You work NVivo by assigning attributes to data, linking comments or memos with data, categorising, modelling, searching.
It is a partially recursive and continuous process that eventually reaches saturation.
Using the tool this way embodies the essence of grounded theory, that is, concepts and ideas are grounded in, emerge from, engagement with the field and what we identify as evidence (often misunderstood as data).
NVivo is a rather open ended platform for doing this, particularly as it allows for native annotation of various kinds/types/sources of evidentiary traces. Again, I'm avoiding the term "data".
I am not going to describe NVivo, the menus, ribbon, mechanics of accessing particular functions or any other product feature walk-through stuff. The training courses I've taken were organised just like this, more about using NVivo that carrying out research. So, no boring disconnected stuff like: creating/opening/saving projects, familiarity with the workspace, adding material, nodes, coding with nodes, adding memos and annotations, importing and analysing spreadsheets (i.e. surveys and demographics), framework matrices, queries collections and search, charts, and reports.
For a functional view of how NVivo works review the getting started guides listed below or search for others.
What I am going to describe is some basic research workflow or work-activity that just happens to be performed using NVivo.
Activity scenarios for a historical research project
Exercises
1. Having found a new primary source document: a large scanned report of around 100Mb pdf...
Add a reference to the document as an 'external' source
Open this new element and write out a skeletal structure, table of contents etc, as a backbone to add your own (indirect) coding (nodes and memos).
2. Having found a new primary source document: a page on a newspaper's website...
Click on the NCapture widget from the browser.
Choose between article as pdf or webpage as pdf. Supposedly selects between downloading a representation of the whole webpage versus the bare article. Expect it to generate a file up to 5Mb or more depending on how much content is captured.
I only use this if I'm feeling lazy or if the whole document is interesting, including ad placements etc.
3. Importing NCaptured pages into the project
Select Data>NCapture. NVivo monitors NCapture downloads and offers of list of them for you to import.
4. How do I how to go about 'folding' a new item into the research?
Add memos, link them to nodes, edit the memos again and again...
Talking point: Why bother linking a memo with a source item or a node?
Hmm, I'm not really sure why we're limited to one memo 'link'. It means a 'memo' is a kind of floating annotation. A 'link' in NVivo seems to be used for a special, unique relationship. However you can use 'node coding' for multiple relationships between the memo, but not directly to source items (internals or externals). In that case you've got to go 'through' an intermediate 'node' relationship...
Talking point: While coding (linking to nodes) a new item I notice a theme emerging
I can make annotations in-line, but annotations are local to the document. What I'm thinking of is a new global item, a memo.
So make a new memo, give it a meaningful title, and edit it to start building up the idea of this theme.
However don't forget to link it back to the coded (noded) passage in the item I was working on before...
For me memos are kernels of analysis and argument that eventual evolve into my research paper. A scrapbook of partially thought-through ideas the I'll copy/paste into a working article.
Talking point: Adding a document to the project.
First problem, should I add it as an internal or an external? It depends.If the document is a huge digital object (100s of Mb or Gb in size) or unreadable by NVivo then use external.
If it is small and readable (e.g. text, pdf, jpg, mp3, mp4, etc) then use internal.
Internal is generally better because you can create in-place links (i.e. hyperlinks) between parts of the document and your coding (nodes, memos etc). Coding an internal or in-situ document is easier because the document itself structures the order and location of your nodes, memos etc. Coding an external document implies that you'll create a skeleton, index, table of contents or other indexing structure for linking coding.
Talking point: Coding, once having added a new source document to the project.
Second problem, this kind of implies that you either have some 'nodes' or that you will spontaneously generate some.As a starting point I took a initial research proposal and 'made up' terms as nodes in the project.
In the first instance you should just lump them all together as top level nodes. They can be organised into hierarchies but leave that job to later. In fact you will definitely reorganise nodes many times over.
A new top level node can be configured to aggregate sub-nodes; just check "aggregate coding from child nodes" if you want this behaviour. You can turn it on or off.
Talking point: Messing around in the project using "find", sets (collections), classifications, node matrices...
A set is just a way of making loose collections or shortcuts within NVivo. Use sets to organise nodes and sources in some way; like a way to collect narrative documents or things node-coded as narrative? You can add or delete set members without affecting the original items.Talking point: What's the point of Memos if you can only link them once?
I don't know. You might have thought that you could do the same thing with memos, that is, 'link' them with various relevant instances of evidence in multiple items... Well hard luck. You can't. From the help system - Understanding Memo Links; "Each source or node can have one memo linked to it and that memo cannot be linked to any other item. You can also have memos that are not linked to any item". :-(Questions:
Lucy
Novice?
What do I do? Total novice?
Basic format you want Word documents in?
What if one is on a PC and the other on a Mac?
What does NVivo do for me?
Marius
Total novice. How do I get started?
How to enter data? How to analyse it, coding techniques?
I want something that does the whole thing? Ironic.
How to do intercoder reliability? Is there a system test or report that can provide data on that?
Jury
Total novice, same questions.
Specifically, what are the pitfalls?
If I ask the system to give me evidence for a specific construct? How reliable is that?
How about linking constructs?
Fiona
I have blocks of data, that look disconnected, random. How to translate that into something that will answer my research question (magically).
Betul
Total novice.
I'm a qualitative researcher, want to to translate into a quantitative format. Some kind of quants?
Content or statistical analysis.
Paul
Advanced beginner.
20 case studies, 50 interviews, industry blocks.
Want to do analysis within and across these elements.
Refs
QSR International's YouTube site
Using NVivo: An Unofficial and Unauthorized Primer by Shalin Hai-Jew
Loads of getting started guides:
NVivo 10 for Windows Getting Started (from QSR)
NVivo Introduction Workbook (Finders University)
NVivo Advanced Workbook (Finders University)
