined in one defined period of time: a current event, a day in the life of a person, a month's diary of a marriage. Whether a month, a week, a day, or even a period as short as an hour, the analysis will be aided by the temporal juxtaposition of events. The snapshot develops, the picture presenting itself as a Gestalt over a tight time frame. The diachronic study shows change over time. I use the term "diachronic" to refer to change over time in preference to the word "longitudinal" principally to avoid confusion with other kinds of longitudinal research. The essence, though, is the same as that in "longitudinal": Data capture occurs at points a, b, c . . . n, and one's interest is in the changes occurring at the two or more data collection points.
For multiple studies the researcher considers additional features of the situation. How can the different studies be used for comparison-for cross-case analysis in Schwandt's (2001) terms? There are two principal means of doing this: first by straightforward comparison between clearly different examples, as in Burgess's (1984) 10 case studies of research in educational settings, and the contrast between and among the cases throws the spotlight on an important theoretical feature. Second, comparison may be of elements within one case-comparison, in other words, of nested elements. With nested studies the breakdown is within the principal unit of analysis-for example, wards within a hospital. A nested study is distinct from a straightforwardly multiple study in that it gains its integrity, its wholeness, from the wider case. For example, a researcher might be looking at three wards within one hospital, but if the one hospital had no significance other than its physical housing of these three wards then the cases would not be seen as nested. The elements are nested only in the sense that they form an integral part of a broader picture.
A further subdivision may be drawn in the multiple study, and this is between parallel and sequential studies. In the first, the parallel study, the cases are all happening and being studied concurrently, while with the sequential study the cases happen consecutively and there is an assumption that what has happened in one or in an intervening period will in some way affect the next.

Integrating the Layers: A Typology
Having separated the classificatory layers drawn in discourse about case study, I now propose a typology in which they are organized and reintegrated. The typology, incorporating considerations about these layers-concerning subject and object, purpose, approach, and process-is summarized in Figure 1. While this perhaps implies sequencing to the choices being made, in most cases much of the decision making will in fact occur simultaneously, particularly in relation to the subject, object, and approach. The typology offers a "flattened-out" view of the thinking that occurs in the process of research design.
As the typology in a sense "unrolls" the various considerations being made in the design of a case study it perhaps implies that a series of separate design choices are being made during the planning of the study. It perhaps thus, at first glance, denies the coherence and simultaneity of the design decisions of which I have just spoken. But for the researcher new to case study the mere existence of these decisions may not have occurred. As a consequence, the variety of design paths will be restricted. A typology encourages a clear articulation of the distinctness and necessity of both subject and object; it encourages consideration of theoretical or illustrative approaches, methodological decisions, and decisions about process: Can the research question be addressed by a single focus on one person or situation, or would a comparison be better? Is there a time element that will be addressed by looking at a sequence of events, or is it better to examine one tightly defined period in time? Would it be helpful to extract a number of nested elements from the main focus and to examine these in detail? It is useful to explore all of these considerations alongside thought about subject, object, theory, and method.

Two Examples: Routes Through the Typology
Using a set of linked case studies, Diamond (2005) develops a complex thesis about how societies collapse. He looks at a range of communities and societies that have collapsed or are currently in the process of collapsing. The title-Collapse: How Societies Choose to Fail or Survive-is the object of the research, while the subject is the set of societies he selects to illuminate and understand the object: today's Montana, Easter Island of the 17th century, Pitcairn, the Mayans, the Vikings, and modern Haiti. Each one is quite different from the others, but they are connected by decline and collapse, the object. The theoretical explanation that "thickens" around the object is that the decline of a society happens because the society's response to change is insufficient to address a range of potential factors, from hostile neighbors to environmental change and degradation.
So, using the typology, Diamond's was a multiple study possessing both subject and object. It involved key examples of collapse. It was of intrinsic interest, containing exploratory elements, and the aim was to explain collapse through theory building. Let me take just one example from his multiple study, that of Easter Island, to show how he proceeded. He theorized about how the Easter Islanders sought and used resources, from the intensification of their agriculture using windbreaks and pits for better growing to the division of the island into a number of territories that eventually competed with one another. Using a variety of methods and data collection techniques from the use of existing data sets-botanical, cultural, historical-to visiting and interviewing he examined Easter Island's history, geography, and culture. He looked at what the people ate, drawing from research using pollen and charcoal remains that gave clues about now-extinct crops. Using archaeological evidence and oral history he looked and the likely structure of the society, at the people's attitudes to death and the afterlife, at how statues might have been sculpted, transported, and erected. He then wove all of this together using intelligent questioning and answering-essentially, the Socratic method-in the context of the other, parallel studies he was collecting to come to a thesis about collapse. While there would have been an integrity to much of the decision making in this design, an analysis of that decision  Figure 3. Zigler and Muenchow's (1992) study of the Head Start education program making can be displayed as a trajectory through the typology, and this is summarized in Figure 2.
As a second example I have taken Zigler and Muenchow's (1992) study of the Head Start education program. It is a much-cited study with a clear subject (Head Start) but perhaps a less easily identifiable object. It is a local knowledge case-these were key figures in the program, coauthoring the presidential report on it. The object could perhaps be said to be the working and the success of the program, with lessons to be learned, but this is not clearly defined. It is, essentially, a chronicle with observations, analyses of what went right and wrong, and a set of conclusions and recommendations. It is not contextualized, as Wieviorka (1992, p. 160) puts it, in an "analytical category or theory"-whether that be expensive national education programs, attempts at top-down change, the operation of early childhood programs, or whatever. What was it, in other words, that was to be explained? What was the explanandum? The study was essentially descriptive and illustrative, and we can say that via various methods-the recollections of the authors, discussions with others, and interrogation of official and unofficial records-this was a single, retrospective study (see Figure 3).
As a rider, one might note that the comparison of the two cases via the typology gives a means for assessing what a case study is trying to do and how it proceeds to do that.

Concluding Comments
The "weak sibling" status of the case study noted by Yin (2009, p. xiii) is due at least in part to the uncertainty felt by intending researchers about structure and method. As the design of the case study is presented often as open-ended and untethered-and methodological eclecticism is emphasized in commentary on design-researchers may feel unguided about structure: Open-endedness is extended to an unwarranted expectation of structural looseness, and, in the absence of a structure that maps out potential routes to follow, important pointers may be missed. I have, therefore, suggested a typology that foregrounds a number of features-classificatory layers-of the study: the distinction between subject and object, the importance of clarifying the purpose of the study, an awareness of the likely analytical approach to be pursued, and an identification of the likely process to be followed in conducting it. I hope that the typology will assist in both the construction and analysis of case studies.

Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding
The author received no financial support for the research, authorship, and/or publication of this article.