When you are sorting women into one pile and men into another pile, you are sorting cases. Use of the term ‘case’ in this context is not intended to imply that you are necessarily undertaking case study research. Rather, cases are employed in NVivo as a unit of analysis in a research study, and a study might involve cases of more than one type.
In the sociological and anthropological literature, a case is typically regarded as a specific and bounded (in time and place) instance of a phenomenon selected for study. … Cases are generally characterized on the one hand by their concreteness and circumstantial specificity and on the other by their theoretical interest or gener- alizability. (Schwandt, 1997: 12)
In the Environmental Change study, Barbara is a case, Charles is a case and Dorothy is a case. In this example, people are cases. Alternatively, you may be working in a project where policies are cases, geographical regions are cases, or less visible entities like mathematical theories are cases. This all depends on your research question, and it could also change as you analyse your data and realize that the case type you started with is not the case type
designing an nvivo database 51
you are currently interested in examining. You may initially be looking at individual people, thinking you’ll compare men and women, and later decide that instead you want to focus on friendship clusters and compare insular and open clusters.
Even if you are doing a single case study (a methodology), you still
might end up with multiple cases (a tool in NVivo).2 For instance,
if you are following a single Olympic athlete’s journey to the summer games, you might be investigating this one person’s experience, with each letter they write home to their family while they are in training being treated as a case. In this regard, when using NVivo you can easily sort the letters written to parents into one subgroup and the letters written to siblings in another and friends in another, so you can focus on a more detailed interpretation of the data (the content of the letters) and compare across these subgroups. It all depends on your research questions, and the data you collect to answer those questions. Your case structure in NVivo should follow your research design, not lead it.
Identifying your cases
Much has been written about selection of cases in qualitative work. Patton (2002) and Flick (2007) each give a thorough overview of the range of sampling and selection possibilities, as do a number of other qualitative authors. Case structures are often simple, but they can also be quite complicated. The main case unit(s) sometimes have illustrative cases embedded within them, for exam- ple, when a corporation is the case and one or more specific departments, or products, are illustrative cases within the study of that corporation (Yin, 2003). Alternatively, they might be layered, for example where schools, classes and the pupils in them are each treated as cases at different levels.
Yin (2003) warns to ‘beware’ cases which are not easily defined in terms of their boundaries – their beginning and end points. ‘As a general guide, your tentative definition of the unit of analysis (and therefore of the case) is related to the way you have defined your initial research questions’, and ‘selection of the appropriate unit of analysis will occur when you accurately specify your primary research questions’ (Yin, 2003: 23, 24). If your ques- tions and conceptual framework are clear, it should take only minutes to define the case type (what kinds of units you are using) and thence the cases for the study (Miles & Huberman, 1994). Even if you do not intend to com- pare subgroups, you will benefit from understanding cases and how they help your work in NVivo.
52 qualitative data analysis with nvivo
Cases in NVivo
A case is a core structural element in NVivo. Each case unites all the different components of qualitative and quantitative data you have about that entity, that unit of analysis, in one place. Cases are incredibly flexible entities in three ways that are important for you to understand at the outset.
First, you will be able to include just a single source, multiple sources, or portions of sources as data for each case. For instance, if you interview 30 people and as a result have 30 transcripts, generally these will be turned into 30 cases. Alternatively, a case could include several waves of data collection for a single person. You could also turn
portions of a single document into a case, such as all the contributions an individual
speaker makes to a focus group. Or you can use a combination of strategies. For example, some or all of your participants came to the focus group as well as complet- ing one or more individual interviews, so you want their cases to contain their contribu- tions to the focus group as well as their interviews.
Second, each case might include only one kind of data, such as text in participant transcripts; or it could bring together information in multiple formats – in addition to an interview, you have photographs taken by the participant and videos of family celebrations.
Third, you can combine related demographic and numeric (attribute) data with the text (or other qualitative data) for each case so that, for example, a flag such as male or
female is applied to all data for the person who is that case, regardless of type, vol-
ume, or how many sources it is spread over.
Thus, for each case in NVivo, you are able to take advantage of the software’s capacity to manipulate multiple data collection points, multiple formats of qualitative data, and quantitative as well as qualitative data.
In NVivo, cases are managed by creating case nodes, with each case node acting as the ‘container’ that holds all data, of all types,
for each case, regard- less of source.3
3 This is an example of where nodes are used for organizational purposes,
rather than for coding thematic content. As a consequence, you will store them in a folder to keep them separate from your thematic nodes (as you saw for the case nodes in the Environmental Change project in Chapter 1).
For an ethnographic study reviewing issues of research production and perfor- mance for academics in the arts, humanities and social sciences disciplines of a university, Pat created a case node for each member of academic staff, sorted by academic unit. Data for the study comprised administrative records of research funding received by each staff member and details of research publications pro- duced by them (originally in two Excel spreadsheets), individually completed sur- veys, web profiles, media releases, field observations, interview notes, other official records, and incidental documentary sources. The case nodes brought
Why does it matter now?
Cases and the case structure can be created at any time in your NVivo project. You can create a case structure before you import data; you can create cases as you import each source; or you can create cases and a case structure after data are imported, and even after all the data are coded (Chapter 6 details alterna- tive methods for creating cases). Creating cases can be one of the first things or one of the last things you do in the database, with no negative effects on the analysis. Given this reality, why are we asking you to consider the issue of cases carefully before you begin?
What you primarily need to consider, at this planning and data preparation stage, is that the need to create a case structure can have implications for how you might best format the data in your sources, particularly for text-based data. Table 3.1 sets out the various possibilities. It is helpful to be aware of these issues even if your case structure is very simple. The second half of this chapter then provides guidance on the practical aspects of data preparation, particu- larly when you need to differentiate multiple speakers within a single source for the purpose of managing cases. It also details other factors to consider in preparing data that are independent of managing cases.
9 Each source will remain intact throughout its NVivo journey, so you will never lose that important element of context, even if you divided its contents to create case nodes.
9 Planning for cases and the attributes you will attach to each case has implications also for how you organize your coding system. Your flags for male and female, for example, can be rapidly applied to all relevant text yet they are kept right out of your coding system, making both attributes and coding more efficient and effective to use.
Planning for attributes
By now you will have realized that attributes (e.g., age group;
location) and attribute values (e.g., 30s, 40s, 50s; rural, urban) are
together data from all or part of the various documents, so she could instantly access everything she knew about a particular academic. Additionally, once all the data were coded for issues raised, and for the scholars’ research areas (interper- sonal violence, pedagogy of mathematics, or religious experience, for example), she could easily discover which academics were interested in which issues, whether there was sufficient interest in any particular topic to create a research group, and who might want to be part of such a group, including details of what their contribution might be.
intimately related to cases. We will be discussing attributes in detail in Chapter 6, but in the meantime, while you are preparing your data, we offer the following suggestions:
Structure of data sources
What qualitative data will be included in each case node? (in
addition to demographic and/or numeric attribute data)
Formatting requirements when preparing data, for most efficient handling of cases (other methods applied
later will be much slower)
Each file (e.g., text of an interview, a video, a picture) contains all the data for one case only.
Multiple sources for each case, where each source is related to one case only, e.g.: the same person has been interviewed several times; your information comes from different people associated with a single case; or you have different types of data, such as an interview and a picture, for each case.
Individually identified speakers in a focus group
or
Interviews with more than one interviewee
or
Notes from a committee meeting.
One source only. No special formatting required, although it is useful to distinguish the text of different speakers (interviewer and interviewee) with some form of identifier at the beginning of each paragraph or by using headings.
Multiple sources. No special formatting required within each file, but ensure that each file for a particular case has a common root name. Then, where they are multiples of the same type of data, follow the name with a unique identifier, e.g.:
(a) Beatrice 1, Beatrice 2;
(b) John_self, John_mother, and John_teacher (where John is the case).
Portions of one or more sources. Speakers need to be uniquely identified, with their name on a separate line and formatted using a heading style, e.g.:
Dagmar (in Heading 2 style)
This and that, that and this (Normal style)
Daniella (in Heading 2 style)
Chatter, chatter (Normal style)
Ricardo (in Heading 2 style)
Structure of data sources
What qualitative data will be included in each case node? (in
addition to demographic and/or numeric attribute data)
Formatting requirements when preparing data, for most efficient handling of cases (other methods applied
later will be much slower)
Any other files that include data relating to multiple cases, e.g., field notes.
Responses to a survey or questionnaire, recorded as a dataset in Excel (see Chapter 9).
Combinations of any of the above, e.g., where you have survey responses as well as individual interviews.
Portions of a source, unless data have been extracted for each case separately.
One or more codable fields in a dataset, one row per case.
Portions of sources in addition to whole sources.
If data relating to each case can be separated within the file, include headings (as above) in the field notes.
The alternative will be to use interactive coding for relevant text – much slower!
Standard row and column identifiers for a dataset.
Format individual files as above.
Ensure that each data item or part-item relating to the same case uses the same unique name to identify the case.
9 Record attribute data (e.g., demographic details) as you gather your qualitative data. Try to think of all the kinds of comparisons you are likely to want to make, and, when you are gathering your data, record the details needed to make those possible. For example, if you want to compare what is said by people from different locations, then you will need to record information about where each person lives.
9 It is much more helpful to have attribute data recorded in checklists or, even better, in a spreadsheet, than to extract them from within the text of interview documents (where it is also a waste of a transcriptionist’s time!).