case study and theory

An official website of the United States government

Official websites use .gov A .gov website belongs to an official government organization in the United States.

Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

• Publications
• Account settings
• Advanced Search
• Journal List

The case study approach

Sarah crowe, kathrin cresswell, ann robertson, anthony avery, aziz sheikh.

• Author information
• Article notes
• Copyright and License information

Corresponding author.

Received 2010 Nov 29; Accepted 2011 Jun 27; Collection date 2011.

This is an Open Access article distributed under the terms of the Creative Commons Attribution License ( http://creativecommons.org/licenses/by/2.0 ), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

The case study approach allows in-depth, multi-faceted explorations of complex issues in their real-life settings. The value of the case study approach is well recognised in the fields of business, law and policy, but somewhat less so in health services research. Based on our experiences of conducting several health-related case studies, we reflect on the different types of case study design, the specific research questions this approach can help answer, the data sources that tend to be used, and the particular advantages and disadvantages of employing this methodological approach. The paper concludes with key pointers to aid those designing and appraising proposals for conducting case study research, and a checklist to help readers assess the quality of case study reports.

Introduction

The case study approach is particularly useful to employ when there is a need to obtain an in-depth appreciation of an issue, event or phenomenon of interest, in its natural real-life context. Our aim in writing this piece is to provide insights into when to consider employing this approach and an overview of key methodological considerations in relation to the design, planning, analysis, interpretation and reporting of case studies.

The illustrative 'grand round', 'case report' and 'case series' have a long tradition in clinical practice and research. Presenting detailed critiques, typically of one or more patients, aims to provide insights into aspects of the clinical case and, in doing so, illustrate broader lessons that may be learnt. In research, the conceptually-related case study approach can be used, for example, to describe in detail a patient's episode of care, explore professional attitudes to and experiences of a new policy initiative or service development or more generally to 'investigate contemporary phenomena within its real-life context' [ 1 ]. Based on our experiences of conducting a range of case studies, we reflect on when to consider using this approach, discuss the key steps involved and illustrate, with examples, some of the practical challenges of attaining an in-depth understanding of a 'case' as an integrated whole. In keeping with previously published work, we acknowledge the importance of theory to underpin the design, selection, conduct and interpretation of case studies[ 2 ]. In so doing, we make passing reference to the different epistemological approaches used in case study research by key theoreticians and methodologists in this field of enquiry.

This paper is structured around the following main questions: What is a case study? What are case studies used for? How are case studies conducted? What are the potential pitfalls and how can these be avoided? We draw in particular on four of our own recently published examples of case studies (see Tables 1 , 2 , 3 and 4 ) and those of others to illustrate our discussion[ 3 - 7 ].

Example of a case study investigating the reasons for differences in recruitment rates of minority ethnic people in asthma research[ 3 ]

Example of a case study investigating the process of planning and implementing a service in Primary Care Organisations[ 4 ]

Example of a case study investigating the introduction of the electronic health records[ 5 ]

Example of a case study investigating the formal and informal ways students learn about patient safety[ 6 ]

What is a case study?

A case study is a research approach that is used to generate an in-depth, multi-faceted understanding of a complex issue in its real-life context. It is an established research design that is used extensively in a wide variety of disciplines, particularly in the social sciences. A case study can be defined in a variety of ways (Table 5 ), the central tenet being the need to explore an event or phenomenon in depth and in its natural context. It is for this reason sometimes referred to as a "naturalistic" design; this is in contrast to an "experimental" design (such as a randomised controlled trial) in which the investigator seeks to exert control over and manipulate the variable(s) of interest.

Definitions of a case study

Stake's work has been particularly influential in defining the case study approach to scientific enquiry. He has helpfully characterised three main types of case study: intrinsic , instrumental and collective [ 8 ]. An intrinsic case study is typically undertaken to learn about a unique phenomenon. The researcher should define the uniqueness of the phenomenon, which distinguishes it from all others. In contrast, the instrumental case study uses a particular case (some of which may be better than others) to gain a broader appreciation of an issue or phenomenon. The collective case study involves studying multiple cases simultaneously or sequentially in an attempt to generate a still broader appreciation of a particular issue.

These are however not necessarily mutually exclusive categories. In the first of our examples (Table 1 ), we undertook an intrinsic case study to investigate the issue of recruitment of minority ethnic people into the specific context of asthma research studies, but it developed into a instrumental case study through seeking to understand the issue of recruitment of these marginalised populations more generally, generating a number of the findings that are potentially transferable to other disease contexts[ 3 ]. In contrast, the other three examples (see Tables 2 , 3 and 4 ) employed collective case study designs to study the introduction of workforce reconfiguration in primary care, the implementation of electronic health records into hospitals, and to understand the ways in which healthcare students learn about patient safety considerations[ 4 - 6 ]. Although our study focusing on the introduction of General Practitioners with Specialist Interests (Table 2 ) was explicitly collective in design (four contrasting primary care organisations were studied), is was also instrumental in that this particular professional group was studied as an exemplar of the more general phenomenon of workforce redesign[ 4 ].

What are case studies used for?

According to Yin, case studies can be used to explain, describe or explore events or phenomena in the everyday contexts in which they occur[ 1 ]. These can, for example, help to understand and explain causal links and pathways resulting from a new policy initiative or service development (see Tables 2 and 3 , for example)[ 1 ]. In contrast to experimental designs, which seek to test a specific hypothesis through deliberately manipulating the environment (like, for example, in a randomised controlled trial giving a new drug to randomly selected individuals and then comparing outcomes with controls),[ 9 ] the case study approach lends itself well to capturing information on more explanatory ' how ', 'what' and ' why ' questions, such as ' how is the intervention being implemented and received on the ground?'. The case study approach can offer additional insights into what gaps exist in its delivery or why one implementation strategy might be chosen over another. This in turn can help develop or refine theory, as shown in our study of the teaching of patient safety in undergraduate curricula (Table 4 )[ 6 , 10 ]. Key questions to consider when selecting the most appropriate study design are whether it is desirable or indeed possible to undertake a formal experimental investigation in which individuals and/or organisations are allocated to an intervention or control arm? Or whether the wish is to obtain a more naturalistic understanding of an issue? The former is ideally studied using a controlled experimental design, whereas the latter is more appropriately studied using a case study design.

Case studies may be approached in different ways depending on the epistemological standpoint of the researcher, that is, whether they take a critical (questioning one's own and others' assumptions), interpretivist (trying to understand individual and shared social meanings) or positivist approach (orientating towards the criteria of natural sciences, such as focusing on generalisability considerations) (Table 6 ). Whilst such a schema can be conceptually helpful, it may be appropriate to draw on more than one approach in any case study, particularly in the context of conducting health services research. Doolin has, for example, noted that in the context of undertaking interpretative case studies, researchers can usefully draw on a critical, reflective perspective which seeks to take into account the wider social and political environment that has shaped the case[ 11 ].

Example of epistemological approaches that may be used in case study research

How are case studies conducted?

Here, we focus on the main stages of research activity when planning and undertaking a case study; the crucial stages are: defining the case; selecting the case(s); collecting and analysing the data; interpreting data; and reporting the findings.

Defining the case

Carefully formulated research question(s), informed by the existing literature and a prior appreciation of the theoretical issues and setting(s), are all important in appropriately and succinctly defining the case[ 8 , 12 ]. Crucially, each case should have a pre-defined boundary which clarifies the nature and time period covered by the case study (i.e. its scope, beginning and end), the relevant social group, organisation or geographical area of interest to the investigator, the types of evidence to be collected, and the priorities for data collection and analysis (see Table 7 )[ 1 ]. A theory driven approach to defining the case may help generate knowledge that is potentially transferable to a range of clinical contexts and behaviours; using theory is also likely to result in a more informed appreciation of, for example, how and why interventions have succeeded or failed[ 13 ].

Example of a checklist for rating a case study proposal[ 8 ]

For example, in our evaluation of the introduction of electronic health records in English hospitals (Table 3 ), we defined our cases as the NHS Trusts that were receiving the new technology[ 5 ]. Our focus was on how the technology was being implemented. However, if the primary research interest had been on the social and organisational dimensions of implementation, we might have defined our case differently as a grouping of healthcare professionals (e.g. doctors and/or nurses). The precise beginning and end of the case may however prove difficult to define. Pursuing this same example, when does the process of implementation and adoption of an electronic health record system really begin or end? Such judgements will inevitably be influenced by a range of factors, including the research question, theory of interest, the scope and richness of the gathered data and the resources available to the research team.

Selecting the case(s)

The decision on how to select the case(s) to study is a very important one that merits some reflection. In an intrinsic case study, the case is selected on its own merits[ 8 ]. The case is selected not because it is representative of other cases, but because of its uniqueness, which is of genuine interest to the researchers. This was, for example, the case in our study of the recruitment of minority ethnic participants into asthma research (Table 1 ) as our earlier work had demonstrated the marginalisation of minority ethnic people with asthma, despite evidence of disproportionate asthma morbidity[ 14 , 15 ]. In another example of an intrinsic case study, Hellstrom et al.[ 16 ] studied an elderly married couple living with dementia to explore how dementia had impacted on their understanding of home, their everyday life and their relationships.

For an instrumental case study, selecting a "typical" case can work well[ 8 ]. In contrast to the intrinsic case study, the particular case which is chosen is of less importance than selecting a case that allows the researcher to investigate an issue or phenomenon. For example, in order to gain an understanding of doctors' responses to health policy initiatives, Som undertook an instrumental case study interviewing clinicians who had a range of responsibilities for clinical governance in one NHS acute hospital trust[ 17 ]. Sampling a "deviant" or "atypical" case may however prove even more informative, potentially enabling the researcher to identify causal processes, generate hypotheses and develop theory.

In collective or multiple case studies, a number of cases are carefully selected. This offers the advantage of allowing comparisons to be made across several cases and/or replication. Choosing a "typical" case may enable the findings to be generalised to theory (i.e. analytical generalisation) or to test theory by replicating the findings in a second or even a third case (i.e. replication logic)[ 1 ]. Yin suggests two or three literal replications (i.e. predicting similar results) if the theory is straightforward and five or more if the theory is more subtle. However, critics might argue that selecting 'cases' in this way is insufficiently reflexive and ill-suited to the complexities of contemporary healthcare organisations.

The selected case study site(s) should allow the research team access to the group of individuals, the organisation, the processes or whatever else constitutes the chosen unit of analysis for the study. Access is therefore a central consideration; the researcher needs to come to know the case study site(s) well and to work cooperatively with them. Selected cases need to be not only interesting but also hospitable to the inquiry [ 8 ] if they are to be informative and answer the research question(s). Case study sites may also be pre-selected for the researcher, with decisions being influenced by key stakeholders. For example, our selection of case study sites in the evaluation of the implementation and adoption of electronic health record systems (see Table 3 ) was heavily influenced by NHS Connecting for Health, the government agency that was responsible for overseeing the National Programme for Information Technology (NPfIT)[ 5 ]. This prominent stakeholder had already selected the NHS sites (through a competitive bidding process) to be early adopters of the electronic health record systems and had negotiated contracts that detailed the deployment timelines.

It is also important to consider in advance the likely burden and risks associated with participation for those who (or the site(s) which) comprise the case study. Of particular importance is the obligation for the researcher to think through the ethical implications of the study (e.g. the risk of inadvertently breaching anonymity or confidentiality) and to ensure that potential participants/participating sites are provided with sufficient information to make an informed choice about joining the study. The outcome of providing this information might be that the emotive burden associated with participation, or the organisational disruption associated with supporting the fieldwork, is considered so high that the individuals or sites decide against participation.

In our example of evaluating implementations of electronic health record systems, given the restricted number of early adopter sites available to us, we sought purposively to select a diverse range of implementation cases among those that were available[ 5 ]. We chose a mixture of teaching, non-teaching and Foundation Trust hospitals, and examples of each of the three electronic health record systems procured centrally by the NPfIT. At one recruited site, it quickly became apparent that access was problematic because of competing demands on that organisation. Recognising the importance of full access and co-operative working for generating rich data, the research team decided not to pursue work at that site and instead to focus on other recruited sites.

Collecting the data

In order to develop a thorough understanding of the case, the case study approach usually involves the collection of multiple sources of evidence, using a range of quantitative (e.g. questionnaires, audits and analysis of routinely collected healthcare data) and more commonly qualitative techniques (e.g. interviews, focus groups and observations). The use of multiple sources of data (data triangulation) has been advocated as a way of increasing the internal validity of a study (i.e. the extent to which the method is appropriate to answer the research question)[ 8 , 18 - 21 ]. An underlying assumption is that data collected in different ways should lead to similar conclusions, and approaching the same issue from different angles can help develop a holistic picture of the phenomenon (Table 2 )[ 4 ].

Brazier and colleagues used a mixed-methods case study approach to investigate the impact of a cancer care programme[ 22 ]. Here, quantitative measures were collected with questionnaires before, and five months after, the start of the intervention which did not yield any statistically significant results. Qualitative interviews with patients however helped provide an insight into potentially beneficial process-related aspects of the programme, such as greater, perceived patient involvement in care. The authors reported how this case study approach provided a number of contextual factors likely to influence the effectiveness of the intervention and which were not likely to have been obtained from quantitative methods alone.

In collective or multiple case studies, data collection needs to be flexible enough to allow a detailed description of each individual case to be developed (e.g. the nature of different cancer care programmes), before considering the emerging similarities and differences in cross-case comparisons (e.g. to explore why one programme is more effective than another). It is important that data sources from different cases are, where possible, broadly comparable for this purpose even though they may vary in nature and depth.

Analysing, interpreting and reporting case studies

Making sense and offering a coherent interpretation of the typically disparate sources of data (whether qualitative alone or together with quantitative) is far from straightforward. Repeated reviewing and sorting of the voluminous and detail-rich data are integral to the process of analysis. In collective case studies, it is helpful to analyse data relating to the individual component cases first, before making comparisons across cases. Attention needs to be paid to variations within each case and, where relevant, the relationship between different causes, effects and outcomes[ 23 ]. Data will need to be organised and coded to allow the key issues, both derived from the literature and emerging from the dataset, to be easily retrieved at a later stage. An initial coding frame can help capture these issues and can be applied systematically to the whole dataset with the aid of a qualitative data analysis software package.

The Framework approach is a practical approach, comprising of five stages (familiarisation; identifying a thematic framework; indexing; charting; mapping and interpretation) , to managing and analysing large datasets particularly if time is limited, as was the case in our study of recruitment of South Asians into asthma research (Table 1 )[ 3 , 24 ]. Theoretical frameworks may also play an important role in integrating different sources of data and examining emerging themes. For example, we drew on a socio-technical framework to help explain the connections between different elements - technology; people; and the organisational settings within which they worked - in our study of the introduction of electronic health record systems (Table 3 )[ 5 ]. Our study of patient safety in undergraduate curricula drew on an evaluation-based approach to design and analysis, which emphasised the importance of the academic, organisational and practice contexts through which students learn (Table 4 )[ 6 ].

Case study findings can have implications both for theory development and theory testing. They may establish, strengthen or weaken historical explanations of a case and, in certain circumstances, allow theoretical (as opposed to statistical) generalisation beyond the particular cases studied[ 12 ]. These theoretical lenses should not, however, constitute a strait-jacket and the cases should not be "forced to fit" the particular theoretical framework that is being employed.

When reporting findings, it is important to provide the reader with enough contextual information to understand the processes that were followed and how the conclusions were reached. In a collective case study, researchers may choose to present the findings from individual cases separately before amalgamating across cases. Care must be taken to ensure the anonymity of both case sites and individual participants (if agreed in advance) by allocating appropriate codes or withholding descriptors. In the example given in Table 3 , we decided against providing detailed information on the NHS sites and individual participants in order to avoid the risk of inadvertent disclosure of identities[ 5 , 25 ].

What are the potential pitfalls and how can these be avoided?

The case study approach is, as with all research, not without its limitations. When investigating the formal and informal ways undergraduate students learn about patient safety (Table 4 ), for example, we rapidly accumulated a large quantity of data. The volume of data, together with the time restrictions in place, impacted on the depth of analysis that was possible within the available resources. This highlights a more general point of the importance of avoiding the temptation to collect as much data as possible; adequate time also needs to be set aside for data analysis and interpretation of what are often highly complex datasets.

Case study research has sometimes been criticised for lacking scientific rigour and providing little basis for generalisation (i.e. producing findings that may be transferable to other settings)[ 1 ]. There are several ways to address these concerns, including: the use of theoretical sampling (i.e. drawing on a particular conceptual framework); respondent validation (i.e. participants checking emerging findings and the researcher's interpretation, and providing an opinion as to whether they feel these are accurate); and transparency throughout the research process (see Table 8 )[ 8 , 18 - 21 , 23 , 26 ]. Transparency can be achieved by describing in detail the steps involved in case selection, data collection, the reasons for the particular methods chosen, and the researcher's background and level of involvement (i.e. being explicit about how the researcher has influenced data collection and interpretation). Seeking potential, alternative explanations, and being explicit about how interpretations and conclusions were reached, help readers to judge the trustworthiness of the case study report. Stake provides a critique checklist for a case study report (Table 9 )[ 8 ].

Potential pitfalls and mitigating actions when undertaking case study research

Stake's checklist for assessing the quality of a case study report[ 8 ]

Conclusions

The case study approach allows, amongst other things, critical events, interventions, policy developments and programme-based service reforms to be studied in detail in a real-life context. It should therefore be considered when an experimental design is either inappropriate to answer the research questions posed or impossible to undertake. Considering the frequency with which implementations of innovations are now taking place in healthcare settings and how well the case study approach lends itself to in-depth, complex health service research, we believe this approach should be more widely considered by researchers. Though inherently challenging, the research case study can, if carefully conceptualised and thoughtfully undertaken and reported, yield powerful insights into many important aspects of health and healthcare delivery.

Competing interests

The authors declare that they have no competing interests.

Authors' contributions

AS conceived this article. SC, KC and AR wrote this paper with GH, AA and AS all commenting on various drafts. SC and AS are guarantors.

Pre-publication history

The pre-publication history for this paper can be accessed here:

http://www.biomedcentral.com/1471-2288/11/100/prepub

Contributor Information

Sarah Crowe, Email: [email protected].

Kathrin Cresswell, Email: [email protected].

Ann Robertson, Email: [email protected].

Guro Huby, Email: [email protected].

Anthony Avery, Email: [email protected].

Aziz Sheikh, Email: [email protected].

Acknowledgements

We are grateful to the participants and colleagues who contributed to the individual case studies that we have drawn on. This work received no direct funding, but it has been informed by projects funded by Asthma UK, the NHS Service Delivery Organisation, NHS Connecting for Health Evaluation Programme, and Patient Safety Research Portfolio. We would also like to thank the expert reviewers for their insightful and constructive feedback. Our thanks are also due to Dr. Allison Worth who commented on an earlier draft of this manuscript.

• Yin RK. Case study research, design and method. 4. London: Sage Publications Ltd.; 2009. [ Google Scholar ]
• Keen J, Packwood T. Qualitative research; case study evaluation. BMJ. 1995;311:444–446. doi: 10.1136/bmj.311.7002.444. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Sheikh A, Halani L, Bhopal R, Netuveli G, Partridge M, Car J. et al. Facilitating the Recruitment of Minority Ethnic People into Research: Qualitative Case Study of South Asians and Asthma. PLoS Med. 2009;6(10):1–11. doi: 10.1371/journal.pmed.1000148. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Pinnock H Huby G Powell A Kielmann T Price D Williams S et al. The process of planning, development and implementation of a General Practitioner with a Special Interest service in Primary Care Organisations in England and Wales: a comparative prospective case study Report for the National Co-ordinating Centre for NHS Service Delivery and Organisation R&D (NCCSDO) 2008 http://www.sdo.nihr.ac.uk/files/project/99-final-report.pdf 15318916
• Robertson A, Cresswell K, Takian A, Petrakaki D, Crowe S, Cornford T. et al. Prospective evaluation of the implementation and adoption of NHS Connecting for Health's national electronic health record in secondary care in England: interim findings. BMJ. 2010;41:c4564. doi: 10.1136/bmj.c4564. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Pearson P, Steven A, Howe A, Sheikh A, Ashcroft D, Smith P. the Patient Safety Education Study Group. Learning about patient safety: organisational context and culture in the education of healthcare professionals. J Health Serv Res Policy. 2010;15:4–10. doi: 10.1258/jhsrp.2009.009052. [ DOI ] [ PubMed ] [ Google Scholar ]
• van Harten WH, Casparie TF, Fisscher OA. The evaluation of the introduction of a quality management system: a process-oriented case study in a large rehabilitation hospital. Health Policy. 2002;60(1):17–37. doi: 10.1016/S0168-8510(01)00187-7. [ DOI ] [ PubMed ] [ Google Scholar ]
• Stake RE. The art of case study research. London: Sage Publications Ltd.; 1995. [ Google Scholar ]
• Sheikh A, Smeeth L, Ashcroft R. Randomised controlled trials in primary care: scope and application. Br J Gen Pract. 2002;52(482):746–51. [ PMC free article ] [ PubMed ] [ Google Scholar ]
• King G, Keohane R, Verba S. Designing Social Inquiry. Princeton: Princeton University Press; 1996. [ Google Scholar ]
• Doolin B. Information technology as disciplinary technology: being critical in interpretative research on information systems. Journal of Information Technology. 1998;13:301–311. doi: 10.1057/jit.1998.8. [ DOI ] [ Google Scholar ]
• George AL, Bennett A. Case studies and theory development in the social sciences. Cambridge, MA: MIT Press; 2005. [ Google Scholar ]
• Eccles M. the Improved Clinical Effectiveness through Behavioural Research Group (ICEBeRG) Designing theoretically-informed implementation interventions. Implementation Science. 2006;1:1–8. doi: 10.1186/1748-5908-1-1. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Netuveli G, Hurwitz B, Levy M, Fletcher M, Barnes G, Durham SR, Sheikh A. Ethnic variations in UK asthma frequency, morbidity, and health-service use: a systematic review and meta-analysis. Lancet. 2005;365(9456):312–7. doi: 10.1016/S0140-6736(05)17785-X. [ DOI ] [ PubMed ] [ Google Scholar ]
• Sheikh A, Panesar SS, Lasserson T, Netuveli G. Recruitment of ethnic minorities to asthma studies. Thorax. 2004;59(7):634. [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Hellström I, Nolan M, Lundh U. 'We do things together': A case study of 'couplehood' in dementia. Dementia. 2005;4:7–22. doi: 10.1177/1471301205049188. [ DOI ] [ Google Scholar ]
• Som CV. Nothing seems to have changed, nothing seems to be changing and perhaps nothing will change in the NHS: doctors' response to clinical governance. International Journal of Public Sector Management. 2005;18:463–477. doi: 10.1108/09513550510608903. [ DOI ] [ Google Scholar ]
• Lincoln Y, Guba E. Naturalistic inquiry. Newbury Park: Sage Publications; 1985. [ Google Scholar ]
• Barbour RS. Checklists for improving rigour in qualitative research: a case of the tail wagging the dog? BMJ. 2001;322:1115–1117. doi: 10.1136/bmj.322.7294.1115. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Mays N, Pope C. Qualitative research in health care: Assessing quality in qualitative research. BMJ. 2000;320:50–52. doi: 10.1136/bmj.320.7226.50. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Mason J. Qualitative researching. London: Sage; 2002. [ Google Scholar ]
• Brazier A, Cooke K, Moravan V. Using Mixed Methods for Evaluating an Integrative Approach to Cancer Care: A Case Study. Integr Cancer Ther. 2008;7:5–17. doi: 10.1177/1534735407313395. [ DOI ] [ PubMed ] [ Google Scholar ]
• Miles MB, Huberman M. Qualitative data analysis: an expanded sourcebook. 2. CA: Sage Publications Inc.; 1994. [ Google Scholar ]
• Pope C, Ziebland S, Mays N. Analysing qualitative data. Qualitative research in health care. BMJ. 2000;320:114–116. doi: 10.1136/bmj.320.7227.114. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Cresswell KM, Worth A, Sheikh A. Actor-Network Theory and its role in understanding the implementation of information technology developments in healthcare. BMC Med Inform Decis Mak. 2010;10(1):67. doi: 10.1186/1472-6947-10-67. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Malterud K. Qualitative research: standards, challenges, and guidelines. Lancet. 2001;358:483–488. doi: 10.1016/S0140-6736(01)05627-6. [ DOI ] [ PubMed ] [ Google Scholar ]
• Yin R. Case study research: design and methods. 2. Thousand Oaks, CA: Sage Publishing; 1994. [ Google Scholar ]
• Yin R. Enhancing the quality of case studies in health services research. Health Serv Res. 1999;34:1209–1224. [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Green J, Thorogood N. Qualitative methods for health research. 2. Los Angeles: Sage; 2009. [ Google Scholar ]
• Howcroft D, Trauth E. Handbook of Critical Information Systems Research, Theory and Application. Cheltenham, UK: Northampton, MA, USA: Edward Elgar; 2005. [ Google Scholar ]
• Blakie N. Approaches to Social Enquiry. Cambridge: Polity Press; 1993. [ Google Scholar ]
• Doolin B. Power and resistance in the implementation of a medical management information system. Info Systems J. 2004;14:343–362. doi: 10.1111/j.1365-2575.2004.00176.x. [ DOI ] [ Google Scholar ]
• Bloomfield BP, Best A. Management consultants: systems development, power and the translation of problems. Sociological Review. 1992;40:533–560. [ Google Scholar ]
• Shanks G, Parr A. Proceedings of the European Conference on Information Systems. Naples; 2003. Positivist, single case study research in information systems: A critical analysis. [ Google Scholar ]
• View on publisher site
• PDF (249.1 KB)
• Collections

Similar articles

Cited by other articles, links to ncbi databases.

• Download .nbib .nbib
• Format: AMA APA MLA NLM

Add to Collections

The theory contribution of case study research designs

• Original Research
• Open access
• Published: 16 February 2017
• Volume 10 , pages 281–305, ( 2017 )

Cite this article

You have full access to this open access article

• Hans-Gerd Ridder 1  

178k Accesses

340 Citations

11 Altmetric

Explore all metrics

The objective of this paper is to highlight similarities and differences across various case study designs and to analyze their respective contributions to theory. Although different designs reveal some common underlying characteristics, a comparison of such case study research designs demonstrates that case study research incorporates different scientific goals and collection and analysis of data. This paper relates this comparison to a more general debate of how different research designs contribute to a theory continuum. The fine-grained analysis demonstrates that case study designs fit differently to the pathway of the theory continuum. The resulting contribution is a portfolio of case study research designs. This portfolio demonstrates the heterogeneous contributions of case study designs. Based on this portfolio, theoretical contributions of case study designs can be better evaluated in terms of understanding, theory-building, theory development, and theory testing.

Similar content being viewed by others

Case Study Research

Avoid common mistakes on your manuscript.

1 Introduction

Case study research scientifically investigates into a real-life phenomenon in-depth and within its environmental context. Such a case can be an individual, a group, an organization, an event, a problem, or an anomaly (Burawoy 2009 ; Stake 2005 ; Yin 2014 ). Unlike in experiments, the contextual conditions are not delineated and/or controlled, but part of the investigation. Typical for case study research is non-random sampling; there is no sample that represents a larger population. Contrary to quantitative logic, the case is chosen, because the case is of interest (Stake 2005 ), or it is chosen for theoretical reasons (Eisenhardt and Graebner 2007 ). For within-case and across-case analyses, the emphasis in data collection is on interviews, archives, and (participant) observation (Flick 2009 : 257; Mason 2002 : 84). Case study researchers usually triangulate data as part of their data collection strategy, resulting in a detailed case description (Burns 2000 ; Dooley 2002 ; Eisenhardt 1989 ; Ridder 2016 ; Stake 2005 : 454). Potential advantages of a single case study are seen in the detailed description and analysis to gain a better understanding of “how” and “why” things happen. In single case study research, the opportunity to open a black box arises by looking at deeper causes of the phenomenon (Fiss 2009 ). The case data can lead to the identification of patterns and relationships, creating, extending, or testing a theory (Gomm et al. 2000 ). Potential advantages of multiple case study research are seen in cross-case analysis. A systematic comparison in cross-case analysis reveals similarities and differences and how they affect findings. Each case is analyzed as a single case on its own to compare the mechanisms identified, leading to theoretical conclusions (Vaughan 1992 : 178). As a result, case study research has different objectives in terms of contributing to theory. On the one hand, case study research has its strength in creating theory by expanding constructs and relationships within distinct settings (e.g., in single case studies). On the other hand, case study research is a means of advancing theories by comparing similarities and differences among cases (e.g., in multiple case studies).

Unfortunately, such diverging objectives are often neglected in case study research. Burns ( 2000 : 459) emphasizes: “The case study has unfortunately been used as a ‘catch –all’ category for anything that does not fit into experimental, survey, or historical methods.”

Therefore, this paper compares case study research designs. Such comparisons have been conducted previously regarding their philosophical assumptions and orientations, key elements of case study research, their range of application, and the lacks of methodological procedures in publications. (Baxter and Jack 2008 ; Dooley 2002 ; Dyer and Wilkins 1991 ; Piekkari et al. 2009 ; Welch et al. 2011 ). This paper aims to compare case study research designs regarding their contributions to theory.

Case study research designs will be analyzed regarding their various strengths on a theory continuum. Edmondson and McManus ( 2007 ) initiated a debate on whether the stage of theory fits to research questions, style of data collection, and analyses. Similarly, Colquitt and Zapata-Phelan ( 2007 ) created a taxonomy capturing facets of empirical article’s theoretical contributions by distinguishing between theory-building and theory testing. Corley and Gioia ( 2011 ) extended this debate by focusing on the practicality of theory and the importance of prescience. While these papers consider the whole range of methodological approaches on a higher level, they treat case studies as relatively homogeneous. This paper aims to delve into a deeper level of analysis by solely focusing on case study research designs and their respective fit on this theory continuum. This approach offers a more fine-grained understanding that sheds light on the diversity of case study research designs in terms of their differential theory contributions. Such a deep level of analysis on case study research designs enables more rigor in theory contribution. To analyze alternative case study research designs regarding their contributions to theory, I engage into the following steps:

First, differences between case study research designs are depicted. I outline and compare the case study research designs with regard to the key elements, esp. differences in research questions, frameworks, sampling, data collection, and data analysis. These differences result in a portfolio of various case study research designs.

Second, I outline and substantiate a theory continuum that varies between theory-building, theory development, and testing theory. Based on this continuum, I analyze and discuss each of the case study research designs with regard to their location on the theory continuum. This analysis is based on a detailed differentiation of the phenomenon (inside or outside the theory), the status of the theory, research strategy, and methods.

As a result, the contribution to the literature is a portfolio of case study research designs explicating their unique contributions to theory. The contribution of this paper lies in a fine-grained analysis of the interplay of methods and theory (van Maanen et al. 2007 ) and the methodological fit (Edmondson and McManus 2007 ) of case study designs and the continuum of theory. It demonstrates that different designs have various strengths and that there is a fit between case study designs and different points on a theory continuum. If there is no clarity as to whether a case study design aims at creating, elaborating, extending, or testing theory, the contribution to theory is difficult to identify for authors, reviewers, and readers. Consequently, this paper aims to clarify at which point of the continuum of theory case study research designs can provide distinct contributions that can be identified beyond their traditionally claimed exploratory character.

2 Differences across case study design: a portfolio approach

Only few papers have compared case study research designs so far. In all of these comparisons, the number of designs differs as well as the issues under consideration. In an early debate between Dyer and Wilkins ( 1991 ) and Eisenhardt ( 1991 ), Dyer and Wilkins compared the case study research design by Eisenhardt ( 1989 ) with “classical” case studies. The core of the debate concerns a difference between in-depth single case studies (classical case study) to a focus on the comparison of multiple cases. Dyer and Wilkins ( 1991 : 614) claim that the essence of a case study lies in the careful study of a single case to identify new relationships and, as a result, question the Eisenhardt approach which puts a lot of emphasis on comparison of multiple cases. Eisenhardt, on the contrary, claims that multiple cases allow replication between cases and is, therefore, seen as a means of corroboration of propositions (Eisenhardt 1991 ). Classical case studies prefer deep descriptions of a single case, considering the context to reveal insights into the single case and by that elaborate new theories. The comparison of multiple cases, therefore, tends—in the opinion of Dyer and Wilkens—to surface descriptions. This weakens the possibility of context-related, rich descriptions. While, in classic case study, good stories are the aim, the development of good constructs and their relationships is aimed in Eisenhardt’s approach. Eisenhardt ( 1991 : 627) makes a strong plea on more methodological rigor in case study research, while Dyer and Wilkins ( 1991 : 613) criticize that the new approach “… includes many of the attributes of hypothesis-testing research (e.g., sampling and controls).”

Dooley ( 2002 : 346) briefly takes the case study research designs by Yin (1994) and Eisenhardt ( 1989 ) as exemplars of how the processes of case study research can be applied. The approach by Eisenhardt is seen as an exemplar that advances conceptualization and operationalization in the phases of theory-building, while the approach by Yin is seen as exemplar that advances minimally conceptualized and operationalized existing theory.

Baxter and Jack ( 2008 ) describe the designs by Yin (2003) and Stake ( 1995 ) to demonstrate key elements of qualitative case study. The authors outline and carefully compare the approaches by Yin and Stake in conducting the research process, neglecting philosophical differences and theoretical goals.

Piekkari et al. ( 2009 ) outline the methodological richness of case study research using the approaches of Yin et al. (1998), and Stake. They specifically exhibit the role of philosophical assumptions, establishing differences in conventionally accepted practices of case study research in published papers. The authors analyze 135 published case studies in four international business journals. The analysis reveals that, in contrast to the richness of case study approaches, the majority of published case studies draw on positivistic foundations and are narrowly declared as explorative with a lack of clarity of the theoretical purpose of the case study. Case studies are often designed as multiple case studies with cross-sectional designs based on interviews. In addition to the narrow use of case study research, the authors find out that “… most commonly cited methodological literature is not consistently followed” (Piekkari et al. 2009 : 567).

Welch et al. ( 2011 ) develop a typology of theorizing modes in case study methods. Based on the two dimensions “contextualization” and “causal explanation”, they differentiate in their typology between inductive theory-building (Eisenhardt), interpretive sensemaking (Stake), natural experiment (Yin), and contextualised explanation (Ragin/Bhaskar). The typology is used to analyze 199 case studies from three highly ranked journals over a 10-year period for whether the theorizing modes are exercised in the practice of publishing case studies. As a result, the authors identify a strong emphasis on the exploratory function of case studies, neglecting the richness of case study methods to challenge, refine, verify, and test theories (Welch et al. 2011 : 755). In addition, case study methods are not consistently related to theory contribution: “By scrutinising the linguistic elements of texts, we found that case researchers were not always clear and consistent in the way that they wrote up their theorising purpose and process” (Welch et al. 2011 : 756).

As a result, the comparisons reveal a range of case study designs which are rarely discussed. In contrast, published case studies are mainly introduced as exploratory design. Explanatory, interpretivist, and critical/reflexive designs are widely neglected, narrowing the possible applications of case study research. In addition, comparisons containing an analysis of published case studies reveal a low degree in accuracy when applying case study methods.

What is missing is a comparison of case study research designs with regard to differences in the contribution to theory. Case study designs have different purposes in theory contribution. Confusing these potential contributions by inconsistently utilizing the appropriate methods weakens the contribution of case studies to scientific progress and, by that, damages the reputation of case studies.

To conduct such a comparison, I consider the four case study research approaches of Yin, Eisenhardt, Burawoy, and Stake for the following reasons.

These approaches are the main representatives of case study research design outlined in the comparisons elaborated above (Baxter and Jack 2008 ; Dooley 2002 ; Dyer and Wilkins 1991 ; Piekkari et al. 2009 ; Welch et al. 2011 ). I follow especially the argument by Piekkari et al. ( 2009 ) that these approaches contain a broad spectrum of methodological foundations of exploratory, explanatory, interpretivist, and critical/reflexive designs. The chosen approaches have an explicit and detailed methodology which can be reconstructed and compared with regard to their theory contribution. Although there are variations in the application of the designs, to the best of my knowledge, the designs represent the spectrum of case study methodologies. A comparison of these methodologies revealed main distinguishable differences. To highlight these main differences, I summarized these differences into labels of “no theory first”; “gaps and holes”; “social construction of reality”; and “anomalies”.

I did not consider descriptions of case study research in text books which focus more or less on general descriptions of the common characteristics of case studies, but do not emphasize differences in methodologies and theory contribution. In addition, I did not consider so-called “home grown” designs (Eisenhardt 1989 : 534) which lack a systematic and explicit demonstration of the methodology and where “… the hermeneutic process of inference—how all these interviews, archival records, and notes were assembled into a coherent whole, what was counted and what was discounted—remains usually hidden from the reader” (Fiss 2009 : 425).

Finally, although often cited in the methodological section of case studies, books are not considered which concentrate on data analysis in qualitative research per se (Miles et al. 2014 ; Corbin and Strauss 2015 ). Therefore, to analyze the contribution of case study research to the scientific development, it needs to compare explicit methodology. This comparison will be outlined in the following sections with regard to main methodological steps: the role of the case, the collection of data, and the analysis of data.

2.1 Case study research design 1: no theory first

A popular template for building theory from case studies is a paper by Eisenhardt ( 1989 ). It follows a dramaturgy with a precise order of single steps for constructing a case study and is one of the most cited papers in methods sections (Ravenswood 2011 ). This is impressive for two reasons. On the one hand, Eisenhardt herself has provided a broader spectrum of case study research designs in her own empirical papers, for example, by combining theory-building and theory elaboration (Bingham and Eisenhardt 2011 ). On the other hand, she “updated” her design in a paper with Graebner (Eisenhardt and Graebner 2007 ), particularly by extending the range of inductive theory-building. These developments do not seem to be seriously considered by most authors, as differences and elaborations of this spectrum are rarely found in publications. Therefore, in the following, I focus on the standards provided by Eisenhardt ( 1989 ) and Eisenhardt and Graebner ( 2007 ) as exemplary guidelines.

Eisenhardt follows the ideal of ‘no theory first’ to capture the richness of observations without being limited by a theory. The research question may stem from a research gap meaning that the research question is of relevance. Tentative a priori constructs or variables guide the investigation, but no relationships between such constructs or variables are assumed so far: “Thus, investigators should formulate a research problem and possibly specify some potentially important variables, with some reference to extant literature. However, they should avoid thinking about specific relationships between variables and theories as much as possible, especially at the outset of the process” (Eisenhardt 1989 : 536).

Cases are chosen for theoretical reasons: for the likelihood that the cases offer insights into the phenomenon of interest. Theoretical sampling is deemed appropriate for illuminating and extending constructs and identifying relationships for the phenomenon under investigation (Eisenhardt and Graebner 2007 ). Cases are sampled if they provide an unusual phenomenon, replicate findings from other cases, use contrary replication, and eliminate alternative explanations.

With respect to data collection, qualitative data are the primary choice. Data collection is based on triangulation, where interviews, documents, and observations are often combined. A combination of qualitative data and quantitative data is possible as well (Eisenhardt 1989 : 538). Data analysis is conducted via the search for within-case patterns and cross-case patterns. Systematic procedures are conducted to compare the emerging constructs and relationships with the data, eventually leading to new theory.

A good exemplar for this design is the investigation of technology collaborations (Davis and Eisenhardt 2011 ). The purpose of this paper is to understand processes by which technology collaborations support innovations. Eight technology collaborations among ten firms were sampled for theoretical reasons. Qualitative and quantitative data were used from semi-structured interviews, public and private data, materials provided by informants, corporate intranets, and business publications. The data was measured, coded, and triangulated. Writing case histories was a basis for within-case and cross-case analysis. Iteration between cases and emerging theory and considering the relevant literature provided the basis for the development of a theoretical framework.

Another example is the investigation of what is learned in organizational processes (Bingham and Eisenhardt 2011 ). This paper demonstrates that the case study design is not only used for theory-building, but can also be combined with theory elaboration. Based on the lenses of the organizational knowledge literature, organizational routines literature, and heuristics literature, six technology-based ventures were chosen for theoretical reasons. Several data sources were used, especially quantitative and qualitative data from semi-structured interviews, archival data, observations, e-mails, phone calls, and follow-up interviews. Within-case analysis revealed what each firm has learned from process experience. Cross-case analysis revealed emerging patterns from which tentative constructs and propositions were formed. In replication logic constructs and propositions were refined across the cases. When mirroring the findings with the literature, both the emergences of the constructs were compared and unexpected types were considered. The iteration of theory and data as well as the consideration of related research sharpened the theoretical arguments, eventually leading to a theoretical framework. “Thus, we combined theory elaboration (Lee 1999 ) and theory generation (Eisenhardt 1989 )” (Bingham and Eisenhardt 2011 : 1448).

2.2 Case study research design 2: gaps and holes

Contrary to “No Theory First”, case study research design can also aim at specifying gaps or holes in existing theory with the ultimate goal of advancing theoretical explanations (Ridder 2016 ). A well-known template for this case study research design is the book by Yin ( 2014 ). It is a method-orientated handbook of how to design single and multiple case studies with regard to this purpose. Such a case study research design includes: “A ‘how’ and ‘why’ question” (Yin 2014 : 14). Research questions can be identified and shaped using literature to narrow the interest in a specific topic, looking for key studies and identifying questions in these studies. According to Yin’s design, existing theory is the starting point of case study research. In addition, propositions or frameworks provide direction, reflect the theoretical perspective, and guide the search for relevant evidence.

There are different rationales for choosing a single case design (Yin 2014 : 51). Purposeful sampling is conducted if an extreme case or an unusual case is chosen and if rarely observable phenomena can be investigated with regard to unknown matters and their relationships. Common cases allow conclusions for a broader class of cases. Revelatory cases provide the opportunity to investigate into a previously inaccessible inquiry, and the longitudinal study enables one to investigate a single case at several points in time. A rationale for multiple case designs has its strength in replication logic (Yin 2014 : 56). In the case of literal replication, cases are selected to predict similar results. In the case of theoretical replication, cases are selected to predict contrasting results but for theoretical reasons. Yin provides several tactics to increase the reliability (protocol; data base) of the study.

Yin ( 2014 : 103) emphasizes that interviews are one of the most important sources of data collection but considers other sources of qualitative data as well. Data triangulation is designed to narrow problems of construct validity, as multiple sources of data provide multiple measures of the same phenomenon. Yin ( 2014 : 133) offers a number of data analysis strategies (e.g., case description; examining rival explanations) and analytic techniques which are apt to compare the proposed relationships with empirical patterns. Pattern-matching logic compares empirically based patterns with predicted patterns, enabling further data analysis techniques (explanation building, time series analysis, logic models, and cross-case synthesis). In analytical generalization, the theory is compared with the empirical results, leading to the modification or extension of the theory.

An appropriate model for this case study design can be identified in a paper by Ellonen et al. ( 2009 ). The paper is based on the emerging dynamic capability theory. The four cases were chosen for theoretical reasons to deliver an empirical contribution to the dynamic capability theory by investigating the relationship of dynamic capabilities and innovation outcomes. The authors followed a literal replication strategy and identified patterns between dynamic capabilities of the firms and their innovation outcomes.

Shane ( 2000 ) is an author who developed specific propositions from a framework and examined the propositions in eight entrepreneurial cases. Using several sources of interviews and archival data, the author compared the data with the propositions using the pattern-matching logic, which concluded in developing entrepreneurship theory.

2.3 Case study research design 3: social construction of reality

So far, the outlined case study research designs are based on positivist roots, but there is richness and variety in case study research stemming from different philosophical realms. The case study research design by Stake ( 1995 , 2000 , 2005 ), for example, is based on constructivist assumptions and aims to investigate the social construction of reality and meaning (Schwandt 1994 : 125).

According to this philosophical assumption, there is no unique “real world” that preexists independently of human mental activity and symbolic language. The world is a product of socially and historically related interchanges amongst people (social construction). The access to reality is given through social constructions, such as language and shared meanings: “The meaning-making activities themselves are of central interest to social constructionists/constructivists, simply because it is the meaning-making/sense making attributional activities that shape action or (inaction)” (Guba and Lincoln 2005 : 197). Therefore, the researcher is not looking for objective “facts”, nor does he aim at identifying and measuring patterns which can be generalized. Contrarily, the constructivist is researching into specific actions, in specific places, at specific times. The scientist tries to understand the construction and the sharing of meaning (Schwandt 1994 ).

According to Stake ( 2005 ), the direction of the case study is shaped by the interest in the case. In an intrinsic case study, the case itself is of interest. The purpose is not theory-building but curiosity in the case itself. In an instrumental case study, the case itself is of secondary interest. It plays a supportive role, as it facilitates the understanding of a research issue. The case can be typical of other cases. Multiple or collective case study research designs extend the instrumental case study. It is assumed that a number of cases will increase the understanding and support theorizing by comparison of the cases.

The differentiation by Stake ( 1995 , 2005 ) into intrinsic and instrumental cases guides the purposive sampling strategy. In intrinsic case studies, the case is, by definition, already selected. The researcher looks for specific characteristics, aiming for thick descriptions with the opportunity to learn. Representativeness or generalization is not considered. In instrumental case study design, purposive sampling leads to the phenomenon under investigation. In multiple case study designs, the ability to compare cases enhances the opportunity to theorize.

A case study requires an integrated, holistic comprehension of the case complexity. According to Stake ( 2005 ), the case study is constructed by qualitative data, such as observations, interviews, and documents. Triangulation first serves as clarification of meaning. Second, the researcher is interested in the diversity of perceptions.

Two methods of data analysis are considered in such qualitative case study design: direct interpretation and categorical aggregation (Stake 1995 : 74). The primary task of an intrinsic case study is to understand the case. This interpretation is offered to the reader, but the researcher has to provide the material in a sufficient way (thick descriptions), so that the reader can learn from the case as well as draw his or her own conclusions. Readers can thus make some generalizations based on personal and vicarious experiences (“naturalistic generalization”). In instrumental case studies, the understanding of phenomena and relationships leads to categorical aggregation, and the focus is on how the phenomenon exists across several cases.

Greenwood and Suddaby ( 2006 ), for example, used the instrumental case study design by Stake, combining network location theory and dialectical theory. They identified new dynamics creating a process model of elite institutional entrepreneurship.

Ituma et al. ( 2011 ) highlighted the social construction of reality in their study of career success. The majority of career studies have been conducted in Western countries and findings have been acknowledged as universally applicable. The authors demonstrated that realities of managers in other areas are constructed differently. As a result of their study, they provided a contextually sensitive frame for the analysis of career outcomes.

2.4 Case study research design 4: anomalies

Identifying anomalies as a basis for further research is common in management and organization research (Gilbert and Christensen 2005 ). In case study research, the extended case study method is used for this case study research design (Ridder 2016 ). Following Burawoy ( 1991 , 1998 , 2009 ), the research question derives from curiosity. Researchers normally look at what is “interesting” and what is “surprising” in a social situation that existing theory cannot explain. Initially, it is not important whether the expectations develop from some popular belief, stereotype, or from an academic theory. The extended case study research design is guided by anomalies that the previous theory was not able to explain through internal contradictions of theory, theoretical gaps, or silences. An anomaly does not reject theory, but rather demonstrates that the theory is incomplete. Theory is aimed to be improved by “… turning anomalies into exemplars” (Burawoy 1991 : 10).

The theoretical sampling strategy in this case study research design stems from the theoretical failure in confrontation with the site. According to the reflexive design, such cases do not favour individuals or isolated phenomena, but social situations in which a comparative strategy allows the tracing of differences across the cases to external forces.

In the extended case study, the researcher deals with qualitative data, but also considers the broader complex social situation. The researcher engages into a dialogue with the respondents (Burawoy ( 1991 , 1998 , 2009 ). An interview is an intervention into the life of a respondent. By means of mutual interaction it is possible to discover the social order under investigation. The observer has to unpack those situational experiences by means of participant observation and mutual interpretation. This situational comprehension aims at understanding divergent “voices”, reflecting the variety of respondents’ understandings of the social situation.

As in other sciences, these voices have to be aggregated. This aggregation of multiple readings of a single case is conducted by turning the aggregation into social processes: “The move from situation to process is accomplished differently in different reflexive methods, but it is always reliant on existing theory” (Burawoy 2009 : 41). Social processes are now traced to the external field as the conditions of the social processes. Consequently, this leads to the question concerning “… how those micro situations are shaped by wider structures” (Burawoy 1991 : 282). “Reflexive science insists, therefore, on studying the everyday world from the standpoint of its structuration, that is, by regarding it as simultaneously shaped by and shaping an external field of forces” (Burawoy 2009 : 42). Such social fields cannot be held constant, which undermines the idea of replication. The external field is in continuous flux. Accordingly, social forces that influence the social processes are identified, shaping the phenomenon under investigation. Extension of theory does not target representativeness as a relationship of sample and population. Generality in reflexive science is to reconstruct an existing theory: “We begin with our favorite theory but seek not confirmations but refutations that inspire us to deepen that theory. Instead of discovering grounded theory, we elaborate existing theory. We do not worry about the uniqueness of our case, since we are not as interested in its representativeness as its contribution to reconstructing theory. Our theoretical point of departure can range from the folk theory of participants to any abstract law. We consider only that the scientist consider it worth developing” (Burawoy 2009 : 43). Such elaboration stems from the identification of anomalies and offers new predictions with regard to the theory.

It is somewhat surprising that the extended case study design has been neglected in the management literature so far, and it appears that critical reflexive principles have to be resurrected as they have been in other disciplines (see the overview at Wadham and Warren 2014 ). Examples in the management and organization literature are rare. Danneels ( 2011 ) used the extended case study design to extend the dynamic capabilities theory. In his famous Smith Corona case, Danneels shows how a company tried to change its resource base. Based on detailed data, the Smith Corona case provides insights into the resource alteration processes and how dynamic capabilities operate. As a result, the paper fills a process gap in dynamic capability theory. Iterating between data collection and analysis, Danneels revealed resource cognition as an element not considered so far in dynamic capability theory. The use of the extended case study method is limited to the iteration of data and theory. First, there is “running exchange” (Burawoy 1991 : 10) between field notes and analysis. Second, there is iteration between analysis and existing theory. Unlike Burawoy, who aims to reconstruct existing theory on the basis of “emergent anomalies” (Burawoy 1991 : 11) considering social processes and external forces, Danneels confronts the dynamic capabilities literature with the Smith Corona case to extend the theory of dynamic capabilities.

2.5 A comparison of case study research processes

Commonalities and differences emerged from the comparison of the designs. Table  1 provides a brief summary of these main differences and the resulting portfolio of case study research designs which will be discussed in more detail.

There is an extensive range between the different designs regarding the research processes. In “no theory first”, there is a broad and tentative research question with some preliminary variables at the outset. The research question may be modified during the study as well as the variables. This design avoids any propositions regarding relationships.

On the contrary, the research question in “gaps and holes” is strongly related to existing theory, focusing on “how and why” questions. The existing theory contains research gaps which, once identified within the existing theory, lead accordingly to assumed relationships which are the basis for framework and propositions to be matched by empirical data. This broad difference is even more elaborated by a design that aims the “social construction of reality”. There is no research question at the outset, but a curiosity in the case or the case is a facilitator to understand a research issue. This is far away from curiosity in the “anomaly approach”. Here, the research question is inspired by questioning why an anomaly cannot be explained by the existing theory. What kind of gaps, silences, or internal contradictions demonstrates the insufficiency of the existing theory?

Various sampling strategies are used across these case study research designs, including theoretical sampling and purposeful sampling, which serve different objectives. Theoretical sampling in “no theory first” aims at selecting a case or cases that are appropriate to highlight new or extend preliminary constructs and reveal new relationships. There is a distinct difference from theoretical sampling in the “anomalies” approach. Such a sampling strategy aims to choose a case that is a demonstration of the failure of the theory. In “gaps and holes” sampling is highly focused on the purpose of the case study. Extreme and unusual cases have other purposes compared to common cases or revelatory cases. A single case may be chosen to investigate deeply into new phenomena. A multiple case study may serve a replication logic by which the findings have relevance beyond the cases under investigation. In “social construction of reality”, the sampling is purposeful as well, but for different reasons. Either the case is of interest per se or the case represents a good opportunity to understand a theoretical issue.

Although qualitative data are preferred in all of the designs, quantitative data are seen as a possible opportunity to strengthen cases by such data. Nevertheless, in “social construction of reality”, there is a strong emphasis on thick descriptions and a holistic understanding of the case. This is in contrast to a more construct- and variable- oriented collection of data in “no theory first” and “gaps and holes”. In addition, in contrast to that, the “anomaly” approach is the only design that receives data from dialogue between observer and participants and participant observation.

Finally, data analysis lies within a wide range. In “no theory first”, the research process is finalized by inspecting the emerging constructs within the case or across cases. Based on a priory constructs, systematic comparisons reveal patterns and relationships resulting in a tentative theory. On the contrary, in “gaps and holes”, a tentative theory exists. The final analysis concentrates on the matching of the framework or propositions with patterns from the data. While both of these approaches condense data, the approach of “social construction of reality” ends the research process with thick descriptions of the case to learn from the case or with categorical comparisons. In the “anomaly” approach, the data analysis is aggregation of data, but these aggregated data are related to its external field and their pressures and influences by structuration to reconstruct the theory.

As a result, it is unlikely that the specified case study designs contribute to theory in a homogeneous manner. This result will be discussed in light of the question regarding how these case study designs can inform theory at several points of a continuum of theory. This analysis will be outlined in the following sections. In a first step, I review the main elements of a theory continuum. In a second step, I discuss the respective contribution of the previously identified case study research designs to the theory continuum.

3 Elements of a theory continuum

What a theory is and what a theory is not is a classic debate (Sutton and Staw 1995 ; Weick 1995 ). Often, theories are described in terms of understanding relationships between phenomena which have not been or were not well understood before (Chiles 2003 ; Edmondson and McManus 2007 ; Shah and Corley 2006 ), but there is no overall acceptance as to what constitutes a theory. Theory can be seen as a final product or as a continuum, and there is an ongoing effort to define different stages of this continuum (Andersen and Kragh 2010 ; Colquitt and Zapata-Phelan 2007 ; Edmondson and McManus 2007 ; Snow 2004 ; Swedberg 2012 ). In the following section, basic elements of the theory and the construction of the theory continuum are outlined.

3.1 Basic elements of a theory

Most of the debate concerning what a theory is comprises three basic elements (Alvesson and Kärreman 2007 ; Bacharach 1989 ; Dubin 1978 ; Kaplan 1998 ; Suddaby 2010 ; Weick 1989 , 1995 ; Whetten 1989 ). A theory comprises components (concepts and constructs), used to identify the necessary elements of the phenomenon under investigation. The second is relationships between components (concepts and constructs), explaining the how and whys underlying the relationship. Third, temporal and contextual boundaries limit the generalizability of the theory. As a result, definitions of theory emphasize these components, relationships, and boundaries:

“It is a collection of assertions, both verbal and symbolic, that identifies what variables are important for what reasons, specifies how they are interrelated and why, and identifies the conditions under which they should be related or not related” (Campbell 1990 : 65).

“… a system of constructs and variables in which the constructs are related to each other by propositions and the variables are related to each other by hypotheses” (Bacharach 1989 : 498).

“Theory is about the connections among phenomena, a story about why acts, events, structure, and thoughts occur. Theory emphasizes the nature of causal relationships, identifying what comes first as well as the timing of such events” (Sutton and Staw 1995 : 378).

“… theory is a statement of concepts and their interrelationships that shows how and/or why a phenomenon occurs” (Corley and Gioia 2011 : 12).

The terms “constructs” and “concepts” are either used interchangeably or with different meanings. Positivists use “constructs” as a lens for the observation of a phenomenon (Suddaby 2010 ). Such constructs have to be operationalized and measured. Non-positivists often use the term “concept” as a more value neutral term in place of the term construct (Gioia et al. 2013 ; Suddaby 2010 : 354). Non-positivists aim at developing concepts on the basis of data that contain richness and complexity of the observed phenomenon instead of narrow definitions and operationalizations of constructs. Gioia et al. ( 2013 : 16) clarify the demarcation line between constructs and concepts as follows: “By ‘concept,’ we mean a more general, less well-specified notion capturing qualities that describe or explain a phenomenon of theoretical interest. Put simply, in our way of thinking, concepts are precursors to constructs in making sense of organizational worlds—whether as practitioners living in those worlds, researchers trying to investigate them, or theorists working to model them”.

In sum, theories are a systematic combination of components and their relationships within boundaries. The use of the terms constructs and concepts is related to different philosophical assumptions reflected in different types of case study designs.

3.2 Theory continuum

Weick ( 1995 ) makes an important point that theory is more a continuum than a product. In his view, theorizing is a process containing assumptions, accepted principles, and rules of procedures to explain or predict the behavior of a specified set of phenomena. In similar vein, Gilbert and Christensen ( 2005 ) demonstrate the process character of theory. In their view, a first step of theory building is a careful description of the phenomena. Having already observed and described the phenomena, researchers then classify the phenomena into similar categories. In this phase a framework defines categories and relationships amongst phenomena. In the third phase, researchers build theories to understand (causal) relationships, and in this phase, a model or theory asserts what factors drive the phenomena and under what circumstances. The categorization scheme enables the researchers to predict what they will observe. The “test” offers a confirmation under which circumstances the theory is useful. The early drafts of a theory may be vague in terms of the number and adequateness of factors and their relationships. At the end of the continuum, there may be more precise variables and predicted relationships. These theories have to be extended by boundaries considering time and space.

Across that continuum, different research strategies have various strengths. Several classifications in the literature intend to match research strategies to the different phases of a theory continuum (Andersen and Kragh 2010 ; Colquitt and Zapata-Phelan 2007 ; Edmondson and McManus 2007 ; Snow 2004 ; Swedberg 2012 ). These classifications, although there are differences in terms, comprise three phases with distinguishable characteristics.

3.2.1 Building theory

Here, the careful description of the phenomena is the starting point of theorizing. For example, Snow ( 2004 ) puts this phase as theory discovery, where analytic understandings are generated by means of detailed examination of data. Edmondson and McManus ( 2007 ) state the starting phase of a theory as nascent theory providing answers to new questions revealing new connections among phenomena. Therefore, research questions are open and researchers avoid hypotheses predicting relationships between variables. Swedberg ( 2012 ) highlights the necessity of observation and extensive involvement with the phenomenon at the early stage of theory-building. It is an attempt to understand something of interest by observing and interpreting social facts. Creativity and inspiration are necessary conditions to put observations into concepts and outline a tentative theory.

3.2.2 Developing theory

This tentative theory exists in the second phase of the continuum and has to be developed. Several possibilities exist. In theory extension, the preexisting constructs are extended to other groups or other contexts. In theoretical refinement, a modification of existing theoretical perspectives is conducted (Edmondson and McManus ( 2007 ). New antecedents, moderators, mediators, and outcomes are investigated, enhancing the explanation power of the tentative theory.

3.2.3 Test of theories

Constructs and relationships are well developed to a mature state; measures are precise and operationalized. Such theories are empirically tested with elaborate methods, and research questions are more precise. In the quantitative realm, testing of hypotheses is conducted and statistical analysis is the usual methodological foundation. Recently, researchers criticize that testing theories has become the major focus of scientists today (Delbridge and Fiss 2013 ); testing theories does not only happen to mature theory but to intermediate theory as well. The boundary between theory development and theory testing is not always so clear. While theory development is adding new components to a theory and elaborating the measures, testing a theory implies precise measures, variables, and predicted relationships considering time and space (Gilbert and Christensen ( 2005 ). It will be of interest whether case studies are eligible to test theories as well.

To summarize: there is a conversation as to where on a continuum of theory development, various methods are required to target different contributions to theory (methodological fit). In this discussion, case study research designs have been discussed as a homogeneous set that mostly contributes to theory-building in an exploratory manner. Hence, what is missing is a more differentiated analysis of how case study methodology fits into this conversation, particularly how case study research methodologically fits theory development and theory testing beyond its widely assumed explorative role. In the following section, the above types of case study research designs will be discussed with regard to their positions across the theory continuum.

This distinction adds to existing literature by demonstrating that case study research does not only contribute to theory-building, but also to the development of tentative theories and to the testing of theories. This distinction leads to the next question: is there any interplay between case study research designs and their contributions to the theory continuum? This paper aims at reconciling this interplay with regard to case study design by mirroring phases of a theory continuum with specific types of case study research designs as outlined above. The importance of the interplay between theory and method lies in the capacity to generate and shape theory, while theory can generate and shape method. “In this long march, theory and method surely matter, for they are the tools with which we build both our representations and understandings of organizational life and our reputations” (van Maanen et al. 2007 : 1145). Theory is not the same as methods, but a relationship of this interplay can broaden or restrict both parts of the equation (Swedberg 2012 : 7).

In the following, I discuss how the above-delineated case study research designs unfold their capacities and contribute differently to the theory continuum to build, develop, and test theory.

4 Discussion of the contribution of case study research to a theory continuum

Case study research is diverse with distinct contributions to the continuum of theory. The following table provides the main differences in terms of contributions to theory and specifically locates the case study research designs on the theory continuum (Table  2 ).

In the following, I outline how these specific contributions of case study designs provide better opportunities to enhance the rigor of building theory, developing theory, testing, and reconstructing theory.

4.1 Building theory

In building theory, the phenomenon is new or not understood so far. There is no theory which explains the phenomenon. At the very beginning of the theory continuum, there is curiosity in the phenomenon itself. I focus on the intrinsic case study design which is located in the social construction of reality approach on the very early phase of the theory continuum, as intrinsic case study research design is not theory-building per se but curiosity in the case itself. It is not the purpose of the intrinsic case study to identify abstract concepts and relationships; the specific research strategy lies in the observation and description of a case and the primary method is observation, enabling understanding from personal and vicarious experience. This meets long lasting complaints concerning the lack of (new) theory in management and organization research and signals that the gap between research and management practice is growing. It is argued that the complexity of the reality is not adequately captured (Suddaby et al. 2011 ). It is claimed that management and organization research systematically neglect the dialogue with practice and, as a result, miss new trends or recognize important trends with delay (Corley and Gioia 2011 ).

The specific case study research design’s contribution to theory is in building concrete, context-dependent knowledge with regard to the identification of new phenomena and trends. Openness with regard to the new phenomena, avoiding theoretical preconceptions but building insights out of data, enables the elaboration of meanings and the construction of realities in intrinsic case studies. Intrinsic case studies will enhance the understanding by researcher and reader concerning new phenomena.

The “No Theory First” case study research design is a classic and often cited candidate for building theory. As the phenomenon is new and in the absence of a theory, qualitative data are inspected for aggregation and interpretation. In instrumental case study design, a number of cases will increase the understanding and support building theories by description, aggregation, and interpretation (Stake 2000 ). New themes and concepts are revealed by case descriptions, interviews, documents, and observations, and the analysis of the data enables the specific contribution of the case study design through a constructivist perspective in theory-building.

Although the design by Eisenhardt ( 1989 ) stems from other philosophical assumptions and there are variations and developments in this design, there is still an overwhelming tendency to quote and to stick to her research strategy which aims developing new constructs and new relationships out of real-life cases. Data are collected mainly by interviews, documents, and observations. From within-site analysis and cross-case analysis, themes, concepts, and relationships emerge. Shaping hypotheses comprises: “… refining the definition of the construct and (…) building evidence which measures the construct in each case” (Eisenhardt 1989 : 541). Having identified the emerged constructs, the emergent relationships between constructs are verified in each case. The underlying logic is validation by replication. Cases are treated as experiments in which the hypotheses are replicated case by case. In replication logic cases that confirm the emergent relationships enhance confidence in the validity of the relationships. Disconfirmation of the relationships leads to refinement of the theory. This is similar to Yin’s replication logic, but targets the precision and measurement of constructs and the emerging relationships with regard to the emerging theory. The building of a theory concludes in an understanding of the dynamics underlying the relationship; the primary theoretical reasons for why the relationships exist (Huy 2012 ). Finally, a visual theory with “boxes and arrows” (Eisenhardt and Graebner 2007 ) may visually demonstrate the emerged theory. The theory-building process is finalized by iterating case data, emerging theory, and extant literature.

The “No Theory First” and “Social Construction of Reality” case study research designs, although they represent different philosophical assumptions, adequately fit the theory-building phase concerning new phenomena. The main contribution of case study designs in this phase of the theory continuum lies in the generation of tentative theories.

Case studies at this point of the theory continuum, therefore, have to demonstrate: why the phenomenon is new or of interest; that no previous theory that explains the phenomenon exists; how and why detailed descriptions enhance the understanding of the phenomenon; and how and why new concepts (constructs) and new relationships will enhance our understanding of the phenomenon.

As a result, it has to be demonstrated that the research strategy is in sync with an investigation of a new phenomenon, building a tentative theory.

4.2 Developing theory

In the “Gaps and Holes” case study research design, the phenomenon is partially understood. There is a tentative theory and the research strategy is theory driven. Compared to the theory-building phase, the existence and not the development of propositions differentiate this design along the continuum. The prediction comes first, out of an existing theory. The research strategy and the data have to be confronted by pattern-matching. Pattern-matching is a means to compare the theoretically based predictions with the data in the site: “For case study analysis, one of the most preferred techniques is to use a pattern-matching logic. Such a logic (…) compares an empirically based pattern–that is, one based on the findings from your case study–with a predicted one made before you collected your data (….)” (Yin 2014 : 143). The comparison of propositions and the rich case material is the ground for new elements or relationships within the tentative theory.

Such findings aim to enhance the scientific usefulness of the theory (Corley and Gioia 2011 ). To enhance the validity of the new elements or relationships of the tentative theory, literal replication is a means to confirm the new findings. By that, the theory is developed by new antecedents, moderators, mediators, or outcomes. This modification or extension of the theory contributes to the analytical generalization of the theory.

If new cases provide similar results, the search for regularities is based on more solid ground. Therefore, the strength of case study research in “Gaps and Holes” lies in search for mechanisms in their specific context which can reveal causes and effects more precisely.

The “Gaps and Holes” case study research design is an adequate candidate for this phase of the theory continuum. Case studies at this point of the theory continuum, therefore, have to outline the tentative theory; to demonstrate the lacks and gaps of the tentative theory; to specify how and why the tentative theory is aimed to be extended and/or modified; to develop theoretically based propositions which guide the investigation; and to evaluate new elements, relationships, and mechanisms related to the previous theory (analytical generalization).

As a result and compared to theory-building, a different research strategy exists. While in theory building the research strategy is based on the eliciting of concepts (constructs) and relationships out of data, in theory development, it has to be demonstrated that the research strategy aims to identify new elements and relationships within a tentative theory, identifying mechanisms which explain the phenomenon more precisely.

4.3 Test of theory

In “Gaps and Holes” and “Anomalies”, an extended theory exists. The phenomenon is understood. There is no search for additional components or relationships. Mechanisms seem to explain the functioning or processes of the phenomenon. The research strategy is focused on testing whether the theory holds under different circumstances or under different conditions. Such a test of theories is mainly the domain of experimental and quantitative studies. It is based on previously developed constructs and variables which are the foundation for stating specific testable hypotheses and testing the relations on the basis of quantitative data sets. As a result, highly sophisticated statistical tools enable falsification of the theory. Therefore, testing theory in “Gaps and Holes” is restricted on specific events.

Single case can serve as a test. There is a debate in case study research whether the test of theories is related to the falsification logic of Karl Popper (Flyvbjerg 2006 ; Tsang 2013 ). Another stream of the debate is related to theoretical generalizability (Hillebrand et al. 2001 ; Welch et al. 2011 ). More specifically, test in” Gaps and Holes” is analogous to a single experiment if a single case represents a critical case. If the theory has specified a clear set of propositions and defines the exact conditions within which the theory might explain the phenomena under investigation, a single case study, testing the theory, can confirm or challenge the theory. In sum Yin states: “Overall, the single-case design is eminently justifiable under certain conditions—where the case represents (a) a critical test of existing theory, …” (Yin 2014 : 56). In their survey in the field of International Business, Welch et al. conclude: “In addition, the widespread assumption that the role of the case study lies only in the exploratory, theory-building phase of research downplays its potential to propose causal mechanisms and linkages, and test existing theories” (Welch et al. 2011 : 755).

In multiple case studies, a theoretical replication is a test of theory by comparing the findings with new cases. If a series of cases have revealed pattern-matching between propositions and the data, theoretical replication can be revealed by new waves of cases with contrasting propositions. If the contrasting propositions reveal contrasting results, the findings of the first wave are confirmed. Several possibilities exist to test the initial findings of multiple case studies using different lenses from inside and outside the management realm (Corley and Gioia 2011 ; LePine and Wilcox-King 2010 ; Okhuysen and Bonardi 2011 ; Zahra and Newey 2009 ), but have not become a standard in case study research.

In rival explanations, rival theoretical propositions are developed as a test of the previous theory. This can be distinguished from theoretical replication where contrasting propositions aim to confirm the initial findings. This can, as well, be distinguished from developing theory where rival explanations might develop theory by the elimination of possible influences (interventions, implementations). The rich data enable one to identify internal and external interventions that might be responsible for the findings. Alternative explanations in a new series of cases enable to test, whether a theory “different from the original theory explains the results better (…)” (Yin 2014 : 141).

As a result, it astonishes that theoretical replication and rival explanations, being one of the strengths of case study research, are rarely used. Although the general debate about “lenses” has informed the discussion about theory contributions, this paper demonstrates that there is a wide range of possible integration of vertical or horizontal lenses in case study research design. Case study research designs aiming to test theories have to outline modes of replication and the elimination of rival explanations.

The “anomaly approach” is placed in the final phase of the theory testing, as well. In this approach, a theory exists, but the theory fails to explain anomalies. Burawoy goes a step further. While Yin ( 2014 ) sees a critical case as a test that challenges or contradicts a well formulated theory, in Burawoy’s approach, in contrast to falsification logic (Popper 2002 ), the theory is not rejected but reconstructed. Burawoy relates extended case study design to society and history. Existing theory is challenged by intervention into the social field. Identifying processes of historical roots and social circumstances and considering external forces by structuration lead to the reconstruction of the theory.

It is surprising that this design has been neglected so far in management research. Is there no need to reflect social tensions and distortions in management research? While case study research has, per definition, to investigate phenomena in its natural environment, it is hard to understand why this design has widely been ignored in management and organization research. As a result, testing theory in case study research has to demonstrate that an extended theory exists; a critical case or an anomaly can challenge the theory; theoretical replication and rival explanations will be means to contradict or confirm the theory; and societal circumstances and external forces explain the anomaly.

Compared to theory-building (new concepts/constructs and relationships out of data) and theory development (new elements and relationships within a tentative theory), testing theory challenges extended theory by empirical investigations into failures and anomalies that the current theory cannot explain.

5 Conclusion

Case studies provide a better understanding of phenomena regarding concrete context-dependent knowledge (Andersen and Kragh 2010 ; Flyvbjerg 2006 : 224), but as literature reviews indicate, there is still confusion regarding the adequate utilization of case study methodology (Welch et al. 2011 ). This can be interpreted in a way that authors and even reviewers are not always aware of the methodological fit in case study research. Case study research is mainly narrowed to its “explorative” function, neglecting the scope of possibilities that case study research provides. The claim for more homogeneity of specified rules in case study research misses the important aspect that a method is not a means in itself, but aims at providing improved theories (van Maanen et al. 2007 ). This paper contributes to the fit of case study research designs and the theory continuum regarding the following issues.

5.1 Heterogeneity of case study designs

Although case study research, overall, has similar characteristics, it incorporates various case study research designs that have heterogeneous theoretical goals and use various elements to reach these goals. The analysis revealed that the classical understanding, whereby case study research is adequate for the “exploration” of a theory and quantitative research is adequate for “testing” theory, is oversimplified. Therefore, the theoretical goals of case study research have to be outlined precisely. This study demonstrates that there is variety of case study research designs that have thus far been largely neglected. Case study researchers can utilize the entire spectrum, but have to consider how the phenomenon is related to the theory continuum.

Case study researchers have to demonstrate how they describe new or surprising phenomena, develop new constructs and relationships, add constructs (variables), antecedents, outcomes, moderators, or mediators to a tentative theory, challenge a theory by a critical case, theoretical replication or discarding rival explanations, and reconstruct a theory by tracking failures and anomalies to external circumstances.

5.2 Methodological fit

The rigor of the case study can be enhanced by considering the specific contribution of various case study research designs in each phase of the theory continuum. This paper provides a portfolio of case study research designs that enables researchers and reviewers to evaluate whether the case study arsenal has been adequately located:

At an early phase of the theory continuum, case studies have their strengths in rich descriptions and investigations into new or surprising empirical phenomena and trends. Researchers and readers can benefit from such rich descriptions in understanding and analyzing these phenomena.

Next, on the theory continuum, there is the well-known contribution of case study research in building tentative theory by eliciting constructs or concepts and their relationships out of data.

Third, development of theories is strongly related to literal replication. Strict comparisons, on the one hand, and controlled theoretical advancement, on the other hand, enable the identification of mechanisms, strengthen the notions of causality, and provide generalizable statements.

Fourth, there are specific circumstances under which case study approaches enable one to test theories. This is to confront the theory with a critical case, to test findings of pattern-matching by theoretical replication and discarding rival explanations. Therefore, “Gaps and Holes” provide the opportunity for developing and testing theories through case study design on the theory continuum.

Finally, testing and contradicting theory are not the final rejection of a theory, but is the basis for reconstructing theory by means of case study design. Anomalies can be traced to historical sources, social processes, and external forces.

This paper demonstrates that the precise interplay of case study research designs and theory contributions on the theory continuum is a prerequisite for the contribution of case study research to better theories. If case study research design is differentiated from qualitative research, the intended contribution to theory is stated and designs that fit the aimed contribution to theory are outlined and substantiated; this will critically enhance the rigor of case study research.

Alvesson, M., and D. Kärreman. 2007. Constructing mystery: Empirical matters in theory development. Academy of Management Review 32: 1265–1281.

Article   Google Scholar  

Andersen, P.H., and H. Kragh. 2010. Sense and sensibility: two approaches for using existing theory in theory-building qualitative research. Industrial Marketing Management 39: 49–55.

Bacharach, S.B. 1989. Organizational theories: some criteria for evaluation. Academy of Management Review 14: 496–515.

Google Scholar  

Baxter, P., and S. Jack. 2008. Qualitative case study methodology: study design and implementation for novice researchers. The Qualitative Report 13: 544–559.

Bingham, C.B., and K.M. Eisenhardt. 2011. Rational heuristics: the ‘simple rules’ that strategists learn from process experience. Strategic Management Journal 32: 1437–1464.

Burawoy, M. 1991. Ethnography unbound . Power and resistance in the modern metropolis: University of California Press.

Burawoy, M. 1998. The extended case method. Sociological Theory 16: 4–33.

Burawoy, M. 2009. The extended case method. Four countries, four decades, four great transformations, and one theoretical tradition . Berkeley: University of California Press.

Burns, R.B. 2000. Introduction to research methods . United States of America: SAGE publications.

Campbell, J.P. 1990. The role of theory in industrial and organizational psychology. In Handbook of industrial and organizational psychology , 2nd ed, ed. M.D. Dunnette, L.M. Hough, and H.C. Triandis, 39–73. Palo Alto: Consulting Psychologists Press.

Chiles, T.H. 2003. Process theorizing: too important to ignore in a kaleidic world. Academy of Management Learning & Education 2: 288–291.

Colquitt, J.A., and C.P. Zapata-Phelan. 2007. Trends in theory building and theory testing: a five-decade study of the Academy of Management Journal. Academy of Management Journal 50: 1281–1303.

Corbin, J.M., & Strauss, A.L. 2015. Basics of qualitative research: Techniques and procedures for developing grounded theory (4th ed). Los Angeles, CA: Sage Publications.

Corley, K.G., and D.A. Gioia. 2011. Building theory about theory building: what constitutes a theoretical contribution? Academy of Management Review 36: 12–32.

Danneels, E. 2011. Trying to become a different type of company: dynamic capability at Smith Corona. Strategic Management Journal 32: 1–31.

Davis, J.P., and K.M. Eisenhardt. 2011. Rotating leadership and collaborative innovation: recombination processes in symbiotic relationships. Administrative Science Quarterly 56: 159–201.

Delbridge, R., and P.C. Fiss. 2013. Editors’ comments: styles of theorizing and the social organization of knowledge. Academy of Management Review 38: 325–331.

Dooley, L.M. 2002. Case study research and theory building. Advances in Developing Human Resources 4: 335–354.

Dubin, R. 1978. Theory building . New York: Free Press.

Dyer, W.G., and A.L. Wilkins. 1991. Better stories, not better constructs, to generate better theory: a rejoinder to Eisenhardt. Academy of Management Review 16: 613–619.

Edmondson, A.C., and S.E. McManus. 2007. Methodological fit in management field research. Academy of Management Review 32: 1155–1179.

Eisenhardt, K.M. 1989. Building theories from case study research. Academy of Management Review 14: 532–550.

Eisenhardt, K.M. 1991. Better stories and better constructs: The case for rigor and comparative logic. Academy of Management Review , 16(3): 620–627.

Eisenhardt, K.M., and M.E. Graebner. 2007. Theory building from cases: opportunities and challenges. Academy of Management Journal 50: 25–32.

Ellonen, H.K., P. Wikström, and A. Jantunen. 2009. Linking dynamic-capability portfolios and innovation outcomes. Technovation 29: 753–762.

Fiss, P.C. 2009. Case studies and the configurational analysis of organizational phenomena. In The SAGE handbook of case-based methods , ed. D.S. Byrne, and C.C. Ragin, 424–440. London/Thousand Oaks: SAGE.

Chapter   Google Scholar  

Flick, U. 2009. An introduction to qualitative research , 4th ed. London: SAGE.

Flyvbjerg, B. (2006). Five misunderstandings about case-study research. Qualitative Inquiry , 12(2): 219–245.

Gilbert, C.G., and C.M. Christensen. 2005. Anomaly-seeking research: thirty years of development in resource allocation theory. In From resource allocation to strategy , ed. J.L. Bower, and C.G. Gilbert, 71–89. Oxford: University Press, Oxford.

Gioia, D.A., K.G. Corley, and A.L. Hamilton. 2013. Seeking qualitative rigor in inductive research: notes on the Gioia methodology. Organizational Research Methods 16: 15–31.

Gomm, R., M. Hammersley, and P. Foster. 2000. Case study method. Key issues, key texts . London/Thousand Oaks: Sage Publications.

Greenwood, R., and R. Suddaby. 2006. Institutional entrepreneurship in mature fields: the big five accounting firms. Academy of Management Journal 49: 27–48.

Guba, E.G., and Y.S. Lincoln. 2005. Paradigmatic controversies, contradictions, and emerging confluences. In The SAGE handbook of qualitative research , 3rd ed, ed. N.K. Denzin, and Y.S. Lincoln, 191–215. London, Thousand Oaks: Sage Publications.

Hillebrand, B., R.A.W. Kok, and W.G. Biemans. 2001. Theory-testing using case studies: a comment on Johnston, Leach, and Liu. Industrial Marketing Management 30: 651–657.

Huy, Q.N. 2012. Improving the odds of publishing inductive qualitative research in Premier Academic Journals. The Journal of Applied Behavioral Science 48: 282–287.

Ituma, A., R. Simpson, F. Ovadje, N. Cornelius, and C. Mordi. 2011. Four ‘domains’ of career success: how managers in Nigeria evaluate career outcomes. The International Journal of Human Resource Management 22: 3638–3660.

Kaplan, A. 1998. The conduct of inquiry. Methodology for behavioral science . New Brunswick, N.J.: Transaction Publishers.

Lee, T.W. 1999. Using qualitative methods in organizational research. Organizational research methods series . Thousand Oaks: Sage Publications.

LePine, J.A., and A. Wilcox-King. 2010. Editors´s comments: developing novel theoretical insight from reviews of existing rheory and research. Academy of Management Review 35: 506–509.

Mason, J. 2002. Qualitative researching , 2nd ed. London, Thousand Oaks: Sage Publications.

Miles, M.B., Huberman, M.A., & Saldana, J. 2014. Qualitative data analysis: A methods sourcebook (3rd ed). Los Angeles, CA: Sage Publications.

Okhuysen, G., and J.P. Bonardi. 2011. The challenges of building theory by combining lenses. Academy of Management Review 36: 6–11.

Piekkari, R., C. Welch, and E. Paavilainen. 2009. The case study as disciplinary convention: evidence from international business journals. Organizational Research Methods 12: 567–589.

Popper, K.R. 2002. Logik der Forschung . Tübingen: Mohr Siebeck.

Ravenswood, K. 2011. Eisenhardt’s impact on theory in case study research. Journal of Business Research 64: 680–686.

Ridder, H.G. 2016. Case study research. Approaches, methods, contribution to theory. Sozialwissenschaftliche Forschungsmethoden , vol. 12. München/Mering: Rainer Hampp Verlag.

Schwandt, T.A. 1994. Constructivist, interpretivist approaches to human inquiry. In Handbook of qualitative research , ed. N.K. Denzin, and Y.S. Lincoln, 118–137. Thousand Oaks: Sage Publications.

Shah, S.K., and K.G. Corley. 2006. Building better theory by bridging the quantitative–qualitative divide. Journal of Management Studies 43: 1821–1835.

Shane, S. 2000. Prior knowledge and the discovery of entrepreneurial opportunities. Organization Science 11: 448–469.

Snow, C.C. 2004. Thoughts on alternative pathways to theoretical development: Theory generation, extension, and refinement. In Workshop on scientific foundations of qualitative research , ed. C.C. Ragin, J. Nagel, and P. White, 133–136. Arlington, VA: National Science Foundation.

Stake, R.E. 1995. The art of case study research . London, Thousand Oaks: Sage Publications.

Stake, R.E. 2000. The case study and generalizability. In Case study method. Key issues, key texts , ed. R. Gomm, M. Hammersley, and P. Foster, 19–26. London/Thousand Oaks: Sage Publications.

Stake, R.E. 2005. Qualitative case studies. In The SAGE handbook of qualitative research , 3rd ed, ed. N.K. Denzin, and Y.S. Lincoln, 443–466. London, Thousand Oaks: Sage Publications.

Suddaby, R. 2010. Editor’s comments: construct clarity in theories of management and organization. Academy of Management Review 35: 346–357.

Suddaby, R., C. Hardy, and Q.N. Huy. 2011. Introduction to special topic forum: where are the new theories of organization? Academy of Management Review 36: 236–246.

Sutton, R.I., and B.M. Staw. 1995. What theory is not. Administrative Science Quarterly 40: 371–384.

Swedberg, R. 2012. Theorizing in sociology and social science: turning to the context of discovery. Theory and Society 41: 1–40.

Tsang, E.W.K. 2013. Generalizing from research findings: the merits of case studies. International Journal of Management Reviews 16: 369–383.

van Maanen, J., J.B. Sørensen, and T.R. Mitchell. 2007. The interplay between theory and method. Academy of Management Review 32: 1145–1154.

Vaughan, D. 1992. Theory elaboration: The heuristics of case analysis. In What is a case? , ed. C.C. Ragin, and H.S. Becker, 173–202. Exploring the foundations of social inquiry: Cambridge University Press, Cambridge, New York.

Wadham, H., and R.C. Warren. 2014. Telling organizational tales the extended case method in practice. Organizational Research Methods 17: 5–22.

Weick, K.E. 1989. Theory construction as disciplined imagination. Academy of Management Review 14: 516–531.

Weick, K.E. 1995. What theory is not, theorizing is. Administrative Science Quarterly 40: 385–390.

Welch, C., R. Piekkari, E. Plakoyiannaki, and E. Paavilainen-Mäntymäki. 2011. Theorising from case studies: towards a pluralist future for international business research. Journal of International Business Studies 42: 740–762.

Whetten, D.A. 1989. What constitutes a theoretical contribution? Academy of Management Review 14: 490–495.

Yin, R.K. 2014. Case study research. Design and methods , 5th ed. London, Thousand Oaks: Sage Publications.

Zahra, S.A., and L.R. Newey. 2009. Maximizing the Impact of Organization Science: theory-Building at the Intersection of Disciplines and/or Fields. Journal of Management Studies 46: 1059–1075.

Download references

Author information

Authors and affiliations.

Institute of Human Resource Management, Leibniz University of Hannover, Königsworther Platz 1, 30167, Hannover, Germany

Hans-Gerd Ridder

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to Hans-Gerd Ridder .

Additional information

I thank two anonymous reviewers for their constructive comments. I am grateful for valuable thoughts generously provided by Ann Kristin Zobel.

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License ( http://creativecommons.org/licenses/by/4.0/ ), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Reprints and permissions

About this article

Ridder, HG. The theory contribution of case study research designs. Bus Res 10 , 281–305 (2017). https://doi.org/10.1007/s40685-017-0045-z

Download citation

Received : 26 January 2016

Accepted : 06 February 2017

Published : 16 February 2017

Issue Date : October 2017

DOI : https://doi.org/10.1007/s40685-017-0045-z

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Case studies
• Research design
• Heterogeneity of case study designs
• Theory continuum
• Methodological fit
• Contribution to theory
• Find a journal
• Publish with us
• Track your research

On the site

Belfer center studies in international security.

• political science

Case Studies and Theory Development in the Social Sciences

by Alexander L. George and Andrew Bennett

ISBN: 9780262572224

Pub date: April 15, 2005

• Publisher: The MIT Press

352 pp. , 6 x 9 in ,

ISBN: 9780262262897

ISBN: 9780262072571

Pub date: April 22, 2005

eTextbook rental

4 months: $26.50, 12 months: $37.10.

• 9780262572224
• Published: April 2005
• 9780262262897
• 9780262072571
• Amazon.co.uk
• Bookshop.org
• Waterstones

Other Retailers:

• MIT Press Bookstore
• Penguin Random House
• Barnes and Noble
• Books a Million
• Description

The use of case studies to build and test theories in political science and the other social sciences has increased in recent years. Many scholars have argued that the social sciences rely too heavily on quantitative research and formal models and have attempted to develop and refine rigorous methods for using case studies. This text presents a comprehensive analysis of research methods using case studies and examines the place of case studies in social science methodology. It argues that case studies, statistical methods, and formal models are complementary rather than competitive.

The book explains how to design case study research that will produce results useful to policymakers and emphasizes the importance of developing policy-relevant theories. It offers three major contributions to case study methodology: an emphasis on the importance of within-case analysis, a detailed discussion of process tracing, and development of the concept of typological theories. Case Studies and Theory Development in the Social Sciences will be particularly useful to graduate students and scholars in social science methodology and the philosophy of science, as well as to those designing new research projects, and will contribute greatly to the broader debate about scientific methods.

Alexander L. George was Graham H. Stuart Professor of Political Science Emeritus at Stanford University and the author or coauthor of many books, most recently Presidential Personality and Performance (1998).

Andrew Bennett is Associate Professor of Government at Georgetown University and the author of Condemned to Repetition? The Rise, Fall, and Reprise of Soviet-Russian Military Interventionism, 1973-1996 (MIT Press, 1999).

In this book, George and Bennett explain how research methods such as process tracing and comparative case studies are designed, carried out, and used as the basis for theory development in social science. They provide an invaluable research guide for any scholar interested in the case study approach. But the book is much more than an account of how to do case study research. The authors also offer a sophisticated discussion of the philosophy of science that will be useful to anyone interested in the place of case-study methods in broader debates about social science methodology, and they give a discerning analysis of policy-relevant theory that is sure to draw the attention of a research community increasingly concerned about the social and political relevance of modern social science. In scope, clarity, and erudition, this book sets a new standard not only in the analysis of case study methods, but also in the study of social science methods more broadly. David Dessler, Associate Professor of Government, College of William & Mary

This book combines clear and concise instructions on how to do qualitative research with sophisticated but accessible epistemological reasons for that advice. The volume provides step-by-step templates on ways to design research, compare across cases, congruence test and process trace, and use typological theories. This guidance is illustrated with dozens of concrete examples. Almost no other methodology text comes close to matching the authors' top-to-bottom synthesis of philosophy of science and practical advice. Colin Elman, Executive Director, Consortium on Qualitative Research Methods, Assistant Professor of Political Science, Arizona State University

This landmark study offers to scholars of all methodological persuasions a philosophically informed, theoretically nuanced, and methodologically detailed treatment of case study analysis. With this book Alexander George and Andrew Bennett help all of us in improving our research, teaching, and disciplinary debates. Peter J. Katzenstein, Walter S. Carpenter, Jr., Professor of International Studies, Cornell University

Case Studies and Theory Development in the Social Sciences makes an indispensable contribution to the growing literature on qualitative methods in the social sciences. It provides a definitive analysis of case study methods and research designs, anchors those methods in contemporary philosophy of science, and argues that case study, statistical, and formal approaches can and should be mutually reinforcing in the development and testing of social theories. Jack S. Levy, Board of Governors' Professor, Rutgers University

Today, more and more social scientists recognize the importance of cases in social and political research and are looking for new ways to make their research more case oriented. George and Bennett show how in this important new work. The beauty of their approach is their careful integration of theory and method and their conviction that the pursuit of empirical knowledge is profoundly theory dependent. Charles Ragin, Professor of Sociology, University of Arizona

Andrew Bennett and Alexander George have written an immensely helpful practical guide to the case method. It offers sharp insight on scientific inference and very useful how-to guidance on doing case studies. Graduate students in social science: don't leave home without it! Stephen Van Evera, Professor of Political Science, Massachusetts Institute of Technology

The history of social science shows that well-designed case studies can be both a fertile source of new theories and a powerful tool for testing them. Case Studies and Theory Development in the Social Sciences raises our understanding of case study methodology to a new level of rigor and sophistication. George and Bennett provide a careful analysis of the virtues and pitfalls of comparative case study research and offer valuable advice for any scholar engaged in qualitative research. The more widely this book is read, the better future social science will be. Stephen M. Walt, Robert and Renée Belfer Professor of International Affairs, John F. Kennedy School of Government, Harvard University

This is an extraordinarily valuable book—a guide written with the practitioner in mind, very sophisticated in its approach to the subject, but loaded with practical advice. George and Bennett show how systematic, rigorous, and above all meaningful case study work is to be done. This is the sort of book scholars—and not just graduate students—will want to come back to over and over again. Marc Trachtenberg, Professor of Political Science, University of California, Los Angeles

Additional Material

Sample Chapter

Related Books

No internet connection.

All search filters on the page have been cleared., your search has been saved..

• Sign in to my profile My Profile

Subject index

This is the most comprehensive guide to the current uses and importance of case study methods in social research. The editors bring together key contributions from the field, which reflect different interpretations of the purpose and capacity of case study research. They address issues such as: the problem of generalizing from the study of a small number of cases; and the role of case study in developing and testing theories. The editors offer in-depth assessments of the main arguments. An annotated bibliography of the literature dealing with case study research makes this an exhaustive and indispensable guide.

Case Study and Theory in Political Science

• By: Harry Eckstein
• In: Case Study Method
• Chapter DOI: https:// doi. org/10.4135/9780857024367.d11
• Subject: Anthropology , Business and Management , Criminology and Criminal Justice , Communication and Media Studies , Economics , Education , Geography , Health , Marketing , Nursing , Political Science and International Relations , Psychology , Social Policy and Public Policy , Social Work , Sociology
• Keywords: inquiry ; knowledge ; law ; probing ; social science ; types of study
• Show page numbers Hide page numbers

[…] Case studies run the gamut from the most microscosmic to the most macrocosmic levels of political phenomena. On the micro level, we have many studies of conspicuous political personalities (political leaders such as Lincoln, Stalin, Gandhi), and of particular leadership positions and small leadership groups (the American presidency, the British Cabinet, the prime minister in British government, the operational code of the Soviet leadership, and so on). […] Beyond that level, one finds a similar profusion of case studies of transnational phenomena: specific processes of and organizations for transnational integration, particular ‘systems’ of international politics, particular crises in international relations, and the like. […]

This plenitude of case studies is not associated with any perception that they are a particularly useful means for arriving at a theoretical understanding of the subject matter of political study. […] One might explain this apparent paradox by holding that political scientists do not place a high value on theory building. No doubt this is true for many of them. But it is much less true nowadays than it used to be, and the volume, or proportion, of case studies in the field has not notably decreased.

It is in order, therefore, to raise three questions: What general role can the case study play in the development of theories concerning political phenomena? How useful is the case method at various stages of the theory-building process? And how is case study best conducted for purposes of devising theories? […]

Case studies, I will argue, are valuable at all stages of the theory-building process, but most valuable at that stage of theory building where least value is generally attached to them: the stage at which candidate theories are ‘tested’. Moreover, the argument for case studies as a means for building theories seems strongest in regard to precisely those phenomena with which the subfield of ‘comparative’ politics is most associated: macropolitical phenomena, that is, units of political [Page 120] study of considerable magnitude or complexity, such as nation-states and subjects virtually coterminous with them (party systems or political cultures). More precisely, the abstract brief in favour of the case study as a means of building theories seems to me to hold regardless of level of inquiry, but at the macrocosmic level practical research considerations greatly reinforce that brief. […]

Definitions

Case study and comparative study.

1. The conception of case study commonly held in the social sciences is derived from, and closely similar to, that of clinical studies in medicine and psychology. Such studies are usually contrasted dichotomously (as if they were antitheses) to experimental ones, which furnish the prevalent conception of comparative study. Contrasts generally drawn between the two types of study cover virtually all aspects of inquiry: range of research; methods and techniques; manner of reporting findings; and research objectives (see Riley, 1963, pp. 32–75).

As to range of research: experimental studies are held to be conducted with large numbers of cases, constituting samples of populations, while clinical studies deal with single individuals, or at most small numbers of them not statistically representative of a populous set. Experimental studies thus are sometimes said to be ‘extensive’ and clinical ones ‘intensive’. These adjectives do not refer to numbers of individuals alone, but also involve the number of variables taken into account. In experimental studies that number is deliberately and severely limited, and pre-selected, for the purpose of discovering relationships between traits abstracted from individual wholes. Clinical study, to the contrary, tries to capture the whole individual - ‘tries to’ because it is, of course, conceded that doing so is only an approachable, not an attainable, end.

As for methods and techniques: the typical experimental study, first of all, starts with, and adheres to, a tightly constructed research design, whereas the typical clinical study is much more open-ended and flexible at all stages. The clinical researcher may have (probably must have) in mind some notions of where to begin inquiry, a sort of checklist of points to look into during its course, or perhaps even a preliminary model of the individual being studied; but actual study proceeds more by feel and improvisation than by plan. Second, the techniques most commonly associated with such inquiry in the case of ‘collective individuals’ (that is, social units) are the loose ones of participant observation (simply observing the unit from within, as if a [Page 121] member of it) and Verstehen (that is, empathy: understanding the meaning of actions and interactions from the members’ own points of view). The typical techniques of experimental inquiry, per contra , are those rigorous and routinized procedures of data processing and data analysis concocted to ensure high degrees of ‘non-subjective’ reliability and validity - the techniques of the statistics texts and research methods primers.

Reports of the findings of clinical study are generally characterized as narrative and descriptive: they provide case histories and detailed portraiture. Such reporting might therefore also be termed synthetic, while that of experimental studies is analytic, since it presents not depictions of ‘whole’ individuals but rather of relations among components, or elements, of them. Beyond description, clinical studies present ‘interpretation’; beyond raw data, experimental ones present rigorously evaluated ‘findings’.

It follows that the objectives of the two types of study also differ. That of experimental study is generalized knowledge: theoretical propositions. These may certainly apply to individuals but never exhaust the knowledge it is possible to have of them. Being general they necessarily miss what is particular and unique, which may or may not be a lot. The objective of clinical study, however, is precisely to capture the particular and unique, for if anything about an individual whole is such, so must be the whole perse . It is conceded that in describing an individual configuration we may get hunches about the generalizability of relations not yet experimentally studied, but only hunches, and even these only by serendipity (Merton, 1962, p. 103). Clinical study is therefore associated more with action objectives than those of pure knowledge. In the case of single individuals, it aims at diagnosis, treatment and adjustment; in that of collective individuals, at policy. This association of clinical study with adjustive action is based on the assumption that therapy and policy can hardly proceed without something approximating full knowledge of its subjects, however much general propositions may help in proceeding from clinical knowledge of a case to the appropriate manipulation of a subject. Clinical and experimental objectives draw near, asymptotically, as ‘pure’ knowledge becomes ‘applied’ (that is, in engineering models), but application is merely a possible extension of experimental knowledge while generally being an intrinsic objective of clinical research. […]

2. However, while the distinction between clinical and experimental studies is useful for contrasting the old and new comparative politics, it does not serve nearly so well in distinguishing the case study from other modes of research. At best, it can provide an initial inkling (but only an inkling) of the differences among them. […] The distinction offers a useful denotative definition of case studies in the social [Page 122] sciences (that is, what people usually mean by the term) but a far from useful connotative and generic one (how the term ought to be used if it is not to raise serious difficulties of meaning and classification and not to define merely one of numerous types of case study).

3. The essential objections to equating case study with clinical and comparative study with experimental inquiry all revolve on one basic point: nothing compels the clustering (hence, dichotomization) of the various characteristics used to distinguish clinical and experimental studies. Although that clustering in fact occurs very frequently in the social sciences, it does so chiefly because of dubious beliefs and assumptions. At most, the characteristics have a certain practical affinity; for example, the fewer the cases studied, the more intensive study may be, other things being equal. But no logical compulsion is at work, and the practical considerations often are not weighty.

We may certainly begin with the notion that case studies, like clinical studies, concern ‘individuals’, personal or collective (and, for tidiness of conceptualization, assume that only one individual is involved). From this, however, it does not follow that case studies must be intensive in the clinician's sense: nothing like ‘wholistic’ study may be attempted, and the researcher may certainly aim at finding relationships between pre-selected variables - unless he assumes, a priori , that this is foolish. The research may be tightly designed and may put to use all sorts of sophisticated research techniques. (An excellent example is Osgood and Luria's [1954] ‘blind analysis’, using the semantic differential, of a case of multiple personality.) Its results need not be cast in narrative form, and its objective can certainly be the development of general propositions rather than portraiture of the particular and unique; nor need case studies be concerned with problems of therapeutic action when they go beyond narration, depiction and subjective interpretation. […]

These points of overlap and ambivalence in the distinction between the clinical and experimental have led to a concerted attack on the dichotomy in psychology itself. One typical attack argues that the dichotomy originates in an archaic and absurd Methodenstreit between ‘mechanistic’ and ‘romantic’ views of human nature (Holt, 1962). Another argues that experimental modes of study can also be used profitably in research into single cases; this is the theme of a notable book of essays, N = 1 (Davidson and Costello, 1969). This work implies the most important definitional point of all: if case study is defined as clinical study in the traditional sense, then we not only construct a messy generic (not necessarily classificatory) concept, but also foreclose the possibility of useful argument about case study as a tool in theory building. The definition answers the question: case study and theory are at polar opposites, linked only by the fortuitous operation of serendipity.

4. This attack on the conventional idea of case study serves a constructive as well as destructive purpose. It provides ammunition for later arguments against highly restrictive views concerning the role of case study in theory building and also points the way toward a better, and simpler, definition of what case studies are.

An unambiguous definition of case study should proceed from the one sure point that has been established: case study is the study of individuals. That is about as simple as one can get - but, because of one major problem, it is too simple. The problem is that one person's single individual may be another's numerous cases. Take an example: in order to help break down the dichotomy between the clinical and experimental, Davidson and Costello (1969, pp. 214–32) reprint a study by Chassan on the evaluation of drug effects during psychotherapy. Chassan argues for the greater power of single-case study over the usual ‘treatment group’ versus ‘control group’ design - in this case, for determining the relative effects of tranquillizers and placebos. Readers can catch the flavour of his argument through two of his many italicized passages:

The intensive statistical study of a single case can provide more meaningful and statistically significant information than, say, only end-point observations extended over a relatively large number of patients.

The argument cited against generalization to other patients, from the result of a single case intensively studied, can actually be applied in a more realistic and devastating manner against the value of inferences…drawn from studies in which extensive rather than intensive degrees of freedom are used.

And so on, in the same vein. The whole paper is an object lesson to those who seek theoretical safety only in numbers. But there is a catch. Chassan studied only one patient, but used a large number of treatments by drug and placebo: ‘frequent observations over periods of sufficiently long duration’. The individuals here are surely not the patients, although they may be for other purposes; it is each treatment, the effects of which are being compared. It is easy enough to see the advantages of administering different treatments to the same person over a long period (hence, safety in small numbers of a sort), as against using one patient per observation (although it is to Chassan's credit that he pointed them out in contrast to the more usual procedure). But N, despite the title of the book, in this case is not one.

If this problem arises with persons, it arises still more emphatically with ‘collective individuals’. A study of six general elections in Britain may be, but need not be, an N = 1 study. It might also be an N = 6 study. It can also be an N = 120,000,000 study. It depends on whether the subject of study is electoral systems, elections or voters.

What follows from this is that ambiguity about what constitutes an ‘individual’ (hence ‘case’) can only be dispelled by looking not at concrete [Page 124] entities but at the measures made of them. On this basis, a ‘case’ can be defined technically as a phenomenon for which we report and interpret only a single measure on any pertinent variable . This gets us out of answering insoluble metaphysical questions that arise because any concrete entity can be decomposed, at least potentially, into numerous entities (not excluding ‘persons’: they differ almost from moment to moment, from treatment to treatment, and consist of highly numerous cells, which consist of highly numerous particles, and so on). It also raises starkly the critical problem of this essay: What useful role can single descriptive measures (not measures of central tendency, association, correlation, variance or covariance, all of which presuppose numerous measures of each variable) play in the construction of theory? […]

Theory and theory building

We will be concerned with the utility of case studies in the development of theories in macropolitics - their utility both in themselves and, to an extent, relative to comparative ( N = many) studies. While nearly everyone in the field at the present time agrees that the development of good theories is the quintessential end of political inquiry, conceptions of theory, and of the processes by which it may be developed, vary extremely in our field. This makes unavoidable a definitional exercise on theory and a review of the normal steps in theory building.

1. Two polar positions on what constitutes theory in our field can be identified. While positions range between them, they have recently been rather polarized, more often on, or very near, the extremes than between them.

On one extreme (the ‘hard’ line on theory) is the view that theory consists solely of statements like those characteristic of contemporary theoretical physics (or, better, considered to be so by influential philosophers of science). A good summary of this view, tailored to the field of political science, is presented in Holt and Richardson's (1968) discussion of the nature of ‘paradigms’, but even better sources are the writings of scientist-philosophers such as Kemeny (1959), Popper (1959) and Hempel (1965).

Theories in this sense have four crucial traits: (1) The concepts used in them are defined very precisely, usually by stating definitions in terms of empirical referents, and are less intended to describe phenomena fully than to abstract from them characteristics useful for formulating general propositions about them. (2) The concepts are used in deductively connected sets of propositions that are either axioms (assumptions) or theorems deduced from them. (3) The object of the propositions is both logical consistency and ‘empirical import’, that is, [Page 125] correspondence to observations of phenomena. And (4) empirical import is determined by tests themselves deduced from the propositions, and these are designed to make it highly probable that the propositions will flunk the tests because confidence in propositions is proportionate to the stiffness of the tests they manage to survive. In [political science], theories of this type are sometimes called ‘formal theories’, mainly because of the large role of formal deduction in their elaboration; and economics is generally taken as the nearest social science model for them, not only in general form but also in regard to substantive ‘rationality’ axioms (Buchanan and Tullock, 1967; Curry and Wade, 1968; Downs, 1965; Riker, 1962).

On the other pole (the ‘soff Une), theory is simply regarded as any mental construct that orders phenomena or inquiry into them. This qualifies as theory many quite diverse constructs, including classificatory schemes that assign individual cases to more or less general classes; ‘analytic’ schemes that decompose complex phenomena into their component elements; frameworks and checklists for conducting inquiry (for example, the ‘systems’ approach to macropolitics, or ‘decision-making’ checklists for the study of foreign policy formation); any empirical patterns found in properly processed data, or anything considered to underlie such patterns (for example, learning processes or class position).

2. If the term theory were always prefaced by an appropriate adjective, wrangling about these, and less extreme, positions could be avoided. But this would not take us off the hook of having to specify how ‘good theory’ as an objective of inquiry in our field should be conceived. The best position on this issue, it seems to me, is neither hard nor soft but does come closer to the hard than the soft extreme. It rests on two major premises.

The first is that it makes no sense whatever to call any mental construct a theory. Such constructs differ vastly in nature and purpose, so that they can hardly be considered to be of the same species. With some of them, not much more can be done than to assign names to phenomena or to order one's filing cabinets. And it can be demonstrated that, strictly speaking, the soft position compels one to regard as theory any statement whatever in conventional or technical discourse.

Second, it makes little more sense to restrict the term to constructs like those of theoretical physics, or those abstracted from that field by philosophers of science. While such constructs have proved extremely powerful in certain senses, one may doubt that they alone possess power (even in these senses). If constructs like them are not attainable in a field such as [political science] at its present stage of development (which is at least an open question, since constructs like them have in fact not been attained), commitment to theory in such a narrow sense [Page 126] may induce one to forego theoretical inquiry altogether. Most important, theories in the ‘hard’ sense are a particular form developed, over considerable time, to realize the purposes - the motivating goals, animus, telos - of an activity; and while they do this very well, it does not follow that they are absolutely required for realizing these purposes.

Consequently, even if the constructs of theoretical physics are taken as a model, it seems unwise to restrict the notion of theory entirely to such constructs. It seems better to label as theory any constructs designed to realize the same ends and formulated with the same animus as those that characterize the fields in which hard theory has been developed - leaving open, anyway provisionally, the forms such constructs may take consistent with reasonable achievement of the ends. On this basis, theory is characterized by a telos , or animus , of inquiry rather than by the particular form of statements. The only requirement (which, however, is far from soft) is that the forms of theoretical statements must be conducive to the goals of theoretical activity. […]

Theories can, of course, be more or less powerful, or ‘good’, depending on the rulefulness of regularity statements, the amount of reliability and validity they possess, the amount and kinds of foreknowledge they provide, and how parsimonious they are. 1 The animus of theoretical inquiry is constantly to increase their power to some unattainable absolute in all these senses. And while that absolute might have a unique ideal form to which the forms of theoretical physics might provide a discerning clue, it should be evident that it can be approached through many kinds of formulations, and always only approached. This is why ‘theory’ is better conceived of as a set of goals than as statements having a specified form.

At the same time, no mental construct qualifies as theory unless it satisfies the goals in some minimal sense. This minimal sense is that it must state a presumed regularity in observations that is susceptible to reliability and validity tests, permits the deduction of some unknowns, and is parsimonious enough to prevent the deduction of so many that virtually any occurrence can be held to bear it out. If these conditions are not satisfied, statements can still be interesting and useful; but they are not ‘theory’.

These are the sort of constructs we want about macropolitics. It should be evident that the pivotal point in the whole conception is that regarding foreknowledge: validity is held to depend on it, parsimony is mainly required for the sake of establishing validity, and regularity statements are not an end unless valid. Any general appraisal of the utility of a method of inquiry must therefore also pivot on that point, as will my brief for the case study method. 2

3. It should also be evident that foreknowledge is most closely bound up with the testing of theories and that the process of theory building [Page 127] involves much that precedes testing and some activities subsequent to it as well. It follows that modes of inquiry might be highly serviceable at one stage of the process but not at others, and this also must be considered in arguments about them.

(a) The process of theory building, needless to say, always begins with questions about experience for which answers are wanted - and raising questions, especially penetrating ones, is anything but a simple matter; indeed it is perhaps what most distinguishes the genius from the dullard (for whom common sense, the sense of ordinary people, leaves few mysteries). It is also an ability that, conceivably, could be sharpened or dulled by various modes of inquiry.

(b) Questions, to be answered by theories, must usually be restated as problems or puzzles . This is a complex process that I have discussed elsewhere (Eckstein, 1964), and it consists essentially of stating questions so that testable rules can answer them (which is not the case for any and all questions) and determirning what core-puzzles must be solved if questions are to be answered. A familiar example is the subtle process by which Weber arrived at the conclusion that the question of his Protestant Ethic (’Why did modem capitalism as an economic system develop spontaneously only in the modern West?’) boils down to the problem ‘What engenders the (unlikely) attitude of continuous, rational acquisition as against other economic orientations?’

(c) The next step is hypothesis: formulating, by some means, a candidate-solution of the puzzle that is testable in principle and sufficiently plausible, prima facie , to warrant the bother and costs of testing. Like the formulation of theoretical problems this initial step toward solving them generally first involves a ‘vision’, then the attempt to state that vision in a rigorous and unambiguous form, so that conclusive testing becomes at least potentially possible. The candidate-solution need not be a single hypothesis or integrated set of hypotheses. In fact, a particularly powerful alternative is what Platt (1966) calls ‘strong inference’ (and considers characteristic of the more rapidly developing ‘hard’ sciences, such as molecular biology and high-energy physics): developing a set of competing hypotheses, some or most of which may be refuted by a single test.

(d) After that, of course, one searches for and carries out an appropriate, and if possible definitive, test . Such tests are rarely evident in hypotheses themselves, especially if questions of practicability are added to those of logic.

Testing is, in a sense, the end of the theory-building process. In another sense, it is not: if a test is survived the process of theorizing does not end. Apart from attempting to make pure knowledge applied, one continues to keep an eye out for contradictory or confirmatory observations, continues to look for more definitive tests, and continues to look for more powerful rules that order larger ranges of observations, [Page 128] or the same range more simply, or subsume the tested rule under one of a higher order, capable of subsuming also other tested rules. […]

Options on the utility of case studies: an overview of the argument

The options.

In taking positions on the value of case studies for theory building, both in themselves and relative to comparative studies, one can choose between six, not all mutually exclusive, options. These have been derived from a review of actual political case studies, the scant methodological literature about them (and counterparts in other social and behavioural sciences), and my own reflections on unconsidered possibilities. They are listed in order of the value seen in case studies, especially as one progresses along the path of theory building - a progression in which, arguably, intuitive vision plays a constantly decreasing role relative to systematic procedure.

Option 1 holds case studies and comparative studies to be wholly separate and unequal. They are separate in that the two modes of inquiry are considered to have so little in common that case studies are unlikely to provide more than a severely limited and crude basis for systematic comparisons (for example, variables of major importance in ‘ N = many’ studies might be wholly ignored in studies of pertinent cases or might not be treated in readily comparable ways, and so on). The two modes of study are unequal in that only comparative studies are associated with the discovery of valid theories; case studies are confined to descriptions and intuitive interpretations.

Option 2 desegregates case studies and comparative studies, but hardly lessens their inequality. It holds that the two modes of inquiry draw near (asymptotically) in the interpretation of cases, because such interpretations can be made only by applying explicit or implicit theoretical generalizations to various cases. Case study, however, remains highly unequal because it is certainly not required, nor even especially useful, for the development of theories. There is an exception to this principle, but it is very limited. One never, or at least very rarely, has all the theories needed to interpret and treat a case; hence, something in the process of case interpretation must nearly always be left incomplete or to intuitive insight (which is why the two modes may approach closely but not intersect). Any aspects of case interpretation in regard to which theory is silent may be regarded as questions on the future agenda of theory building, as any intuitive aspects of interpretation may be regarded as implicit answers to the questions.

Option 3 grows out of the exception to Option 2. It holds that case studies may be conducted precisely for the purpose of discovering questions and puzzles for theory and discovering candidate-rules that might solve theoretical puzzles. The idea is simply that, if subjects and insights for comparative study are wanted, case study can provide them, and that case study might be conducted precisely for that purpose; and perhaps satisfy it by something less chancy than serendipity, or at least by affording larger scope to serendipitous discovery than studies that sacrifice intensive for extensive research. This still confines the utility of case study to the earlier stages of theorizing and makes it a handmaiden to comparative study. But it does tie case study into the theory-building process by something less contingent than possible feedback flowing from the ‘clinician’ to the ‘experimentalist’.

Option 4 focuses on the stage in theory building at which one confronts the question whether candidate-rules are worth the costs (time, effort, ingenuity, manpower, funds, and so on) of testing. It holds that, although in the final analysis only comparative studies can really test theories, well-chosen case studies can shed much light on their plausibility, hence whether proceeding to the final, generally most costly, stage of theory building is worthwhile. This clearly involves something more than initial theoretical ideas. It begins to associate case study with questions of validity, if only in the grudging sense of prima facie credibility.

Option 5 goes still another step further, to the testing (validation) stage itself. It might be held (no revelation forbids it) that in attempting to validate theories, case studies and comparative studies generally are equal, even if separate, alternative means to the same end. The choice between them may then be arbitrary or may be tailored to such non-arbitrary considerations as the particular nature of theories, accessibility of evidence, skills of the researcher, or availability of research resources. A corollary of this position is, of course, that case studies may be no less systematic in procedure and rigorous in findings than comparative studies.

Option 6 is the most radical from the comparativist's point of view. It holds that case studies are not merely equal alternatives at the testing stage, but, properly carried out, a better bet than comparative studies. It might even be extended to hold that comparative studies are most useful as preliminary, inconclusive aids to conclusive studies: that is, the former may suggest probabilities and the latter clinch them. (Beyond this, of course, lies the still more radical possibility that comparative studies are good for nothing, case studies good for everything. But all inquiry suggests that this is wrong, and while the history of ideas also suggests that the unthinkable should be thought, there is no point in doing so unless a good case can be made for Option 6.)

Arguments on the options: an overview

The options discussed above tell us how we might answer the first two questions posed in the introduction to this chapter, while the answer to the third depends on the others. Since the answers to be proposed are complex and the manner of the presentation is far from simple (being intended to present others’ views as well as my own), I will outline them before arguing them, as a sort of map to the discussion.

1. First of all, a taxonomic point should be emphasized. This is that ‘case study’ is in fact a very broad generic concept, whether defined technically as ‘single-measure’ study or by the simpler ‘single-individual’ criterion. The genus can and, for our purposes, must be divided into numerous species, some of which closely resemble, some of which differ vastly from, the model of clinical study. The species that need distinction are: configurative-idiographic studies, disciplined-configurative studies, heuristic case studies, case studies as plausibility probes and crucial case studies . There may be still other types, but these five occur most frequently or are of most consequence to us.

Two things are notable about these species. They are intimately associated with the options on the utility of case studies in theory building: each option is linked to a special type of case study (except that Options 5 and 6 make no difference to the type of case study used). And as the utility attributed to case study increases, especially in progression through the phases of theory building, the associated type of case study increasingly departs from the traditional mode of clinical research and, except for numbers of individuals studied, increasingly resembles that of experimental inquiry.

2. As for choice among the options, and associated types of case study, it seems that the modal preference of contemporary political scientists is the third and/or second (not so different, except in nuance, that they preclude being chosen in conjunction); that few choose the fourth (more for reasons of unfamiliarity than methodological conviction); that Options 5 and 6 are not chosen by anyone, or at least by very few. The evidence for this is mostly what political scientists actually write, reinforced by reactions to a preliminary version of this paper by a pretty fair cross-section of fellow professionals and a desultory poll among colleagues and students (only one of whom chose any option beyond the third, and that only because he reckoned that no one would list other possibilities unless up to tricks).

The prevailing preferences seem worth challenging on behalf of the options more favourable to case study. The latter appear to be rejected (better, not considered) for reasons other than full methodological deliberation, more as a result of overreaction against one weak type of case study than because of full consideration of the whole range of [Page 131] alternatives. In consequence, potentially powerful types of case study are neglected, and case studies are carried out less rigorously than they might be. Arguably, as well, this incurs liabilities in the conclusiveness of theories and the definitiveness of findings.

I propose to conduct the argument to this effect by evaluating each option, and associated type of case study, seriatim . In gist, my argument runs as follows:

(a) Option 1 is hardly worth arguing against. Its basic premise - that comparative and case studies are, for all intents and purposes, antithetical - has been exploded for good and all by Verba (1967) in [political science], and has been widely attacked in other social sciences as well (Davidson and Costello, 1969; Holt, 1962). Nevertheless, it is worth discussing because the type of case study associated with it was once dominant (and is still fairly common), and still provides the most widely prevalent notion of what case study is all about and of its potential for theory building.

(b) All the other optional are tenable, but only because there are different types of case study that have different power in regard to theory building, and because the utility of case studies is not fully determined by logic (abstract methodology) but depends also on practical considerations (for example, characteristics of one's subject matter).

(c) Options 2 and 3 identify perfectly legitimate uses of case study and methods of carrying them out. They are implicit in a host of meritorious political studies, but these studies do not come near exhausting the utility of case study for theory building. Case studies may be used not merely for the interpretive application of general ideas to particular cases (that is, after theory has been established) or, heuristically, for helping the inquirer to arrive at notions of problems to solve or solutions worth pursuing, but may also be used as powerful means of determining whether solutions are valid.

(d) Option 4 deserves special consideration, for two reasons. It identifies an objective for which case study of a particular type is eminently serviceable and which can be of vast importance in theory building, but which is rarely pursued, by case study or other means. In addition, the utility of case study for that objective prepares the ground for arguing the case for the more radical options remaining.

(e) Option 5 will be held to state the logically most defensible position: to attain theory in political inquiry, comparative studies and case studies should be considered, by and large, as alternative strategies at all stages, with little or nothing to choose (logically) between them. Since that argument will be most difficult to sustain - at least against the conventional wisdom - for the testing stage, the argument will concentrate on the type of case study suitable to it.

(f) When practical considerations are added to logic, Option 6 seems still more sensible, at least for studies of politics on the macro level.

Case study is generally a better choice than comparative study for testing theories in macropolitics, but the type of case study useful for this purpose requires a kind of prior knowledge for which preliminary comparative study (of a limited kind) may often be useful or even necessary. This amounts to saying that comparative study can, in some circumstances, be treated as a handmaiden to case study, not vice versa, and thus, in a sense, stands the popular Option 3 on its head.

Before working through all this in detail, I want particularly to emphasize two points. First, nothing that follows should be regarded as an attack on the utility of comparative study in theory building, simply because case studies are defended. […] Second, it is not to be inferred that just any case study will do for the purposes of theory building. Some readers of a draft of this essay concluded that it constituted a defence of ‘traditional’ political studies against the ‘behaviouralists’. This is ludicrous, but it occurred. The discussion presents an argument both for case studies and for carrying them out in a particular way . Since the type of case study for which it argues is very demanding, implying great rigour of thought and exactitude of observation, it is hardly ‘anti-behaviouralist’; and since that type of case study, to my knowledge, is as yet virtually non-existent in [political science], the argument can hardly be ‘traditionalist’.

Types and uses of case study

Configurative-idiographic study.

1. In philosophy and psychology a distinction has long been drawn between nomothetic (generalizing, rule-seeking) and idiographic (individualizing, interpretive) types of, or emphases in, science. The philosophic progenitor of this terminology (and, in part, the ideas that underlie it) is Windelband ([1982]/1998), the most notable contemporary defender of the distinction is Gordon Allport. 3 Idiographic study is, in essence, what was earlier described as clinical study, and configurative-idiographic study is its counterpart in fields, like macropolitics, that deal with complex collective individuals. (Verba calls them configurative-idiosyncratic studies, but the difference in terminology is of no consequence.)

The configurative element in such studies is their aim to present depictions of the overall Gestalt (that is, configuration) of individuals: polities, parties, party systems, and so on. The idiographic element in them is that they either allow facts to speak for themselves or bring out their significance by largely intuitive interpretation, claiming validity on the ground that intensive study and empathetic feel for cases provide authoritative insights into them.

If configurative-idiographic studies are made from philosophic conviction, then the following assumptions usually are at work (see Holt, 1962, pp. 388–97): (a) In the study of personalities and the collectivities they form, one cannot attain prediction and control in the natural-science sense, but only ‘understanding’ (Verstehen) - and thus, from understanding, limited, non-deductive conceptions of probable futures and prudent policy. (b) In attaining understanding, subjective values and modes of cognition are crucial, and these resist quantification. (c) Each subject, personal or collective, is unique, so generalizations can at most be only about their actions (persons) or interactions (collectivities). And (d) the whole is lost or at least distorted in abstraction and analysis - the decomposition of the individual into constituent traits and statements of relations among limited numbers of these; it is ‘something more’ than an aggregate of general relations, rather than ‘nothing but’ such an aggregate. […]

2. Configurative-idiographic studies are certainly useful, and, at their best, have undeniably considerable virtues. They may be beautifully written and make their subjects vivid. They may pull together and elegantly organize wide and deep researches. The intuitive interpretations they provide may be subtle and persuasive and suggest an impressive feel for the cases they treat.

Their most conspicuous weakness is that, as Verba puts it, ‘they do not easily add up’ - presumably to reliable and valid statements of regularity about sets of cases, or even about a case in point (Kaufmann, 1958; Lowi, 1964). This is plain in regard to sets of cases; the summation regarding them is at most factual (information about similar subjects, for example, legislatures, parties, and so on, in different contexts) and, because of idiosyncrasies in fact collecting and presentation, rarely involving even the systematic accumulation of facts. Anyone who has used secondary sources for compiling comparable data on numerous cases knows this to his pain, and, even more painfully, that inventories of interpretive propositions culled from case studies usually contain about as many distinguishable items as studies. The point is less plain, but just as true, for regularity statements concerning individual cases. The interpretations, being idiosyncratic, rarely come to an agreed position, or even to a point of much overlap. For example, in the configurative-idiographic literature on France there seems to be overlap on the position that there are ‘two Frances’, but nearly everyone has his own conception of what they are and where they are found. This situation is hardly surprising: in configurative-idiographic study the interpreter simply considers a body of observations that are not self-explanatory, and, without hard rules of interpretation, may discern in them any number of patterns that are more or less equally plausible.

The criticism that configurative-idiographic study does not add up to theory, in our sense, is mitigated by the fact that its capability to do [Page 134] so was never claimed by its exponents; in fact it is often explicitly repudiated. What is really troublesome about configurative-idiographic study is the repudiation itself (that is, the claim that case study in the behavioural and social sciences can only be idiographic) and its consequences for the way in which the nomothetic utility of case studies in these fields is regarded.

For a thorough refutation of the idiographer's position, and a broad attack on the distinction between the nomothetic and idiographic itself, readers should consult Holt (1962). His argument, in gist, is

(a) that both the position and distinction have ‘peculiar origins’ - misunderstandings of Kant by lesser German philosophers and ‘romantic’ assumptions prevalent during the early nineteenth century (‘Teutonic ghosts’ raised against classical ideas and styles) that led to unreasonably sharp lines between nature and mechanisms, on one hand, and behaviour and organisms, on the other; and, more important,

(b) that none of the postulates of idiographic study, as outlined above, withstands examination. As for the consequences of the claims of idiographers, the most stultifying has been the association of nomothetic study in macropolitics with study different from that favoured by idiographers in all respects: not only study based on more systematic methods of collecting and processing data and on explicit frameworks of inquiry intended to make for cumulation, but ‘comparative’ (that is, multi-case, cross-national, cross-cultural) studies.

If case study could only be configurative-idiographic in character, then the conclusions that case studies and comparative studies are wholly antithetical and that theories about politics require comparative study, or are unattainable, could not be avoided. But case study need not have that character, and the comparativists themselves have pointed the way to other varieties - without, however, overcoming a fundamental bias against case study of any kind in theory construction, largely anchored to the archetype of such studies in our field.

Disciplined-configurative study

1. The comparativist's typical reaction to the theoretical poverty of configurative-idiographic studies is to hold that, while theories cannot be derived from case interpretations, such interpretations can, and should, be derived from theories. ‘The unique explanation of a particular case’, says Verba (1962, p. 114), ‘can rest on general hypotheses’. Indeed, it must rest on them, since theoretical arguments about a single case, in the last analysis, always proceed from at least implicit general laws about a class or set to which it belongs or about universal attributes of all classes to which the case can be subsumed.

The logic involved has been succinctly stated in Hempel's discussion of ‘scientific’ explanation, the essence of which is the explanation of particular phenomena (in my terms ‘case interpretation’) ‘by showing that [their] existence could have been inferred - either deductively or with a high probability - by applying certain laws of universal or of statistical form to specified antecedent circumstances’ (Hempel, 1965, pp. 299–303). Those who consider this the only way of interpreting cases scientifically hold that the theoretical bases of case interpretations should always be made explicit, and that ad hoc additions to a framework of case interpretation should always be made as if they were general laws, not unique factors operating only in the case in point. The bases of case interpretation, in other words, should be established theories or, lacking them, provisional ones, and such interpretations can be sound only to the extent that their bases are in fact valid as general laws. Case studies so constructed are ‘disciplined-configurative studies’. The terminology is Verba's, who recommends such studies to us. […]

Disciplined-configurative studies need not just passively apply general laws or statements of probability to particular cases. A case can impugn established theories if the theories ought to fit it but do not. It may also point up a need for new theory in neglected areas. Thus, the application of theories to cases can have feedback effects on theorizing, as Hempel recognizes. In addition, it is unlikely that all aspects of a case can be nomologically explained. As in the field of engineering, where general theories are applied to achieve conscious ends in particular circumstances, there are nearly always elements of prudence, common sense or ‘feel’ in case interpretations. Theory building, however, aims at the constant reduction of those elements, by stating notions that fit particular cases as general theoretical rules and subjecting them to proper theoretical tests.

In essence, the chain of inquiry in disciplined-configurative studies runs from comparatively tested theory to case interpretation, and thence, perhaps, via ad hoc additions, newly discovered puzzles and systematized prudence, to new candidate-theories. Case study thus is tied into theoretical inquiry - but only partially, where theories apply or can be envisioned; passively, in the main, as a receptacle for putting theories to work; and fortuitously, as a catalytic element in the unfolding of theoretical knowledge. This is, of course, still close to the clinician's conception of his role, and configurative studies that are disciplined in intent are not always easy to distinguish from unadulterated idiography. The two types of often intermixed and easily blend together.

2. The essential basis of Verba's argument about the relations between general theory and particular case interpretation is surely correct. If the interpretations of a case are general laws correctly applied to the case, [Page 136] the interpretations may be valid or invalid, depending on whether the laws are valid; otherwise, their validity simply cannot be known at all. Moreover, if cases are complex, the number of possible alternative interpretations, equally plausible because not at variance with the facts of the case, is usually vast, so that undisciplined case interpretation in much-studied cases usually yields large inventories of quite different propositions, none of which is clearly superior to any other. Preferences among them depend on personal tastes or general intellectual fads. […]

It remains to add a point insufficiently stressed in writings on disciplined-configurative studies. The application of theories in case interpretation, although rarely discussed, is not at all a simple process, even leaving aside the question of how valid theory is to be developed. Such applications only yield valid interpretations if the theories permit strict deductions to be made and the interpretations of the case are shown to be logically compelled by the theories. In the case of revolutions, for instance, it is not enough to know that a regularity exists and that a case somehow ‘fits’ it (that is, does not manifestly contradict it). One should also be able to demonstrate, by correct reasoning, that, given the regularity and the characteristics of the case, revolution must have occurred, or at least had a high probability of occurring. Not all theories permit this to be done, or at least equally well. For example, a theory attributing revolution to aggressions engendered by social frustrations will hardly fail to fit any case of revolution, nor tell us exactly why any case of it occurred. Unless it specifies precisely how much and what sort of frustration engenders revolution, on whose part, and under what complex of other conditions (Gurr, 1970), the frustration-aggression theory of revolution, applied, say, to the French Revolution, can yield about as many plausible case interpretations as can configurative-idiographic study (there having existed many sources of frustration in the ancien régime , as in all regimes).

This point brings out a major utility of attempting disciplined case interpretation. Aiming at the disciplined application of theories to cases forces one to state theories more rigorously than might otherwise be done - provided that the application is truly ‘disciplined’, that is, designed to show that valid theory compels a particular case interpretation and rules out others. As already stated, this, unfortunately, is rare (if it occurs at all) in political study. One reason is the lack of compelling theories. But there is another, which is of the utmost importance: political scientists reject, or do not even consider, the possibility that valid theories might indeed compel particular case interpretations. The import of that possibility, assuming it to exist, lies in the corollary that a case might invalidate a theory, if an interpretation of the case compelled by the theory does not fit it.

But this goes too far ahead, toward a crucial argument that will require much discussion below. The point for the present is merely that [Page 137] exponents of disciplined-configurative study have insufficiently considered both the difficulties and promises of the relations between general theories and particular case interpretations. 4

Heuristic case studies

1. Disciplined-configurative study assumes that ‘general laws’ are available. It is not thought of as a part of the process of theory building as such, except in that the interpretation of cases may lead to ad hoc , serendipitous additions to existing theories in order to cover puzzling aspects of a case. However, the feedback effect in Verba's recommended sequence of inquiry can be isolated from the rest of the sequence and case study deliberately used to stimulate the imagination towards discerning important general problems and possible theoretical solutions. That is the essence of heuristic case studies (heuristic means ‘serving to find out’). Such studies, unlike configurative-idiographic ones, tie directly into theory building and therefore are less concerned with overall concrete configurations than with potentially generalizable relations between aspects of them; they also tie into theory building less passively and fortuitously than does disciplined-configurative study, because the potentially generalizable relations do not just turn up but are deliberately sought out.

Heuristic case studies do not necessarily stop with one case, but can be conducted seriatim, by the so-called ‘building-block technique’ (Becker, 1968) [in other words, analytic induction - Editors’ Note] , in order to construct increasingly plausible and less fortuitous regularity statements. This technique is quite simple in principle. One studies a case in order to arrive at a preliminary theoretical construct. That construct, based on a single case, is unlikely to constitute more than a slim clue to a valid general model. One therefore confronts it with another case that may suggest ways of amending and improving the construct and achieve better case interpretation; and this process is continued until the construct seems sufficiently refined to require no further major amendment or at least to warrant testing by large-scale comparative study. Each step beyond the first can be considered a kind of disciplined-configurative study, but is better regarded as heuristic case study proceeding with increasingly refined questions and toward increasingly more specific ends. It is important not to confuse the whole process with comparative study. The latter seeks regularities through the simultaneous inspection of numerous cases, not the gradual unfolding of increasingly better theoretical constructs through the study of individuals. Of course, comparative studies can also employ the building-block technique by successively refined theories through a series of multi-case studies. […]

2. The justification for heuristic case studies runs as follows: (a) Theories do not come from a vacuum, or fully and directly from data. In the final analysis they come from the theorist's imagination, logical ability and ability to discern general problems and patterns in particular observations, (b) There are ineffable differences in such imaginative and other abilities, but various aids can be used to stimulate them: among them, the printouts from data banks or other comparative studies (which, however, never obviate the use of theoretical imagination, for example, for interpreting the print-outs into proper regularity statements and for determining what data banks should contain or how comparative studies should be designed in the first place). (c) The track record of case studies as stimulants of the theoretical imagination is good. (d) One reason it is good is precisely that, unlike wide-ranging comparative studies, case studies permit intensive analysis that does not commit the researcher to a highly limited set of variables, and thus increases the probability that critical variables and relations will be found. The possibility of less superficiality in research, of course, also plays a role here.

3. Arguments in favour of heuristic case studies surely have merit. Whatever logic might dictate, the indubitable fact is that some case study writers in macropolitics have come up with interpretations notably incisive for their cases and notably plausible when taken as generalizations for sets of them, with or without the benefit of special frameworks or approaches. See, for example, the works of such men as Tocqueville, Bagehot, Halévy, Bryce and Bodley, or, in another field, anthropologists too numerous to mention.

Nevertheless, one may argue that too much is made of heuristic case studies, for two related reasons. One is that those who defend them sometimes seem to do so simply because they can see no more ambitious function to be served by case study. The other is that, not wishing to make other claims but to defend case studies, they claim too much for such studies as heuristic tools, especially in comparison to ‘N = many’ studies. Scenting a valid claim, they exaggerate it - and miss the possibility that a more persuasive brief might be based on a greater sense of limitation at the heuristic stage of theory building and a lesser one at others.

The point that case studies are good for more than getting clues will concern us later. But the anticipation of that point in the previous section can be supplemented here by a further suggestive argument. Case studies intended to serve a heuristic function can proceed much in the manner of ‘clinical’ study, that is, with a minimum of design or rigour, and tackle any case that comes to hand. In that event, however, nothing distinguishes the study from configurative-idiographic study, except the researcher's hopes and intentions, and results can only turn [Page 139] up by good fortune - which the bright will seize and the dull miss, but which the researcher can do nothing to induce. The alternatives are to use at least a modicum of design and rigour in research and not to choose just any case on any grounds but a special sort of case: one considered likely to be revealing, on some basis or other. The suggestive point in this for later argument is not that case study may often depart markedly from the archetype of clinical study (although that is noteworthy), but that certain kinds of cases may be regarded as more instructive for theory building than others . Actual heuristic case studies seem in fact generally, even if often just implicitly, to make that claim for the cases selected. The grounds are often obscure, and the claim often seems post hoc and intended to disarm charges of idiography. The point nevertheless remains that the brief for heuristic case study is strong only to the extent that cases especially instructive for theory, and subject to rigorous inquiry, can be identified. And if that possibility exists, then the further possibility arises that some cases might be especially instructive also at other stages of the theory-building process.

If the prevalent emphasis on heuristic functions is too modest, in what senses does it also exaggerate? First of all, the fact that case study writers have often spawned ideas notable as generalizations proves nothing. The Tocquevilles or Bagehots might have been successful in spawning plausible theories without writing case studies, since their imagination and incisiveness clearly matter more than the vehicles chosen for putting them to work. If they had used comparative studies they might have been even more successful, and more successful still if they had had available modem technology for accumulating, coding, storing and processing data - not to mention the fact that they do always make implicit, sometimes explicit, use of comparisons in their case studies (for example, Bagehot's contrasts between Britain and America, Tocqueville's between America and France), even if only to demonstrate that factors used to interpret their cases do in fact differ in different cases. Moreover, for every case study that has notably succeeded in spawning theory, there are scores that have notably failed - and this does not refer to idiography alone. Case study certainly furnishes no guarantee that theoretical abilities will be awakened or sharpened. And comparativists have been at least as successful in spawning theories as configurativists; for every Tocqueville or Bagehot we can produce an offsetting Aristotle, Machiavelli, Mosca, Pareto or Weber.

Second, the benefit of being able to take into consideration more variables in case study incurs the cost of highly circumscribed breadth of inference. And it is probable that the number of hypotheses suggested, hence also the number of invalid ones to be pursued, will be proportional to the number of variables considered. Heuristic case studies have a demonstrable tendency, as in the case of studies of the French Revolution, to spawn a crushing and chaotic number and variety [Page 140] of candidate generalizations, or hypercomplex multivariate theories, especially when these studies are made by imaginative people. And, unlike comparative studies, they cannot even yield initial clues about the generalizability of relations selected from all those that constitute the case - unless, to repeat, the case is considered, on some good basis, especially revealing for sets of phenomena, that is, one for which breadth-of-inference problems may be claimed to be slight.

These problems have led some to identify ‘grounded theory’ (theory that is initially derived from observations, not spawned wholly out of logic and imagination) with comparative inquiry rather than case study (Glaser and Strauss, 1967). The reasons for doing so are rather convincing. But the more sensible position surely is that, if we are really only concerned with the initial formulation of candidate-theories as a phase of theoretical inquiry (and not theory leaping full-blown out of data), case study is useful but by no means indispensable, as also is comparative study or any other exercise of the theoretical imagination. It is manifestly more useful for some people than others. It also would be generally more useful than it has been if more case studies were deliberately undertaken as exploratory means for arriving at candidate-theories, rather than simply allowing these to occur fortuitously, and if special characteristics of heuristically instructive cases could be specified and something like a heuristic ‘method’ could be developed.

If nothing more were to be said for case studies than that they may be helpful in initially formulating candidate regularity statements, we could only conclude that there is not special reason for either making or not making such studies. It follows that if there is a strong justification for case studies as tools in developing theories, it must be found in the special utility of such studies at some later stage of the sequence of inquiry by which theories are established, or, at a minimum, their availability as reasonable alternatives to comparative studies during the later, no less than earlier, stages of the theory-building process.

Plausibility probes

1. After hypotheses are formulated, one does not necessarily proceed immediately to test them. A stage of inquiry preliminary to testing sometimes intervenes and ought to do so far more often than it actually does in political study (or in other social sciences). It involves probing the ‘plausibility’ of candidate-theories. Plausibility here means something more than a belief in potential validity plain and simple, for hypotheses are unlikely ever to be formulated unless considered potentially valid; it also means something less than actual validity, for which rigorous testing is required. In essence, plausibility probes involve attempts to determine whether potential validity may reasonably be [Page 141] considered great enough to warrant the pains and costs of testing, which are almost always considerable, but especially so if broad, painstaking comparative studies are undertaken.

Such probes are common in cases where costly risks have to be run. These probes are roughly analogous to the trials to which one subjects a racehorse before incurring the costs of entering and preparing it for a major race: success cannot be guaranteed, but some kind of odds (ratios between certain coasts and probable benefits) can be established. The simple principle at work is that large investments in less likely outcomes are worse propositions than large investments in more likely outcomes. Here the analogy between theorizing and horse racing becomes a little specious, for in probing the plausibility of a theory we can hardly expect to know much, or anything, about previous performance or to have exact estimates of probability like those given by a stopwatch. But we do not lack means for at least getting a reasoned, not merely intuitive, ‘feel’ for the odds against a theory.

At a minimum, a plausibility probe into theory may simply attempt to establish that a theoretical construct is worth considering at all, that is, that an apparent empirical instance of it can be found. I take that (together with heuristic objectives) to be the purpose of Dahl's (1961) influential study of power in New Haven. Dahl, as I read him (contrary to some other interpreters of his work), wants to establish that power in democracy may be ‘pluralistic’, or may not be ‘monolithic’, not that it must be the former and cannot be the latter. The study certainly succeeds in that regard, although it would succeed even more if New Haven had been selected for study because it is typical of a specified class of cases.

Some ways of surmising the plausibility of a theory beyond that minimal point are non-empirical, and since they entail only the cost of thought, these should generally be used before, or instead of, empirical probes. We may have confidence in a theory because it is derived logically from premises that have previously yielded valid theory in a field or because it is derived from premises contrary to those that have led to major failures. We may also have confidence in a theory if it is able to account for both strengths and weaknesses in existing relevant hypotheses or otherwise seems to organize considerable volumes and varieties of unexplained data. An example of both these methods of estimating plausibility is furnished by those passages of my monograph on stable democracy that show the grounding of its main proposed regularity statement in (as I then thought) validated psychological theories and those that try to show how the strengths and weaknesses of three alternative hypotheses, all rather powerful yet flawed in certain ways, can be explained by the main proposed regularity statement (Eckstein, 1966). Demonstrating logically that proposed regularity statements can potentially explain data not yet explained, [Page 142] and/or provide a common foundation for previously validated but quite discrete and unconnected hypotheses, and/or extend assumptions found powerful in some areas to other areas, all create presumptions in favour of testing the statements independently, even by costly means.

Plausibility probes can also be directly empirical, that is, in the nature of preliminary, rather loose and inconclusive, but suggestive tests, before more rigorous tests are conducted. Such probes confront theories with lesser challenges that they must certainly withstand if they are not to be toppled by greater ones. If, for example, it were posited that democratic power structures are normally monolithic (which is in fact often done in political theories) and one had strong reason to believe that New Haven was unlikely to be a deviant case (which is also arguable), then Dahl's study of its power structure would establish much more than that the counter-idea of pluralism in democratic power is not completely vacuous (Dahl, 1961). It would cast serious doubt on the posited regularity. Such empirical probes are especially important where non-empirical probes yield very uncertain results, and there is also reason to use them, as additions to others, as cheap means of hedging against expensive wild-goose chases, when the costs of testing are likely to be very great.

2. There is no reason why empirical plausibility probes should not take the form of modest or rather diffusely designed comparative studies, as preludes to more ambitious and tighter ones. Indeed, most systematic comparative studies in macropolitics make more sense as plausibility probes (or as ‘heuristic comparative studies’) than as what they are generally claimed or regarded to be: that is, works presenting definitive results. Almond and Verba's The Civic Culture (1963) is surely a case in point. The sample of cases covered by the study is hardly large and dubiously representative; the regularity statements about ‘democratic stability’ emerging in its final chapter could certainly be made more exact, are mainly afterthoughts imputed to the evidence, and are hardly conclusively compelled by that evidence. But they seem sufficiently rooted in data and reasoning to warrant their statement in more precise form and their thorough testing, preferably by logically deduced predictions about findings in a project specifically designed not to get interesting data but to get those crucial to establishing the validity of the work's central propositions. (One may consider it reprehensible that so many comparativists are willing to stop where only that much, or little more, has been accomplished, and then go on to other, still merely plausible, ideas on different subject matter. We have no right to bewail the fact that others do not take up our ideas if we ourselves drop them far short of the point to which they could be taken.)

The essential point for us is that, as empirical plausibility probes, case studies are often as serviceable as, or more so than, comparative [Page 143] ones - and nearly always a great deal cheaper - a prime consideration in probing plausibility. The economic case for them is strongest where required information is not readily available in aggregate data or good secondary sources and is intrinsically hard to get. Case studies can certainly serve the purpose well if well selected, that is, if they are such that a result, for or against a theory, cannot readily be shrugged off. It is true that case studies have been little used in political studies as plausibility probes, but this is largely due to the fact that the idea of any sort of plausibility probe is foreign to the field, plus the fact that comparative studies to amass data from which finished theories supposedly emerge have been its dominant contemporary genre. (Comparative studies as plausibility probes are equally uncommon, except only in the sense that some of them appear better tailored for that purpose than the purposes they pretend to serve.) […]

3. And here again we come to a critical possibility. If studies of well-selected cases, no less than comparative studies, can serve the purposes of plausibility probes (the idea of which is, after all, to form estimates of probable validity), could they not also serve, painstakingly selected and rigorously carried out, as tests of validity itself, with similar economies in the work required? The possibility should at least be entertained and the case for it argued, since the potential practical gains could be enormous. It arises, at bottom, from the obvious fact that cases are not all equal in their import, even for the modest purposes of heuristic exploration. The question is whether their inequality extends to the point where certain types of cases, and modes of case study, can serve to test theories for validity - the step most demanding on rigour and in which breadth-of-inference problems seem most damaging to case studies.

To explore this question further, we shall have to look more closely at a suggestion made in the discussion on disciplined-configurative study: that if theory can compel particular case interpretations, then particular cases could invalidate or confirm theories.

Crucial case studies

1. The position that case studies are weak or useless for testing theories rests, at bottom, on the mistaken application of a correct principle - a principle that applies more, but still imperfectly, to the discovery of theories in data than to their testing.

We can think of theory formulation as a process that leads one to postulate a curve or line to which observations of reality are expected to correspond; and we can think of theories as valid if the curves or lines that best fit relevant observations in fact match the theoretical [Page 144] expectation, or, put in a different manner, if the points yielded by measurements of relevant observations fall on or very near the postulated curve at logically specified locations. The principle that seems to rule out case study for the purpose of finding valid theory is the elementary one holding that any single instance of a relationship yields only one observed point, and that through any single point an infinite number of curves or lines can be drawn. (A less abstract variant of this principle is the argument above that any number of different explanations, not contrary to fact and thus at least minimally plausible, can be offered for any political event.)

The principle is, as stated, incontestable. But we are not constrained to conclude from it that comparative studies are indispensable to the development of valid theories and case studies useless for the purpose, unless we inject between the premise and conclusion a major fallacy that apparently dies hard: the inductive fallacy. The essence of that fallacy is the belief that theories, being contained in phenomena, can be fully derived from observations by simple inspection or, at any rate, sophisticated data processing. This is fallacious in several senses that should be disentangled, although the fallacies are all of a piece and usually all committed at once.

(a) One aspect of the fallacy involves confusion between the discovery of candidate-theories and their testing: in deriving theory from observations (‘grounded theory’) one may be tempted to think that curves suggested by comparative data are themselves valid theories. This hazard is not logically inherent in comparative study, but contemporary political science, among other fields, suggests that it is extreme in practice, most of all where the behavioural sciences’ model of experimental study is closely followed. Such study, regardless of how punctiliously carried out, cannot, in and of itself, reveal general laws guaranteed to be valid. It only provides more or less powerful clues as to what they are, that is, helps to discover them. In some cases these clues may be so powerful that testing may seem superfluous or not worth the cost, but this is highly exceptional. 5

Strictly speaking, generalizations directly inferred from data only hold (probabilistically) for the phenomena observed under the conditions prevailing during observation. If the observations are voluminous and accurate, and if the conditions of observation are highly various or controlled, then one may have very high confidence that the curve that best fits the observation in fact manifests, in graphic form, a valid theory. Nevertheless, the element of surmise in going from data to theory always is considerable, and the ‘epistemic gap’ between them, as Northrop calls it, ineluctable. And such great limitations usually exist, in practice, on the volume, accuracy and variety or control of observations (including, of course, the obvious limitation that we cannot observe the future) that the risk in identifying an empirical generalization with a [Page 145] theoretical rule usually cannot be defended unless special testing of the presumed rule is carried out.

(b) A second aspect of the fallacy concerns the discovery of candidate-theories in the first place. It is the principles that give rise to empirically discovered curves that constitute theories, not the curves as such: these only represent the principles, that is, show them at work. When an empirically grounded curve has been drawn, therefore, the principles it expresses must still be elucidated. Often this is not much of a problem, and statistical techniques (like causal path analysis) can help solve it. Nevertheless, curves can deceive as well as instruct, regardless of such techniques; and they most resist the discerning of simple regularities governing phenomena because the data from which they are constructed usually express all the complex interactions of factors in the concrete world, or sometimes even the laboratory. Nature, as Bacon knew, is a tough adversary capable of innumerable disguises. More than routine method is often required to strip off these disguises.

It seems, in fact, unlikely that the more powerful laws of physical science could have been discovered (their testing aside) by the mechanical processing of observations, however ‘sophisticated’. Certainly one is struck by the small role played by systematic comparative observation in both their formulation and critical testing - in effect, by the thorough lack of correspondence between the psychologist's and physicist's conception of ‘experimentation’. As illustration, take that touchstone of ancient, modem and contemporary physics: conceptions of gravity and the closely associated law of the velocity of freely falling objects.

The Galilean challenge to the Aristotelian conception of free fall (the heavier the object, the faster it falls), accepted as gospel for nearly two millennia, did not grow out of observation at all but out of a simple ‘thought-experiment’ (simple in retrospect, but apparently not at all obvious until performed). In gist (and with apologies for a layman's bowdlerization): if Aristotle is correct, then two bricks of the same weight, dropped at the same instant from the same height, must strike the ground at the same instant. If the two bricks are dropped side by side, as if cemented, the rate of fall of each must be the same as if dropped separately; but if cemented, they would be twice as heavy as a single brick and must therefore drop much faster; hence, since both conclusions cannot be right, the theory must be wrong. And the only way to square the two conditions logically is to make the weight of the falling objects irrelevant to acceleration in free fall, with the relevant variables being only the gravitational forces that account for falling and, possibly, the duration of the fall.

There is no observation here at all (and much doubt even about whether Galileo ever climbed the Tower of Pisa to check out, by a single ‘probe’, the plausibility of his conclusion). Had systematic comparative measurements been used, anomalies in the Aristotelian [Page 146] conception would certainly have turned up, at least in the fall of objects ‘heavy’ above a certain threshold. But if a well-chosen sample of objects had been dropped from a well-selected sample of heights under a well-selected sample of wind and other conditions, the likely conclusions would surely have been something like this: that the whole process of falling, like macropolitical phenomena, is ‘immensely complicated and cannot be accounted for by one or two simple causes’ (Verba, 1967, p. 114), that weight is a factor (as it is at certain heights and other conditions); and that weight, size, shape and density of object, and wind conditions account for such and such a (no doubt high) percentage of the variance in rates of fall, singly or in various combinations. A radical, deductively fertile simplification of the whole complex process might, but almost certainly would not, have emerged.

The Galilean notion was widely disputed until a crucial experiment could be conducted to check it out. Objects of different weights did demonstrably fall at different, sometimes vastly different, rates. So the Galileans’ extraneous factors, having no place in the law, remained other people's favoured explanatory variables. Only with the invention of the air pump, about 1650, was a definitive experiment possible: dropping a heavy object (coin) and a light one (feather) in an evacuated tube. Again, no systematic comparative measurement was used, only a single experimental observation that foreclosed weaseling out by ad hocery to Galileans and Aristotelians alike. […]

Comparative observations may be significant in fleshing out basic conceptions of regularities (for example, determining that the velocity of freely falling objects is described by s = gt 2 , or measuring any specific value g). But they are far from necessary, and quite likely to deceive, when these basic conceptions - critical variables and their basic relationship - are to be formulated.

(c) The inductive fallacy has a third facet, pertaining exclusively to testing. It might be conceded that discovering and testing theories are different processes, but not that testing requires data different from those that help in discovery. The analysis of data may be so convincing that one might not consider it worthwhile to test rules derived from them but, despite this, the experiences in light of which theories are constructed cannot be used again as tests of them. Testing involves efforts to falsify, and anything giving rise to a theory will certainly not falsify it; nor will any body of replicated observations do so, if replication indeed is faithful. (Replication pertains to reliability, not validity.) The object of testing is to find observations that must fit a theory but have a good chance of not doing so. Nothing that suggests a theory, therefore, can also test it.

2. Having established a need for the independent testing of theoretical curves (on the grounds that the discovery and testing of theories [Page 147] are intrinsically different activities, that no method of discovery can guarantee validity, that even painstakingly gathered and analysed data can deceive, and that data which suggest regularities cannot also validate them), we come to the crux: the argument of the fifth option that, in principle, comparative and case studies are alternative means to the end of testing theories, choices between which must be largely governed by arbitrary or practical, rather than logical, considerations .

Comparative studies can certainly be used to test theories. If we use them for this purpose, our object, as stated, is to demonstrate that a curve that fits their results well in fact closely coincides with a curve postulated by theory, however that may have been worked out. In the case of a law like that of the velocity of falling objects, for example, one might try to demonstrate that the curve yielded by a set of observations sustains the expectation that the postulated law is an increasingly better predictor as one increasingly approximates the conditions under which the law is considered to hold absolutely.

However, there is available (not necessarily in all cases, but in many) an alternative to that rather cumbersome procedure, and it involves a kind of case study. One can use a well-constructed experiment, conducted to simulate as closely as possible the specified conditions under which a law must hold, and compare its result with that predicted by the law. In the history of science the decisive experiments have been mostly of that kind, a fact that makes one wonder how the comparative observation of unmanipulated cases could ever have come to be regarded as any sort of equivalent of experimental method in the physical sciences. (The main reasons are, by my reading, the influence of J.S. Mill's Logic , the intuitive decision reached by some influential contemporary social scientists that their regularity statements must unavoidably be ‘probabilistic’ in form, origin and testing, and the fact that much experimentation in the physical sciences is simply hopeful fishing for regularities in masses of data.) And if a well-constructed experiment can serve the purpose, then so may a well-chosen case - one that is somehow as crucial for a theory as are certain experiments, or indeed natural observations, in the physical sciences.

This argument is not at all impugned by the incontestable principle regarding the relations between points and curves with which we started. For there is another principle about those relations that is equally incontestable. This is that any given point can fall only on an infinitesimal fraction of all conceivable curves: it will not fall on any of the curves, the number of which is also infinity, that do not in fact pass through the point. (A less abstract variant of this principle is that for every plausible explanation of a political event, there is an infinite number that are not even minimally plausible.) The fact that a point falls, or does not fall, on a curve, therefore, is not at all insignificant. If the curve is not constructed to pass through the point but [Page 148] preconstructed to represent a theory, and if, given the nature of a case subsequently examined, we can predict, according to the theory, that it must fall on, or very near, the curve at a specified location, the fact that it does so is of the utmost significance, and its location far from the predicted point will impeach the theory no less than the tendency of several points to describe a divergent curve. At any rate, this is the case if the bases for predicting the location of an unknown point are really compelling - which is the object of crucial case study. In such case study, the compelling instance ‘represents’ a regularity as, in comparative study, a sample of individuals ‘represents’ a population.

3. Crucial case study presupposes that crucial cases exist. Whether they do or not in macropolitics can hardly be settled abstractly. All one can say on the subject is the following: (i) If they do not, no reasonable alternative to testing theory by comparative study exists, (ii) The inability to identify cases crucial for theories may be the result not of their non-existence but of the loose way theories are stated, their relative lack of […] ‘rulefulness’. (iii) Any a priori assumption as to their non-existence manifestly is a self-fulfilling prophecy, and it is difficult to see with what compelling reasoning such an assumption might be justified, (iv) If that reasoning rests on the inability to use controlled laboratory experiments in macropolitics, it suffices to point out that crucial measures in the physical sciences can be of natural observations (as, for example, the confirmation of Einsteinian relativity). And, obviously most important, (v) both hypothetical and actual examples of (apparently) crucial observations in the social sciences, including observations of complex collective individuals, can be found - and would almost certainly be found more often if deliberately sought more often.

A more important question, therefore, is how a crucial case can be recognized. What guidelines can be used?

The essential abstract characteristic of a crucial case can be deduced from its function as a test of theory. It is a case that must closely fit a theory if one is to have confidence in the theory's validity, or, conversely, must not fit equally well any rule contrary to that proposed. The same point can be put thus: in a crucial case it must be extremely difficult, or clearly petulant, to dismiss any finding contrary to theory as simply ‘deviant’ (due to change, or the operation of unconsidered factors, or whatever ‘deviance’ might refer to other than the fact of deviation from theory perse) and equally difficult to hold that any finding confirming theory might just as well express quite different regularities. One says difficult and petulant because claims of deviance and the operation of other regularities can always be made. The question is therefore not whether they are made but how farfetched or perverse the reasons for them (if any) are. […]

An alternative is to focus inquiry on ‘most-likely’ or ‘least-likely’ cases - cases that ought, or ought not, to invalidate or confirm theories, if any cases can be expected to do so. The best-known example in political study probably is Michels’ inquiry into the ubiquitousness of oligarchy in organizations, based on the argument that certain organizations (those consciously dedicated to grass-roots democracy and associated ideologies, representing classes whose interest lies in such democracy, having highly elaborate and pure formal democratic procedures, and leaders from the same social strata as the membership) are least likely, or very unlikely, to be oligarchic if oligarchy were not universal. (One may argue with Michels’ choice of social units, his methods, or his findings - see Willey, 1971 - but the principle of the idea is surely sound.) Another example is Malinowski's (1926) study of a highly primitive, communistic (in the anthropological sense) society, to determine whether automatic, spontaneous obedience to norms in fact prevailed in it, as was postulated by other anthropologists. The society selected was a ‘most-likely’ case - the very model of primitive, communistic society - and the finding was contrary to the postulate: obedience was found to result from ‘psychological and social inducements’. A similar example is Whyte's (1943) study of Boston slum gangs, collective individuals that should, according to prevailing theory, have exhibited a high level of ‘social disorganization’, but in fact exhibited the very opposite.

The ‘least-likely’ case (as in Michels) seems especially tailored to confirmation, the ‘most-likely’ case (as in Whyte and Malinowski) to invalidation. […]

4. To this point, the discussion has presented the case only for Option 5: that case and comparative studies are best conceived as equally useful, alternative means for testing theories. The utility of case study and the weaknesses of comparative studies have been stressed only because the reverse is far more common in [political science].

If logic does not intrinsically favour one method or another, the method to be used must be selected for other reasons, that is, out of practical, prudential considerations. One such reason may be the unavailability of clearly crucial cases. But assuming that not to be the case, probably costs and benefits become the pertinent calculus. On that calculus rests the case for the sixth option on case study, which pertains only to macropolitics. Inquiry into macropolitical units involves problems of scale and of sound comparison that point strongly toward crucial case study as the preferable method; the same considerations might also apply to stages of inquiry other than testing but are less telling there since rigour is at a lesser premium.

The most manifest practical advantage of case study is, of course, that it is economical for all resources: money, manpower, time, effort. [Page 150] The economies are not strictly equal to 1/N, where N is the number of cases studied comparatively, since some resources usually have to be devoted to the identification of crucial cases, and some work needed to prepare rigorous case study is similar to that required by rigorous comparative studies. But even so they are likely to be considerable. This economic advantage is especially important, of course, if studies are inherently costly, as they are if units are complex collective individuals. Sociologists of knowledge might note, in this connection, that the growth of comparative studies coincided with the influx of unprecedented research monies and other facilities (research institutions, crowds of postgraduate students) into political study, and that my revisionism coincides with a sudden shrinkage in these factors. If that shrinkage compels us to develop less costly means to the same ends, it will be a blessing in disguise.

A second practical advantage involves access to the subjects of study. Samples of macropolitical units are always likely to be poor and highly uncertain in result: small in number (the population being small) despite the likelihood of a sizeable range of variation. If, as is usually the case, they consist of contemporary cases, they are also bound to be a badly biased sample for all cases relevant to general laws. These problems in principle are compounded by practical problems of access resulting from the political exclusion of researchers (as in Burma), the inaccessibility of subjects in other cases, the lack of local research facilities (for example, survey research organizations), and language problems for foreign researchers, among many other factors. As a result, contemporary comparative studies in macropolitics predominantly have one or more of three characteristics: (1) small numbers of cases chosen by intuition or for convenience; (2) the use, in wide-ranging studies, of readily available, aggregate data that are often quite untrustworthy and dubious indicators of traits they supposedly represent; or (3) think-pieces based on discussions of cases by ‘country-experts’ in light of a common framework that are usually not at all well coordinated. Crucial case study may, of course, also suffer from problems of access. However, since crucial cases rarely appear singly, the likelihood of being unable to study properly even one of them seems considerably smaller than the likelihood of working with poor samples in comparative studies or that of having to tailor theoretical research to practical possibilities rather than the far more desirable vice versa.

A huge practical problem in comparative research involves special knowledge of the cultures being studied. The arguments made by the German exponents of the view that the Naturwissenschaften and Kulturwissenschaften are ineluctably different, and the arguments of the clinicians inspired by them are surely right insofar as they hold that ‘social facts’, personal or collective, are embedded in widely varying, even in each case unique, cultural systems of meaning and value, and [Page 151] that one can neglect these only at great peril. Their position may not imply that ‘social laws’ are therefore unattainable, but they do imply that the cultural sciences impose a requirement of special cultural ‘understanding’ that does not exist in natural science. Crucial cases can often be selected to satisfy this requirement for individual researchers, and it is always possible to acquire a great deal of cultural Verstehen in the course of in-depth study of a case if one does not already possess it. Even if we reject the notion of a special inherent requirement for cultural science, this point can still be made to rest convincingly on the question of special language skills and special historical and sociological knowledge of cases, for lack of which comparativists are often justly criticized.

At the very least, one can obviously, if other things are equal, go more deeply into a single case than a number of them and thus compensate for loss of range by gains in depth: to that extent, at least, the clinicians have a foolproof case. In crucial case study, the advantages of traditional scholarship, as displayed in configurative-idiographic studies, can thus be combined with those of modem technique and rigour. And it is also more possible to apply in crucial case study certain techniques developed in social science for overcoming the imperfections of single measures, especially the ‘triangulation of imperfect measures’ techinique developed in social psychology and applied, impressively, by Greenstein and Tarrow (1971) in political socialization research. […]

At this point it should be clear that the practical advantage of crucial case study does not lie merely in resources. Case studies yield methodological payoffs as well. This is in large part due to the fact that they help avoid difficulties that are hard to reduce or abolish in cross-cultural research. Not the least of these are two related difficulties not yet mentioned: the problem of the proper cross-cultural translation of research instruments, a subject rapidly becoming a methodological field in itself and one that absorbs ingenuity and thought better devoted to theory construction and testing themselves. In addition, if we conduct crucial case studies, we are far more likely to develop theories logically and imaginatively, rather than relying on mechanical processing to reveal them. More important still, we are more constrained to state them tightly and in proper form, suitable to testing: that is, in a manner that permits their deductive and predictive application to cases. Sloppiness in the forms of theory compatible with the criteria developed in the section on ‘definitions’ above is not inherent in comparative studies (certainly not in the ‘disciplined’ variety), but crucial case study involves far more compelling practical demands for the proper statement of theories, or else exposes far more manifestly when theories are not properly stated: that is, when nothing - or a great number of different things - can be deduced for any case from regularity statements about it.

More thought, more imagination, more logic, less busy work, less reliance on mechanical printouts, no questions about sampling, possibly firmer conclusions (including that extreme rarity in political study, the conclusively falsified hypothesis), fewer questions about empathy: these surely establish a heavy credit. It remains to see whether any debits may cancel it out.

Objections to the argument and replies to the objections

A number of arguments that might be, or have been, raised against my brief for crucial case studies should now be considered. None seem unanswerable, except in ways not very damaging to the argument. But others may see more merit in the objections than in the ways they are answered - and it is also possible that the really telling objections have been missed, or subconsciously avoided to restrict discussion to those that can be answered. The most telling objection of all, of course, would be that crucial cases simply are not available in macropolitics, but that has already been ruled out as unlikely for most, or many, theories. It is true that the literature of political science is not rich in crucial cases, but neither does it abound in efforts to find them; the most likely reason is that the very idea of crucial case study is alien to the field.

Objection I

Comparative studies have the advantage over case studies of allowing one to test for the null hypothesis that one's findings are due to chance. Case studies may turn up validating or invalidating results fortuitously, not because theories are actually valid or invalid, but because one cannot determine by single measures whether or not this is so.

(a) The possibillty that a result is due to chance can never be ruled out in any sort of study; even in wide comparative study it is only more or less likely. Now, it is surely very unlikely that, out of all possible states of affairs (which normally are vastly more than the two faces of a coin and sometimes approach infinity), just that predicted by theory should fortuitously turn up in a case carefully chosen as crucial for the theory, and also improbable that in such a case the predicted result should, just by chance, be greatly out of line with actual observation. The real difference between crucial case study and comparative study, [Page 153] therefore, is that in the latter case, but not the former, we can assign by various conventions a specific number to the likelihood of chance results (for example, ‘significant at the .05 level’). Thus, if a theorist posits that democratic stability varies directly with level of economic development and inversely with rate of economic growth, and finds a case of extreme instability with high E d and low E r , he cannot rule out plain bad luck, but the presumption that he has an invalid theory surely is vastly greater.

(b) Any appreciable likelihood of unlucky chance findings in a crucial case study arises from the fact that very short-run fluctuations generally occur in any measure of a variable. (Think of air temperature, or rainfall, or the climate of a marriage.) For example, a polity generally high in performance will probably experience some peaks and troughs in its level, and peaks and troughs will also occur in variables used to explain levels of performance. If we then measure a dependent variable at a peak and an independent variable at a trough, a deceptive result will certainly be obtained. But the remedy is obvious: observe over a reasonable period of time. 6

(c) There is, of course, also a possibility of observer bias in the observation of a case (seeing only what one wishes to see), hence of misleading, if not literally fortuitous, measures. That problem exists also in comparative studies, but not so acutely because of the prophylaxis provided by statistical measures of significance. But again, simple remedies are available. The most obvious is to recognize that falsifying a theory is to be reckoned as success rather than failure, and thus to redefine what one generally wants; knowing what is valid tells one more than knowing what is not, but knowing something to be invalid does signify progress and often provides very powerful clues as to what is valid. It is true that the reward structure of the social sciences overvalues positive findings, especially in publication - which may be why methods that maximize the probability of some sort of positive result (for example, multiple regression) are so widely used. But such prizing of positive results, however tenuous, indicates scientific immaturity or insecurity and ought not to be perpetuated. It works like Gresham's law: bad theory crowds out good. (In fact, the question most frequently, and fearfully, asked about the preliminary version of this paper was: ‘What do you do if a prediction about a crucial case fails?’ Answer: You publish the result - if editors permit and the failure is informative, as it is almost bound to be - and you go on, trying to do better.) Apart from that fundamental point, the problem of observer bias arises more in configurative-idiographic studies than in the more rigorous varieties of case study - hence the stress on it in existing critiques of case study (Becker, 1968) - and can certainly be reduced in the study of collective individuals by the same methods used to reduce it in comparative studies, such as the correct sampling of the micro-units that constitute the case.

(d) In crucial case studies a powerful substitute for the null hypothesis can be put to work: testing for a theory's ‘counter-theories’, that is, likely alternative solutions if a theory is invalid, or a theory's ‘antithesis’, if one is available. (This can also be done in comparative research, but practical considerations make it more feasible in case study, especially if theory and counter-theories cannot be tested by exactly the same data.) The process of testing simultaneously for alternative hypotheses ('strong inference’) has been held persuasively by John Platt as the correct way to put Baconian empiricism to work and, also persuasively, as the hallmark of the most rapidly developing hard sciences (high-energy physics and molecular biology). […] A crucial case study can readily be designed not only to determine whether a case lies off a predicted point on one curve but also whether it lies on, or nearer, a predicted point on a crucial counter-curve. Since only one case is involved, the cost of doing both will not be much greater than that of performing one operation alone. Several advantages accrue. We may not merely establish that a theory is false but also why, at bottom, it is false, and what sort of theory would serve better. Furthermore, a finding near a predicted point on one curve but far off such a point on the counter-curve adds to one's case enormously. One may thus shed not only special light on one's theory, but also more general light on the more fundamental bases for further theory construction. And if both theories are confirmed, a false contradiction is exposed; if neither is, the same result is obtained, or sloppy deduction is unmasked. All this takes one far beyond the mere void of statistical nullness.

Objection 2

If preliminary comparative studies are required to identify crucial cases (for example, cases extreme on pertinent measures, or highly changeable on a measure, or having the characteristics of natural experiments), the practical advantages of case study are severely reduced. And if they are severely reduced, certain practical advantages of comparative studies, such as their ability to provide data for reanalysis or simply data from numerous contexts, tend to tip the scales in their favour.

(a) Independent comparative study is not always required to identify crucial cases, simply because in an ongoing discipline the evidence needed to identify a crucial case often is already available.

One may want to recheck that evidence or try to improve on it, but that is not tantamount to starting from scratch.

(b) Even if one starts from scratch, comparative studies specifically designed to uncover crucial cases can be very limited in scope, even confined to a single variable, and so much reduced in costs of all kinds. Common knowledge, no less than disciplinary knowledge, can also reduce problems of sampling in the search for such cases. For example, if the object were to discover a long stable polity, it would take more than ordinary ignorance to include, say, Germany in the search. Comparative studies to uncover crucial cases thus have little in common, in regard to required breadth of study or data requirements, with comparative studies as currently conducted.

(c) The fact that comparative studies provide ‘extensive’ data from many contexts can be offset by the usual claim for ‘intensive’ study: that it can provide more varieties of data (and is likely to do so not only if study is clinical but also if strong inference procedures are used). Such data, moreover, are as much subject to reanalysis as any others. They may not suit well the purposes of others, but then neither might those produced by comparative study; and virtually any body of data has import for a variety of purposes.

Objection 3

Several crucial case studies are always better than one. Some degree of additional safety is always provided by additional numbers. If, therefore, the intent is to be conclusive, crucial case study ends as comparative study anyway.

(a) The basic problem here again is the equation of success with confirmation. A single crucial case may certainly score a clean knock-out over a theory (as Galileo's thought-experiment would have, had it been a real experiment, and as the falling feather and coin in the evacuated tube later did). The problem arises only if confirmation occurs. Because distrust is a required element of the scientific culture, confirmation only eggs us on to allay our own always remaining doubts and disarm those of caping adversaries; and thus we may want to know whether a theory that fits crucial case X also fits cases Y and Z, assuming these are also crucial, and whether, despite precautions and great unlikelihood, chance has tricked us after all. But the further examination of other cases can be restricted much more than in comparative studies that rest their case on [Page 156] sampling, and in such studies ‘added confirmation’ may also be deemed advisable - in fact, is necessary if the studies merely establish curves rather than matching them.

(b) Conceivably, the most powerful study of all for theory building is neither the presently common from of comparative study (of cases studied randomly, or intuitively selected, or simply studied because they seem readily available or accessible) nor the study of single crucial cases, but, so to speak, ‘comparative crucial case studies’. The case for such studies, however, is strong only to the extent that the most crucial cases available are not very crucial, so that high confidence in the results they yield needs the increment of other crucial case findings. Thus, the feather and coin falling in a vacuum leave virtually no doubt to the sceptic or the inquirer devoted to the tested theory, while a case of change in governmental performance highly unlikely to be due merely to the disappearance of performance-depressing factors, as against the factor posited to be required for high performance, probably leaves enough doubt to both to make desirable a further study or two of equal import. The study of such a more tenuous case might also, in some instances, be considered an especially powerful ‘plausibility probe’, warranting the (costly) comparative testing of a probabilistic hypothesis like that logically implied in the congruence theory of governmental performance that ‘in all cases, the correlation between performance and congruence will be high’ (Eckstein, 1966, p. 282). The comparativist may treat any and all crucial case studies as plausibility probes, warranting the costs of using his favoured method. The notion of the crucial case study was, after all, devised largely from that of the plausibility probe. The point is that he need not do so, unless the crucial case falls far short of the ideal.

Objection 4

Crucial case studies turn out to be comparative studies in disguise. For instance, when dropping a coin and feather through an evacuated tube we take two simultaneous measures and compare them; or, when studying the correlates of a change from low to high governmental performance we again take two measures at different points in time and compare them. The distinction between comparative studies and case studies thus vanishes, along with that between the clinical and experimental modes of inquiry.

(a) Not all crucial measures are like that. Observing the deflection of light near the rim of the sun compares nothing with anything (unless it [Page 157] is claimed that it compares deflection with non-deflection). The same holds true if only high governmental performance, not change towards it, is the critical observation.

(b) It is by no means sophistic to maintain that the supposedly dual measures above are single measures, that is, measures of the amount of change in performance between an earlier and a later period or the amount of difference in the rate of fall of two objects. Such changes and differences can be used as points on a curve no less than measurements of static conditions at a particular point in time, and thus satisfy the exacting technical definition of case study.

(c) Measures of ‘more than’, ‘less than’ and ‘equal to’ do presuppose two anchoring measures (see also Objection 5, Answer (d) below), but are not to be confused with comparative measures of samples, and N = 2 always suffices to establish them. Thus the distinction between case studies and comparative studies is watered down little, even if points (a) and (b) are disregarded.

Objection 5

Social science, especially on the macro level, does not have available measures precise and discriminating enough to make the sort of predictions needed for crucial case study.

(a) If this is true, the fact must bedevil comparative studies as much as any others, unless there is some magic by which many poor measures are equal to one that is good. Numerous poor measures can, of course, cancel one another out, or increase confidence in any one of them. But they can also make for increased distortion, that is, reinforce one another, and will certainly do so if a measuring instrument contains a consistent bias.

(b) What gospel ordains that social measures must be highly inexact and undiscriminating? That of experience? Perhaps; but perhaps only because of the prevalent assumption that nothing precise can be done in social study - surely a self-fulfilling prophecy if ever there was one. And while there is research there is hope: most of the natural sciences had to live long, and managed quite well, with rather imprecise measures too, and ours have been improving.

(c) Highly discriminating measures are not required to put crucial case study to work. If the measure to be predicted is, let us say, the level of democratic stability in postwar Germany, it is not necessary to be able [Page 158] to say that the level is at 112.56 […] with reasonable assurance that it is not then at 112.57. It may quite suffice to say that the measure must come out somewhere between 8 and 10 on a ten-point scale, either because theory permits or because of recognized possible error in a measuring technique. The possibility of disconfirmation then still exists, and is, after all, about four times as likely as confirmation. There must, of course, be a limit on imprecision. The minimum requirement is that measures must not be so inexact that any measure considered to validate (or invalidate) a theory could also, because of inherently possible measurement error, be taken to imply the opposite. If we cannot do much better than that in the social sciences, we might as well not measure at all, in any kind of study. Therefore, arguments about imprecision impugn quantitative social science, not crucial case study only.

(d) If no more than a single point is measured, however, crucial case study does presuppose interval measures (even if measurement techniques do not allow discrimination between minute intervals). If only ordinal measures are available, then (and only then) one must have two measurements to confirm, or invalidate, the prediction that a variable will have a higher, or lower, value at one time than another, or under one condition than under another. Ordinal measures only state ‘more than’, less than’ or ‘equal to’, and that always requires two points of reference, as stated above. And, as also already stated, this still concedes next to nothing to comparative study, and perhaps nothing at all if the predicted measure is interpreted as the measure of a difference of some discernible magnitude.

Objection 6

Crucial case studies cannot confirm multivariate theories, in which one deals with one dependent and several independent variables. The social sciences (especially on the macro level, where crucial case study is most advantageous) deal with multivariate phenomena: phenomena in which a variety of determinants converge upon observed experience.

(a) Again one wants to know: What gospel ordains that social phenomena must be multivariate, or decisively more so than any others? One might answer, the phenomena themselves: look, for instance, at all the factors associated with revolution, or authoritarian political behaviour, or political instability, or non-voting. True - but not decisive, and quite [Page 159] probably pernicious. In the natural sciences, too, ‘causes’ converge in phenomenal experience, but notable successes have been achieved in cutting through the phenomenal complexities to simple theoretical constructs that are powerful tools in explaining particular occurrences or, by engineering, for bringing them about. (For example, the law of velocity of freely falling objects consists of one dependent variable, velocity, and one independent variable, time, gravitational force being a constant; but actual ‘falling’ depends on many more factors, although some operate only with infinitesimal effects.) The problem of multivariate complexity largely dissolves if theory is thought of as a tool of explanation of the behaviour of concrete individuals rather than as total explanation. And the probable perniciousness of the assumption that theories must be multivariate if phenomena are resides precisely in the fact that then they will be, thus missing beautiful and powerful simplicities, even if they might be found.

(b) Multivariate theories do not necessarily rule out crucial case study in testing, provided that one does not simply list independent variables that affect a dependent variable (x has some relationship to a, b, c ,… n ) but specifies precisely the relationship of each to the dependent variable and their effects on one another. Newton's theory of gravity, for example, is multivariate: the dependent variable, gravitational force, is determined by two independent variables, mass and distance. But it specifies a direct relationship to one and an inverse relationship to the square of the other. Given the constant necessary to turn these ideas into an equation, predictions can be made for any case that may conclusively confirm or invalidate. The problem then lies more in the way multivariate theories are stated than in multivariation as such. The job of avoiding that problem is immensely difficult (most of us probably need not apply) but it ought to be tackled, even though here again the prevalent reward structure of the social sciences discourages the attempt to do the better work that is more likely to fail, or to be perceived as failure.

(c) A real problem is that a case finding may be the result of complex ‘interaction effects’. The careful choice of a case may allow one to discount that as a probability, but never altogether, and the fact that the problem might also queer comparative findings (more factors are nearly always interacting than a research design takes into account or allows one to separate) does not abolish the difficulty. The only sensible response is to treat the possibility as reason for continued doubt of some magnitude or other, and thus for further research. If the findings confirm a theory, that simply implies that one might want additional assurance in another pertinent instance. The point here is exactly the same as that regarding the possibility of ‘chance’ results. If the findings disconfirm, and one has strong prior reasons to consider a hypothesis valid (for example, because of various sorts of estimates of plausibility), [Page 160] the sensible course is simply not to give in all at once but to try another crucial test. In neither case is comparative study the required solution. The responses simply involve added confirmation, or added disconfirmation, by further crucial case study. And studies of additional cases for added assurance are not, strictly speaking, ‘comparative’ studies.

Objection 7

Crucial case studies cannot test probabilistically stated theories.

(a) Agreed.

(b) Theories need not be probabilistic, and the more powerful are not, even if the occurrence of phenomena is. Here, once again, the difficulty lies in confusion between theory as a tool of explanation and theory as the full explanation of concrete events. (That confusion is especially reprehensible in this case because probability statements, inherently, are not total explanations either.) The position rests also on two further fallacies: that if something is true probabilistically of a numerous set of cases, then the probability of its being true is equal for each individual in the set (which is true only in rare cases, like tossing fair coins); and that no mere probability can be deduced from a ‘law’ (it can, to the extent that the conditions under which a law is supposed to hold absolutely do not in fact exist, or to the extent that a law treats variables as constants).

(c) Probability statements are used more often than they need be in political study because of the uncompelled belief that they must be, which works, like other methodological assumptions, as self-fulfilling prophecy.

Objection 8

Even if all these objections are answerable, it is highly suspicious that so many should arise. Case study seems more susceptible to challenge than comparative study, in regard to which most of these problems are not even raised.

The essential difference here is not the volume of issues, but that the issues differ because the methods differ. Moreover, comparativists have [Page 161] only recently begun to raise important difficulties inherent in their method, especially on the macro level. But, in a relatively short time, an impressive number of difficulties in the method have turned up. After all that has recently been written about difficulties in comparative cross-national study, or even in micro-level studies - problems concerning the selection and proper number of cases, the feasibility and trustworthiness of research instruments (like survey research schedules), the comparability of data, or the utility of various data-processing techniques and modes of inferring regularities from numerous data (for example, various types of significance tests, attributions of causal paths to correlations, attributing longitudinal characteristics to synchronous data) - it is impossible to take seriously the position that case study is suspect because problem-prone and comparative study deserving of benefit of doubt because problem-free.

Case study in macropolitics begins in idiography and is rooted in the traditional conception of clinical study. In recent years the position that case study cannot be ‘nomothetic’ has been increasingly attacked in psychology, the very field that made the distinction between idiographic and nomothetic study sharpest and most insuperable. But the notion of nomothetic case study has not been taken far. If not conceived as the application of established theory to case interpretation, it has merely been represented as case study in which rigorous methods, similar to those of ‘experimental’ study, are used and/or in which individual experience is used to help find clues to general theories. If more has been claimed, as by Chassan, it has turned out that the term case study (’N = 1’ study) is indefensibly applied, by confusing a case with a concrete person rather than a measure (see Davidson and Costello, 1969).

My object has been to take the argument for nomothetic case study far beyond this point, following up clues provided by examination of more modest arguments in favour of it. The point has been to relate ‘N = 1’ studies to all phases of theory building and particularly to stress the utility of case study where rigour is most required and case studies have been considered least useful. Comparative studies have not been attacked, except on practical grounds in limited fields of inquiry; nor is it claimed that appropriate case studies are always available for all theoretical purposes, and absolutely not that any kind of case study will serve all purposes. The types of case study are numerous, and that recommended for going beyond formulating candidate-theories is extremely rare in [political science] or related disciplines.

The argument thus is mainly abstract. There is no track record worth mentioning. But if the horse is run, the results just might be [Page 162] astounding - or, possibly, abysmal. The point is that trials seem in order, not in place of but alongside comparative researches.

It should be evident that case study can be nomothetic only if cases are not selected for the theoretically trivial reasons that nowadays predominate in their selection: because one knows the language, finds a culture congenial to live and work in, can get money for study in it through an affluent area programme, considers the case important for foreign policy or otherwise publicly marketable, finds it exotic, and the like. Considerations of congeniality or publicity are well and good if other things are equal, not otherwise. And not the least advantage of crucial case studies is that they may permit one to study intensively attractive or convenient cases without sacrifice of disciplinary conscience.

1 The power of theories can also be assessed by another criterion, not included in the text because it strikes me as something for which theorists generally hope, rather than something at which they consciously aim. This is ‘deductive fertility’: not just unexpected knowledge but knowledge in unexpected areas, that is, reliable and valid accounts of observations outside the original fields of interest. That criterion can, of course, be consciously pursued when one asks whether a single regularity statement can account for observations that several separate ones cover, or whether the separate regularities can be deduced from a higher-order rule.

2 The foreknowledge criterion also seems pivotal for economic theorists. Thus, Friedman writes: The belief that a theory can be tested by the realism of its assimiptions independently of the accuracy of its predictions is widespread and the source of much of the perennial criticism of economic theory.’ It seems pivotal as well in contemporary philosophy of science, for the emphases on the deductive elaboration of propositions and parsimony are mainly attributable to the stress on non-intuitive foreknowledge (that is, foreknowledge that is not prophecy or clairvoyance, but rigorously deduced from ‘rules’) as the crucial test of theories. I agree with these positions, except for holding that the ends sought do not manifestly require a unique form of theory. For a somewhat different ‘hard’, but not over-demanding, view of theory - also based on natural science models but making central ‘generality’ (that is, range of applicability) and parsimony (that is, the number of factors needed for complete explanation of a class of events) - see Przeworski and Teune (1970).

3 For bibliography, see Holt (1962, p. 402).

4 If the essence of disciplined-configurative study is its application to cases of pre-established theories or to tools for building theories, we can distinguish four subspecies of this type of case study. They are quite different and need to be differentiated.

(1) Nomological case studies , as they might be called, are studies of the sort Verba has in mind: studies that interpret cases on the basis of theories considered generally valid.

(2) Paradigmatic case studies involve the application to a case of a pre-established framework, or checklist, for analysis, such as Almond's functional framework or the decision-making framework used in some studies of foreign policies. In (3) methodical case study rigorous methods associated with experimental study are applied to the study of individual cases, as in the studies reprinted in Davidson and Costello (1969). In (4) therapeutic case study validated theories and rigorous methods are used to diagnose problems and difficulties [Page 163] and to arrive at likely ways of eliminating or reducing them. (The labels used here are mine and arbitrary; they seem apt for the types.) The first subspecies is uncommon in all the social sciences, except economics, for reasons mentioned in the text. The third is common in psychology, the fourth more common in psychiatry than in other human sciences. In political science, the second subspecies predominates, exceeded in frequency only by configurative-idiographic studies.

5 These statements do not imply that acting on dues provided by empirically drawn curves is foolish. Usually it is wise - for example, not smoking if one wants to keep one's health - and often nothing better is available for making prudent decisions. The preeminent function of statistical analyses, as Wallis and Roberts emphasize, is in fact to help one to make wise decisions in the face of uncertainty. Hence such analyses are best used to help cope with problems of action (for example, traffic control problems, public health problems, problems of increasing agricultural yields, and the like) in which valid theories provide no better guide than plain statistics and in regard to which common sense is inadequate. As the field of statistics has become more and more powerful for this purpose, statistical findings have become increasingly confused with theories, but the distinction remains.

6 What constitutes a reasonable period of time for observations cannot be specified even in general terms. It depends on what one is studying, and it can generally be determined only by a combination of reasoning and reflection on findings. ‘Reasonable’ time spans make it unlikely that findings will be distorted by fortuitous short-term events. For example, I have argued, upon reasoning, that valid measures of performance require observations of polities over about a ten-year period at the least. Gurr and McClelland in Political Performance (1971), an empirical follow-up study, suggest, upon evidence, that a shorter time span might be serviceable for certain measures of performance.

From H. Eckstein (1992) Regarding Politics: Essays on Political Theory, Stability and Change . Berkeley: University of California Press.

Case Study and Generalization

Case and Situation Analysis

Sign in to access this content

Get a 30 day free trial, more like this, sage recommends.

We found other relevant content for you on other Sage platforms.

Have you created a personal profile? Login or create a profile so that you can save clips, playlists and searches

• Sign in/register

Navigating away from this page will delete your results

Please save your results to "My Self-Assessments" in your profile before navigating away from this page.

Sign in to my profile

Please sign into your institution before accessing your profile

Sign up for a free trial and experience all Sage Learning Resources have to offer.

You must have a valid academic email address to sign up.

Get off-campus access

• View or download all content my institution has access to.

Sign up for a free trial and experience all Sage Learning Resources has to offer.

• view my profile
• view my lists