To refer to this article use this url: http://www.gjss.nl/vol03/nr02/a04

---

GJSS, Volume 3, Issue 2, 2006

The Highs and Lows of using Case Studies in Student Research Projects: practical problems versus dynamite data

Kim McKee

Abstract

The case study is a key tool for qualitative researchers, facilitating the in depth exploration and understanding of participants’ views, experiences and perspectives. It can provide rich contextual detail, generate bottom up in vivo concepts, and help unravel the layers of complexity that can obscure research questions. A cheap and manageable option especially for student researchers, case study research is not however problem free. It is a time consuming approach, which relies on good relations between the researcher and their participants, an organised approach to time and work management, and a persistent approach to securing access to potential cases. Above all, the onus is on the researcher to provide a transparent and reflexive account of their methodological approach so that the reader might be convinced of their arguments.

Based on the author’s personal experience of case study research via current doctoral research on the Glasgow housing stock transfer, this paper aims to discuss and illustrate both the merits of adopting a case study approach and also the potential practical problems and difficulties that may be encountered.

Introduction

Case study research has been popular across the social science disciplines and throughout the ages, memorable twentieth century examples of which include classic sociological studies such as Whyte’s Street Corner Society, Holdaways’ Inside the British Police Force, Patrick’s A Glasgow Gang Observed, and Fielding’s research on the National Front (Hammersley and Atkinson 1995). Yet whilst case study research has proved the basis for some of the most in depth, detailed and vivid accounts within the social sciences, especially within the qualitative tradition, it remains a disparaged research method that takes second position to the more elevated and rigorous statistical approaches of the quantitative tradition. This imbalance, and indeed the criteria on which case study research is judged is one that needs addressed. The purpose of this paper is therefore to discuss the highs and lows of case study research drawing on my own live experience as a doctoral researcher studying changing governance relations following the Glasgow housing stock transfer in 2003. This is not the place for an in depth discussion of the findings of my own research and indeed is not necessary to understand the discussions that follow (such discussions do however exist elsewhere – see for example McKee 2006).

In order to provide practical advice and a realistic account of ‘doing’ case study research this paper will address the following issues: selection of the case(s) and negotiating access; field relations, including data collection, relations with gatekeepers, and ethical concerns; and the organisation, management and analysis of data, with an emphasis on producing a reflexive and documented account of the research process. Whilst case studies are a popular method within the social sciences (especially amongst students who do not have vast financial resources or teams of researchers at their disposal) practical advice on conducting them is of variable quality, comes with different theoretical baggage and is spread across a number of disparate sources. Therefore I hope this paper can serve as an orientating device and provide practical advice for those new to this research design.

Case Study Research: the verdict so far

Case study research has been presented by some commentators as a distinct research method in itself (Yin 2003). Yet as Stake (1998) proposes, it is perhaps better considered more of an approach that poses the epistemological question: what can we learn from the single case? As a heuristic tool, and conscious element of research design, the case study draws on a variety of data and mobilises a plethora of research methods from participant observation, through documentary analysis and in depth interviews, to statistical surveys. Although often confused and conflated with qualitative methods in general, case study research need not be non-quantitative and indeed can adopt a mixed-methods approach (Yin 2003; Platt 2006). Having been in decline from its height in the 1950s and its elevated position within the Chicago School of Sociology (Mitchell 1983; Stoecker 2006), the revival of qualitative research in recent decades has also seen the increased popularity of the case study (Platt 2006). Whilst it has long been a staple research tool across the social sciences, as Platt comments its precise use and definition has varied within different disciplines, making it a slippery concept to define:

…writers tend to have in mind, even when making quite general statements, the sort of case and the sort of method most salient in their own intellectual settings. Psychologists think first of individuals in treatment, anthropologists of events in whole small societies, and political scientists of polities (Platt 2006: 277).

The precise form of what constitutes a case(s) may alter, but what unites competing definitions is a focus on the unique, the particular and the specific within a bounded real-life context (see for example, Stake 1998; Yin 2003). Whilst the case may have an ‘instrumental’ purpose: simply an illustration of a more abstract phenomenon, it may also be ‘intrinsically’ interesting in its own right (Stake 1998: 89). A familiar aspect of research and teaching within the social sciences the case study remains however discussed in disparaging terms: ranging from a ‘second best practice’ and little better than ‘journalism’ (Stoecker, 2006: 325-7) to the more pleasant jibe of an ‘attractive nuisance’ (Miles, 1979: 590).

Critics of the case study have focussed primarily on issues relating to bias and the lack of generality (not to mention the time, costs and workload involved particularly for lone researchers). With regards to the former, concern is expressed about the lack of distance between the researcher and the researched, especially the possibility of the researcher going native (Stoecker 2006); this problem is of course grounded in the belief that objective and neutral scientific research is both an achievable and desirable goal. In terms of the latter the common question posed is how can we infer more generally and answer questions about external validity? As Mitchell (1983) argues, only probability samples meet the high standards of representative criteria; consequently the case fails to attain this level of rigour as Stoecker elaborates:

These problems stem from the ‘N of 1’ problem – that there is only one case and, therefore, objectivity is difficult to maintain, falsifiability criteria are more difficult to meet, and generalization is impossible…The attempts to meet the standards set by quantitative scientific sociology do little more than emphasize the inability of case study research to live up to those standards (Stoecker, 2006: 327-329).

Yet if case studies were as limited and useless as some commentators would have us to believe surely they would not have survived across the decades to become a mainstay of social research, particularly in the qualitative tradition. Not only do case studies offer the potential for rich, detailed, persuasive accounts (Miles 1979; Miles and Huberman 1994; Platt 2006), but the critics who lambaste their lack of representativeness would do well to consider that the laws of statistical sampling are irrelevant and of little concern in the qualitative tradition where purposive sampling and the advancement of theory reign supreme (Silverman 2005). As Stake remarks: ‘the purpose of the case study is not to represent the world, but to represent the case’ (1998: 104). This position is reinforced by Mitchell (1983) who stresses it is the strength and quality of the analysis that matters most when making the creative link from the one to the many.

What is clear in this muddled and divided debate is the case study confers both potential advantages and also inherent problems as compared to other research designs. Drawing on my own research experience I will aim to take these debates forward and further discuss the highs and lows of using case studies.

Selecting the Case and Negotiating Access

Selecting the case(s) to be studied is a critical event when conducting case study research, yet this difficult process is unaided by the fact proponents of case study research offer conflicting advice on correct procedure. For example, whilst Yin (2003) perhaps the leading authority on case study research advocates the extensive screening of cases through an intensive piloting process involving data collection, this has been challenged by proponents who argue the ‘opportunity for learning’ (Stake, 1998: 101) and the ‘explanatory power’ (Mitchell, 1983: 203) of the case are of greater importance. Here ease of access and hospitality are elevated as more important attributes than whether the case is typical of some wider population (Stake 1998), especially as inference relies upon a sound theoretical framework opposed to statistical significance and random sampling:

We infer that the features present in the case study will be related in a wider population not because the case is representative but because our analysis is unassailable (Mitchell, 1983: 200).

This does not imply we should select cases on the bases of simple ease and convenience but that we should keep in mind practical considerations whilst also thinking critically about the wider population from which our case is to be drawn. Employing purposive or theoretical sampling may be useful here. What differentiates these latter two concepts is the presence of a theoretical rationale at the outset, yet both encourage us to select cases on the basis that they possess a feature or process that is of interest to us and will help answer our basic research questions (Mitchell 1983; Silverman 2005; Stake 1998).

In my own research into how housing governance arrangements have changed in Glasgow following the city’s housing stock transfer^[1] in 2003, case selection was initially a daunting prospect. Following the transfer of ownership of the housing stock from the city council to the newly formed Glasgow housing association (GHA), sixty Local Housing Organisations (LHOs) were established across the city responsible for delivering housing services on a daily basis through a delegated management structure (for further discussion see, McKee 2006). The LHOs offered a vehicle for understanding the dynamics of change post-transfer: not only could they provide insight into housing governance at the local level, but through their relations with the GHA and other external agencies could also enhance my understanding of relations of power at a citywide level. However, as there were potentially sixty organisations from which to choose narrowing down my options was a difficult first step. Consequently I opted for an ‘exemplary case’ approach - providing a strong and positive example of the phenomenon of interest (Yin, 2003: 13).

I achieved this by concentrating my attention on the high performing LHOs already involved in a pilot programme to devolve ownership as well as tenant control to local communities (see table 1 for details). Those involved in the pilot had to meet a range of criteria to participate in the process: meeting key performance indicators; having established governance/tenant participation mechanisms; and possessing housing stock with a long-term sustainable future.

By selecting my cases from the pilot programme my options were quickly narrowed from potentially sixty organisations to seven. Within the seven, my desire was to introduce variety by incorporating the two main types of LHO: those operating in partnership with existing organisations, and newly formed organisations that had emerged out of the city council’s tenant participation strategy^[2] (see table2 over leaf).

As table 1 indicates this selection process further narrowed my options, as there was only one newly formed organisation involved in the pilot, therefore making it an automatic choice. From the other six, I tried to select an organisation of a broadly similar stock size and which operated in a different area of the city (using the GHA’s own crude division of areas into North-West, South and East). The size of organisation is a key factor as it can influence the degree of tenant involvement, a critical issue I was interested in finding more about; selecting a different area was simply to avoid the research being skewed by localised factors. This narrowed the choice again to two potential options, with the final decision driven as much by practical concerns relating to access and sponsorship as to sampling decisions.

The next challenge was selecting a comparator organisation out with the GHA and LHO network, in order to allow a judgement to be made about whether changes in housing governance were in fact due to the stock transfer. The difficulty I faced was two-fold: there were few established community housing organisations operating in the city out with the GHA context, and my LHO cases were large in size compared to the majority of Glasgow’s housing associations: sixty-five percent of which have less than 1000 units^[3] (Communities Scotland 2005). To narrow my options I referred to annual statistics produced by the housing regulator Communities Scotland. After selecting out those organisations based in Glasgow, because of the large size of my pilot organisations I preceded to rule out those with a stock size of less than 1000 units; using information supplied by the GHA I then ruled out those housing associations who were LHOs which left me with four potential options.

Much time (too much time in hindsight) was spent deliberating between these four options for a comparator. Statistical information was collected in order to compare their characteristics with my LHO cases; two were ruled out immediately because a) they had no community basis, but had a dispersed stock across the region and b) because they operated in an affluent area with little council housing. Of the two cases that remained either would have been a viable option, with the final decision instigated by practical issues: I opted for the organisation from which I received the most positive interest and timely response to my invitation to be involved in the study. What would have been disappointing would have been to spend so much time selecting the best case only to have the door slammed in my face as it were and access denied. Practical concerns therefore deserve as much attention as methodological or theoretical issues. This brings us back to Stake’s (1998) advice that what is important is not having the most typical case study, but the one that offers the greatest potential for learning:

My choice would be to take that case from which we feel we can learn the most. That may mean taking the one we can spend the most time with. Potential for learning is a different and sometimes superior criterion to representativeness. Often it is better to learn a lot about an atypical case than a little from a magnificently typical case (Stake, 1998: 101).

Even when cases have been identified along purposive, theoretical or practical grounds gaining access is no easy task and can itself be a protracted and difficult process, which requires much patience and diplomacy (Hammersley and Atkinson 1995). This is especially true of student research projects, and in my own case I had few professional networks to draw upon on in order to sponsor or assist my access to the organisations I was keen to observe. The reluctance of organisations to become involved in social research is perhaps not unexpected – senior staff will be concerned that the organisation may be presented in a negative fashion, and there may be little in the way of return for the hours of (unpaid) staff time that may be required to satisfy researcher’s lines of enquiries, or the inconvenience of having them ‘hanging about’. Here impression management may help. As Hammersley and Atkinson (1995) emphasise, potential research participants are more likely to be concerned about the type of person a researcher is than what the research is actually about; thought should therefore be paid to initial introductions in terms of demeanour and dress (see also Silverman 2005). Yet even when initial access to a private setting is granted issues remain – consent is not a one off process but needs to be continually renegotiated during the life of the project. Physical presence in the setting does not guarantee the availability of all the people you may wish to speak to, all the records and documents you want to read, or all the situations and settings you wish to observe (Hammersley and Atkinson 1995; Silverman 2005). In my own research, the key contacts in all my cases were with senior management. Not surprisingly front-line staff were often sceptical of my presence in the setting and were initially guarded when speaking to me: concerned their comments would be relayed to management. Eventually my extended presence in the research setting and efforts at sociability convinced participants I was a ‘reasonable person’ and trust was developed. These relations however take time and trust to be built.

In the Field

A further important facet of case study research is to provide a documented and reflexive account of the choices made during the research process, or what Yin identifies as a ‘chain of evidence’ (1981: 63). Qualitative research is not experimental in nature and therefore not easily replicable: the path one researcher would take is not necessarily that which would be adopted by another. Yet by making clear what was done, how it was done and why, this enhances the readers’ trust in the analysis and provides helpful guidance for other researchers who may be entering the field with similar data of their own (Silverman 2005; Seale 1999).

How I collected my data

Each of the case studies was visited approximately two-three times a month, normally for a half day, with regular email contact also maintained with the key informant. In total field contact for the case study phase lasted approximately six months from August 2005 -January 2006^[4], although as previously indicated several months were spent making contact and negotiating access before fieldwork formally started, and even after it ended good relations maintained. Within the case studies a range of qualitative methods were employed (see table 3 overleaf). The purpose of this was not ‘triangulation’ in the sense of corroborating participants’ meanings and understandings (Yin 2003), that is, there is no single fixed reality to which all the data refers, rather it was a means of gaining insight into how the same phenomena can be interpreted and understood in different ways, and ensuring the inferences made of the data valid (Hammersley and Atkinson 1995).

Prior to the commencement of fieldwork a quota guide was drawn up and consideration given to the types of individuals or groups who needed to be interviewed (see figure 1 over leaf). Furthermore as both table 3(above), and figure 1 (over leaf) indicate, the case study research was complemented by external key-actor interviews, observation of events and documentary analysis; this was to facilitate analysis at the citywide level as well as locally within the LHOs. The actual number of interviews etc carried out was dependant on how quickly ‘saturation’ was achieved (Strauss and Corbin 1998), however this was mediated by practical issues such as access and available time/resources.

For transparency purposes it is also worth elaborating here on the way particular research methods were employed in relation to the research design:

• Interviews – interviewing is by far the most popular research method employed in the social sciences, yet this is not to under estimate the importance of simply talking to people and placing their personal accounts at the centre of the research (Legard et al 2003). In this design the interviews were of the semi-structured variety and a topic guide was used to ensure some comparability across the cases. An effort was made to make the interview as ‘naturalistic’ as possible (i.e. like a normal conversation) in order to put participants at ease. Yet as Legard et al (2003) emphasise the success of interviewing as a research method hinges greatly on the personal and professional capabilities of the researcher; a skill that only comes through practice and experience.
• Observation - was a useful complement to the other research methods because when considering relationships between different actors, non-verbal communication can be just as illuminating as the detailed and rich responses constructed during the interview process. To use one case as an illustration, despite interviewees presenting staff-committee relations as generally positive observation of the meetings suggested otherwise. Whilst no formal seating plan was in place participants continually sat in the same seats next to the same people, with a clear divide and physical barrier of empty chairs between staff and tenants. Although it is logical to assume that my presence may have influenced the way individuals behaved at these meetings (whether for the better or worse), it was hoped my extended stay within these organisations would enhance participants’ familiarity with me, and my ‘observer’ presence forgotten. Indeed, I did succeed in building a good rapport in all my cases resulting in amongst other things, me being invited to Christmas parties and lunches, with this informal contact also written up in field notes and coded as ‘data’.
• Documentary analysis – qualitative content analysis was employed in order to identify key themes and facilitate interplay between conceptual frameworks, data collection and analysis; unlike quantitative content analysis which adopts a rigid coding structure this is a more reflexive approach (Bryman 2001).
• Tenant focus groups - without a doubt the most difficult aspect of the research. At all stages, whether it be their organisation, management or analysis I vastly under-estimated how difficult and time consuming these processes could prove. The most difficult aspect by far was getting tenants involved and interested enough to turn up. Whilst raffling a prize helped, it could not overcome the problem of small numbers with two of the five focus groups consisting only of four individuals, and one group in the comparator area having to be cancelled due to lack of interest. Yet there was also blissful moments where I could sit back and listen to participants debate the issues I was interested, and they would challenge and correct each other with little need for intervention or steering from myself.

Field relations: the importance of critical space?

The gate keepers I initially made contact with when trying to negotiate access to the organisations remained vital contacts throughout my fieldwork. As alluded to earlier, gaining access is not a one-off event but requires continual re-negotiation. These gate keepers facilitated this process whether this be encouraging other members of their organisation to give their time to be interviewed, or simply helping with the practicalities of having a quiet space in which to conduct the interviews and cups of tea on hand. Furthermore when the time came for me to leave the field and begin organising interviews with key actors at the citywide and national level, these gate keepers again proved invaluable in vouching for me in order I might secure an interview with an important but unknown stranger, and recommending to me people who would be worthwhile talking too. Whilst Hammersley and Atkinson (1995) warn of a potential conflict of interest between the researcher and their sponsor/gate keeper, my own experience was much more positive. Once access to the organisational setting was secured, all requests for access to particular individuals, settings and documents were granted and no obvious blocking or steering was encountered.

Critics of case study research (and qualitative research in general) often denounce these close personal relationships and rapport as a lack of scientific distance. Yet this ignores two critical factors. Firstly, neutral research may not be an attainable or even desirable outcome of qualitative enquiry: the social world is not something out there waiting to be measured as the researcher is ultimately part of that which they are investigating. Secondly, it is precisely because of these personal relations that participants’ respond to the researcher in the way that they do (Stoecker 2006). Here the ascribed characteristics of the researcher are important, and certainly in my own case I felt these worked to my advantage. Firstly, as a young, female student my presence on site was non-threatening and my endless questions tolerated and even expected; indeed many participants commented to me that they had children or grandchildren my age at university and were keen to help me in my studies. One advantage of being a young aspiring academic is that it is easier to adopt the role of an ‘outsider’ or ‘incompetent’ necessary for understanding and asking questions about an unfamiliar organisational setting; the relations I was able to strike up with my participants may have been very different to that of a middle aged professor for example (Hammersley and Atkinson 1995: 97). Secondly originating from the West of Scotland and having been a tenant of social housing all my life, as well as a university graduate with an interest in housing, I had common interests and talking points with the participants I was interviewing and observing whether they be local residents or housing professionals.

Whilst these inter-personal relationships are important for the success of not only the current research project, but also potentially future projects, it is important to retain some element of critical space in which to conduct analysis. Spending so much time with my local level case studies it was near impossible for me not to empathise with their plight post-transfer and the powerless position they felt they were in, but I was also mindful of the need to maintain some level of distance and be open minded to the views and experiences of other key groups and actors involved in the stock transfer process. Here the second external phase of the research proved a useful balance and complement. Yet it is not going native itself that is the problematic process here, for example in my own research my close relations with local actors resulted in me being invited to amongst other things lunches, Christmas parties and other social occasions all of which provided rich, detailed insights that would not have been accessed by straight forwarding interview techniques. Rather it is being caught unaware and not recognising that this process of adopting participants’ perspective is happening which can have detrimental effects on the research; maintaining some kind of balance in terms of the range of stakeholder involved is therefore critical in order to ensure multiple views on the topic as opposed to over identification with one particular group.

The importance of ethics

Glasgow’s stock transfer at the time of data collection was a very politically sensitive issue with different stakeholders having a lot invested in the process and not all of them entirely happy with the outcomes so far. As the focus of my research was ultimately relations between different stakeholders both at the local and citywide level within Glasgow’s social housing scene, care had to be taken to avoid inflaming what were already tense relations. In such a context attention was given to issues relating to confidentiality and anonymity, especially as some respondents (particularly external key actors) expressed concerns about being directly quoted, with one organisation even refusing to have the interview on tape. The decision was therefore taken to not use individual names but only to refer to the organisation from which the individual was from (e.g. a senior council official was simply referred to as senior representative, Glasgow City Council). In smaller organisations where a lean structure made individuals easily identifiable the organisation’s name was also omitted and a descriptor used (e.g. Housing association/tenant training and support agencies). I deemed this commitment to confidentiality and anonymity as a reasonable trade off for rich, insider accounts which individuals seemed more happy to given following assurances their name would not appear in print. As Stake emphasises researchers are ‘guests in the private spaces of the world’ (1998: 105), and therefore must have a strong sense of ethics and good manners to avoid harm for participants: a real prospect given reputations, relations and potentially jobs were on the line. All participants were consenting adults and once the interview began, they had ultimate control over what information they wanted to divulge and what they wanted to remain private. As a request was made to tape-record all interviews (which was denied on only two occasions) all participants were asked to sign a consent form to confirm that they understood what their involvement in the research required and that they consented to their information being quoted. To further reassure participants’ copies of transcripts were offered to allay in any concerns, allowing them to further block any comments they wished to remain private (which none of them did incidentally). I considered the potential loss of direct quotes a lesser evil than not having their insightful comments to draw on at all. Furthermore at the local level draft copies of interim findings were offered to key contacts in order to have their input into the accuracy of the description and interpretation of settings, events and relations. This not only positively enhanced field relations by building trust, but also provided further useful data in terms of the comments received. Whilst participant validation can be a difficult and nerve-wracking process it is not something to be feared or shied away from. In fact it is a useful test to determine just how grounded researchers’ accounts are.

Organising, Managing and Analysing the Data

The huge body of field notes collected from observations and documentary analysis, coupled with the mass of transcripts generated from the audio-recordings of the interviews and focus groups were extensive. The ‘time cost’ here is important (Miles, 1979: 592): the only way to record qualitative data is through extensive field notes and the transcription of tape-recorded interviews both of which involve laborious processes to transform the raw data into a form ready for analysis. Therefore when I was not collecting data, I was either transcribing it, summarising it or coding it. Miles and Huberman (1994) estimate this stage takes two to five times longer than data collection; if anything this is perhaps an under-estimation. Overload must therefore be guarded against at all times by a proactive approach to work management.

Yet this time and effort does pay dividends: the positive adjectives applied to qualitative data and research are all true. It does facilitate rich, insider accounts with an emphasis on context, process and the preservation of participants’ voices.

Qualitative data are sexy. They are a source of well-grounded, rich descriptions and explanations of processes of identifiable locations…good qualitative data are more likely to lead to serendipitous findings and to new integrations; they help researchers get beyond initial conceptions and to generate of revise conceptual frameworks. Finally, the findings from qualitative studies have a quality of ‘undeniability’ (Miles and Huberman, 1994: 1).

To avoid becoming over whelmed by the sheer mass of data two key principles of data management were adopted at the outset of the project: generating contact summaries, and entering the data into the Computer Assisted Qualitative Data Analysis Software (CAQDAS) package Nvivo (see for example, Richards 2000). Firstly as Miles and Huberman (1994) advocate, a field contact was written up following every interview, field visit, document reading or focus group. This provided an immediate summary of the emergent issues and themes as well as a description of what was actually happening, and proved a key tool not only in guiding planning for the next contact but also for enhancing familiarity and sensitivity with the data. Secondly, from the outset of the project all data was entered into and managed electronically from within Nvivo. Given the sheer volume of paper collected this was a critical time and space saving device, allowing data to be easily and quickly coded, retrieved and manipulated. Whilst CAQDAS packages cannot replace the analytical and creative work of the researcher, they can aid the analysis process in terms of data reduction, organisation and management.

Analysis is not a separate phase of the research that begins when all the data is gathered but must occur simultaneously so that it can shape and guide data collection in an iterative process, energising fieldwork and avoiding the scenario described below.

…coding is hard, obsessive work. It is not nearly as much fun as getting more good stuff in the field. Trying to do coding all at one time tempts the researcher to get sloppy, resentful, tired and partial. This attitude damages the robustness of the data and the quality of the analysis (Miles and Huberman, 1994: 65).

Data was coded as soon as it started coming in, vital given the volume and type of research methods being employed. A small list of about half a dozen pre-conceived codes were derived from the broad research questions, with the overwhelming majority emerging from the data in a bottom up fashion some of which were in vivo in nature (e.g. tenants described tenant participation prior to the transfer as the setting of wish-lists). This allowed analysis to remain grounded, and for myself to gain a good understanding of what was actually happening. Codes were initially descriptive in nature, although as time progressed became more interpretive. Whilst some commentators propose case study research should be guided by theory from the outset (see for example, Mitchell 1983; Silverman 2005) my preferred approach is to get a basic grasp of what is actually happening before introducing and developing concepts. It is an approach to analysis influenced by grounded theory (Strauss and Corbin 1998), but not in a slavish uncritical fashion^[5]. Whilst CAQDAS can help with analysis, particularly in terms of data reduction and organisation, it is no substitute for the creative and interpretive thought processes of the researcher. It is therefore an aid, not a mechanical substitute, for reflective analysis.

The software provides no automatic coding process. It always remains the task of the ethnographer to exercise his or her intellectual imagination to decide upon the analytically relevant codes to be used. Conceptually speaking, therefore, the task of coding for microcomputer applications is no different from manual techniques (Hammersley and Atkinson, 1995: 198).

The coding structure underwent a process of constant revision with the definition and content of codes constantly revisited. I was further mindful not to get too carried away with this revision process and over prune my coding structure: the relational structure of the codes must always be maintained if they are to make any sense or serve a useful analytical purpose.

Figure 2 (above), and figure 3 (over leaf), highlight how the coding structure changed and ultimately became more sophisticated, from the outset of the project to the present day. The data became more saturated with concepts, and the coding schema (or nodes as they are called in Nvivo) became more differentiated and organised in a tree-like relational structure. The next step in analysis is to take these largely empirical and grounded codes and think about them more conceptually in terms of a key organising theme (Strauss and Corbin 1998) – this has been identified as fragmentation, and derived from this the sub-themes of: centralisation; the mobilisation/regulation of tenant involvement; ownership of assets.

Other key steps in the analysis process were the acts of memoing and the production of interim findings reports. The former was initiated at an early stage and represented an attempt to generate ideas and think more conceptually about my data as the excerpt below from the memo ‘ownership of assets’ illustrates (see figure 4). Such insights should be made immediately when they occur, with no need to produce them in polished form.

The latter (interim reports) were produced at the end of each key stage in the fieldwork timeline (e.g. case studies, external setting, tenant focus groups) and they proved a useful means of documenting initial impressions and addressing blind spots in the data collection and early analysis that perhaps needed revisited. The analysis process is therefore not limited to the coding and memoing that occurs within Nvivo, but is also embodied in the act of writing up. The traditional approach of separating a PhD into planning/literature review (year 1); data collection (year 2); and data analysis/writing up (year 3) is therefore narrow and unhelpful. They key to avoid stress, panic and being drowned in a sea of data is to start fieldwork as early as possible and begin analysis of it almost immediately (see for example, Hammersley and Atkinson 1995; Silverman 2005).

Conclusion

Whilst case study research requires a conscientious and organised approach to work management and can be a stressful research design given the potential danger for overload at every stage, it is nonetheless an extremely rewarding approach, which can produce excellent results. The richness of the data and holistic understanding of the context under study cannot be under-estimated, especially as cases provide in-depth insights into research problems that more superficial research designs can only aspire to. Perhaps the maxim to remember here is it is better ‘to say a lot about a little’ (Silverman 2005: 80); it is not surprising therefore that after a period of decline the resurgence of case study research is now firmly under way.

I would suggest the key to doing case studies successfully is four-fold: a) think critically about cases, whilst theoretical and methodological issues have to be balanced against practical constraints do not just select any case and hope it will suffice; b) when on site good field relations are critical: sociability and impression management can go a long way here; c) document all the steps involved in sampling, data collection, and analysis – a transparent account is necessary if readers are to be convinced of the strength and quality of the analysis; d) begin data analysis as soon as possible, delaying the inevitable will simply result in a veritable data mountain which will be very time consuming and disheartening to analyse all at once.

Whilst methodological textbooks and qualitative research methods papers such as this are a useful guide and orientating device prior to the beginning of the research, the only way to learn is through doing – therefore the skills discussed in this paper are best cultivated through practical application and research experience.

Notes

University of Glasgow, Department of Urban Studies, 25 Bute Gardens, Glasgow, G12 8RS, k.mckee.l@research.gla.ac.uk

Within housing, stock transfer refers to the sale of assets out of the public sector (e.g. local authority or Scottish Homes) into the private or independent sector (e.g. housing association or co-operative).

As per table 2 a third category of LHOs formed from amalgamating existing Tenant Management Co-operatives were discarded from case selection due to their small size.

This figure was arrived at by amalgamating the smallest three of the six size categories used by the housing regulator Communities Scotland; whilst not equidistant it is easier to work with existing data.

The tenant focus groups were conducted later in the year (March-April 2006) to avoid the dark, cold winter months and therefore maximise participation.

Grounded theory is an approach to analysis that emphasises the interdependence between data and theory, whilst there is no scope to elaborate here such discussions do occur elsewhere (Seale 1999).

References

Bryman, A. (2001), Social Research Methods. Oxford: Oxford University Press.

Communities Scotland (2005), Registered Social Landlord Statistics 2003/04. Edinburgh: Scottish Executive.

Hammersley, M. and Atkinson, P. (1995), Ethnography: principles in practice 2^nd edition. London: Routledge.

Legard, R. et al (2003), ‘In-depth Interviews’ pp. 138-169, in J. Ritchie and J. Lewis (Eds.). Qualitative Research Practice: a guide for social science students and researchers. London: Sage.

McKee, K. (2006), ‘Community Ownership: the Glasgow housing stock transfer’, conference paper presented at, Cities in City Regions, European Urban Regeneration Association bi-annual conference, University of Warsaw: Poland, 11-14 May 2006.

Miles, M.B. (1979), ‘Qualitative Data as an Attractive Nuisance: the problem of analysis’, Administrative Science Quarterly 24 (4), pp. 590-601.

Miles, M.B. and Huberman, A.M. (1994), Qualitative Data Analysis: an expanded sourcebook 2nd edition. London: Sage.

Mitchell, J.C. (1983), ‘Case and Situation Analysis’, Sociological Review 31 (2), pp. 187-211.

Platt, J. (2006), ‘What can Case Studies do?’ pp. 274-296, in M. David (Ed.) Case Study Research, Volume IV. London: Sage.

Richards, L. (2000), Using Nvivo in Qualitative Research. Australia: QSR International.

Seale, C. (1999), The Quality of Qualitative Research. London: Sage.

Silverman, D. (2005), Doing Qualitative Research 2nd edition. London: Sage.

Stake, R.E. (1998), ‘Case Studies’ pp. 86-109, in N.K. Denzin and Y.S. Lincoln (Eds.) Strategies of Qualitative Inquiry. London: Sage.

Stoecker,R. (2006), ‘Evaluating and Rethinking the Case Study’ pp. 325-348, in M. David (Ed.) Case Study Research, Volume IV. London: Sage.

Strauss, A. and Corbin, J. (1998), Basics of Qualitative Research: techniques and procedures for developing grounded theory. London: Sage.

Yin, R.K. (1981), ‘The Case Study Crisis: some answers’, Administrative Science Quarterly 26(1), pp. 58-65.

Yin, R.K. (2003), ‘Applications of Case Study Research’. London: Sage.

Notes

University of Glasgow, Department of Urban Studies, 25 Bute Gardens, Glasgow, G12 8RS, k.mckee.l@research.gla.ac.uk

• Amsterdam University Press i.s.m. Digitaal Productiecentrum