Some nodes are easy to see as one of a larger group (in the park is one of sev- eral possible locations), and some you see as an instance of a more general concept (conflict is one of several interpersonal issues). It is natural to categorize and organize objects so that the category at the top of a hierarchy describes the contents, in general terms, of the items below. In NVivo, nodes can be structured in a branching tree system with categories, subcategories and sub- subcategories. So, just as folders and subfolders in your computer filing sys- tem help you to organize your files, nodes in ‘trees’ will allow you to organize your categories into conceptual groups and subgroups.
In this chapter:
Learn how to best organize your nodes into a structure with categories and subcategories.
Discover how to use auto coding tools for routine coding tasks.
Decide if and when word- and text-searching features might assist coding.
Think about how a department store organizes its mail-order catalogue. There will be major sections dealing with clothing, kitchen, linen, entertainment, and so on. Within the clothing section, there will be subsections for shirts, dresses, pants, underclothes, shoes. This doesn’t mean that the different shirts bear any particular relation to each other, or that this dress is worn with that dress – only that those in the same section are the same kind of thing, and it is useful for clarity and com- parison to have them all located in the same place. A coding system that reflects this arrangement might look as shown in Figure 5.1. The shopper, after scanning the various sections, could then put together a coordinated outfit that comprises shirt, pants, jacket and shoes. Putting together the outfit is more like making theoretical connections between the nodes: these things go together to build a larger con- cept, they occur together, or they impact on each other in some way – these kinds of connections will be shown in another way.
Figure 5.1 The shopping catalogue ‘tree’
Hierarchical, branching systems are often the subject of debate, are open to revision, and may be more or less helpful, but the principles on which they are built remain consistent. Things in one group are more similar to each other, with respect to critical variables, than they are to things not in that group. In the world of Australian plants, Eucalypts will be classified separately from
Banksias – not much argument there, but as botanists gained further knowl-
going on with coding 97
perceived, and the categorization of some species was changed from Casuarina to Allocasuarina. If there is a change in the critical variables around which the system is built, the final categories might look different, but the system, or taxonomy
will have been constructed using the same principles.
The coding system in NVivo works as a filing system or catalogue for nodes. And like other cataloguing or filing systems, the organization of nodes in trees can be the subject of debate, open to revision based on new understandings, and more or less helpful for particular purposes. Those coming to the same data from different perspectives or with different questions will be almost cer- tain to create differently labelled and organized coding systems. The structure for your nodes will evolve over time, particularly when you engage in a period of review to check the consistency and salience of each one. Nevertheless, each will find a place with other concepts of that sort; there will be a logical fit.
� Later you will use sets, queries, relationship nodes and models to identify and record connections between concepts of different types, such as a combination of place, event and response. While these tools (described in later chapters) are independent of the coding system, having your nodes effectively organized as branches in a tree greatly facilitates their use.
Why bother sorting nodes into trees?
If the trees don’t show the relationships (theoretical connections) between nodes, then why bother with them? What is the point of just listing things in groups? Using trees to create a structured organizing system for concepts brings a number of benefits:
Organization: The hierarchies help to create order out of randomness or chaos. The logic of the system means you can find nodes and you can see where to put new nodes.
Conceptual clarity: Organizing nodes helps to give meaning to them; sorting them into hierarchies prompts you to clarify your ideas, to identify common properties, see miss- ing categories, and sort out categories that overlap. And you clearly see what kinds of things your project is dealing with – the structure of your data. The coding system, when established, will ‘tell’ your project (Richards, 2009). Indeed, when someone approaches Pat for assistance with a project, there are just two things she asks them to send ahead: their research question(s) and a list of nodes.
Prompt to code richly: Well-organized trees provide a useful tool for ensuring the thor- oughness of your coding, as you progress. You stop to code a passage because an interesting issue is raised in the data. Capture that, but before you leave the passage, run an eye over your coding structure as a quick visual prompt to see if there are other nodes that are relevant. Should you also note (code) who the key players are, what the context is, how people felt or otherwise responded? Do you need reminding to always code the outcome? Making sure that the text is coded at nodes across all relevant trees allows for more effective and more complete answers to queries.
can make a significant contribution to an emergent analysis. If all events are sorted
98 qualitative data analysis with nvivo
together and all responses are in another tree, for example, it becomes a simple mat- ter to set up a query to identify the overall pattern of which events give rise to what responses. Additionally, it will be possible to see what each of those responses might look like in relation to particular events, and so, for example, although it occurs for both, enjoyment is expressed differently depending on whether it is in response to working on a task or playing sport.
Building your coding structure
To create some conceptual order in your coding system (‘point-and-click’ instructions for moving nodes around follow):
Start with a thinking-sorting process, to decide how the nodes might be arranged. When you ask yourself why you are interested in a particular category or concept (abstracting), ask yourself also what sort of thing this concept is about – what group it belongs to. Work through each of your nodes in this way, making organized lists of things (if you’re struggling to do this, look for ideas in the section on Kinds of trees, below).
� Create a model which holds each of your nodes, so that you can push each around the screen to help sort them into groups of like things – see Sorting
concepts to build a hierarchical code system in Chapter 10.
� ‘Tell’ your project to someone. Both the telling and the discussion will help you work out the main kinds of things you are dealing with.
Decide on what might be top-level nodes in your Nodes area, as needed. You may already have some nodes which represent broad categories (e.g., emotions, attitudes, environmental issues) that will be suitable to keep in position as top- level nodes. Others will need to be created especially for the purpose of providing a structure, rather like coat-hangers for others to hang from – the way Clothing, Furniture, Linen and Whitegoods are in Figure 5.1.
Rearrange your nodes by placing some under others, so that their arrangement makes sense for you. Not all nodes will immediately find a place in the system – that’s OK. Some nodes may need to be copied into two trees because they embrace two ideas. In that case, they will also need renaming once they are in their new locations, to reflect each of the two aspects.
Once you have a structure, check that it serves the main ideas you set out with, and the research questions you want to answer. If part of the structure doesn’t fit, ask whether you need to modify the structure, or your original purpose and questions.
� When you create a new node now, you should be able to see where it will go. � Experience teaches that projects typically don’t have more than about ten
trees, and that tree structures usually don’t go more than two or three layers deep; it just isn’t possible to subcategorize much more than that without starting to confuse what class of thing you are dealing with. There may be a large variation in the number of nodes within a tree, although if you get too many (say, 30+) at a single level in a tree, it could be time to group them into further branches, or to see if some of them could be merged.
Regard your coding structure as a work in progress. Some nodes may take longer to group (and may never be grouped), and others will move from one tree to another until you crystallize just what it is they are ‘a kind of’. Some start out as a fairly random collection within a tree that you return to later, to sort into branches. We often have an Issues group (for things about which there might be debate for this topic) which later turns into issues of various types as we gradually realize that we are dealing, say, with structural issues, political issues, interpersonal issues, and so on; these are then set up as subgroups under the top-level node of issues, with the specific issues at a third level (or each main type of issue becomes a new tree).