X In List View, Right-click > New Query > Word Frequency.
X For Finding matches, use Exact match, or Including stemmed words. The latter combines different forms of the same word into the same ‘find’, as shown in Figure 5.6.
X Set the options you want for where to search in the Word Frequency Query dialogue. Usually you will Search in: Text, but you might want to restrict where NVivo searches (Of > Selected items or Items in Selected Folders).
X Indicate how many finds you want displayed, and whether you want to set a minimum size for found words (usually 3 or 4 characters is a good minimum).
X NVivo will display the found words in order of frequency (Figure 5.6); you can change the sort order to alphabetical by clicking on Word in the header row.
Figure 5.6 Word frequency query with stemmed words
You now have a number of options for viewing and working with the results from the word frequency query:
X Double-click a word of interest and view it in context. NVivo will provide 5 words either side as default; to see paragraphs, select one or more finds, then Right- click > Coding Context > Broad. Code relevant content to a node in your coding system as desired (not every find will be relevant). Or
X Save all the finds for a word as a node. Select the word from the list, Right-click > Create As Node, then Select Location (i.e., a folder or parent node for it), and
112 qualitative data analysis with nvivo
(Continued)
provide a Name for the new node.
� Incorporate something in the name or the description for the node to indicate it was created as a result of an automated search, rather than through interactive coding – at least until you have ‘cleaned up’ the finds recorded there.
Viewing and adjusting stop words
Stop words are not found as results when you run either a word frequency query or text search. These include such English words as ‘any’, ‘the’ and ‘yours’. In most cases, these words would be regarded as ‘clutter’ in the results.
X Select the text search language and view the default list of stop words for that language in File > Info > Project Properties > General >
Stop Words. Stop words are project-specific.
X Words can be added by typing them in to the list. Alternatively, in the results of a word frequency query, Right-click > Add to Stop Words
List.
X To remove a word from the default list, select and Delete. We recommend that you remove conditional words like ‘because’ and ‘if’ as these can be useful to follow up and view in context. Also, to compare use, say, of ‘the baby’ with ‘my baby’ (in a text query), you will need to have removed ‘the’ and ‘my’ from the list, at least temporarily.
? Search Help for Run a word frequency query.
Coding with text search queries
The capacity to search through sources and identify passages where a particular word, phrase or a set of alternative words is used as a pointer to what is said about a topic unquestionably offers the hint of a ‘quick fix’ to coding, at least where an appropriate keyword can be identified. The conundrum is that searching text offers much more than that, but also much less.
Searching for ‘sleeping’ and ‘feeding’ quickly identified passages (paragraphs) relevant to those tasks in infant care, in the first stage of Katey De Gioia’s (2003) study of the continuity of microcultural care behaviours between home and child care for culturally diverse families. Analysis of those passages then revealed that this was not so much an issue for the parents, but that what was more important was the quality of communication between the child-care centre staff and the families. This new direction then became the focus of further interviewing and analysis.
going on with coding 113
and save them as a node. As a strategy for locating, viewing and coding text, a text search can:
Code passages based on repetitive features, such as speaker initials appearing at the beginning of a paragraph. It can, thereby, provide you with an alternative to auto cod- ing based on headings, although it is much less convenient to run, so if you can auto code using headings, take that option in preference.
Code topics, people, places or groups identified by a keyword, for example, in a tour- ism project, identifying passages (paragraphs) about regularly mentioned locations or environmental features.
� If you are using NVivo to analyse field notes, whether written in Word or NVivo, facilitate rapid coding of those notes by strategically placing routine keywords within the notes you write (or dictate). This would be particularly useful if you are in a situation where you are unable to code as you go.
As a tool for coding unstructured text, a text search can be less adequate, gen- erating what we typically call ‘quick and dirty’ coding suitable primarily for exploration, or for use in an emergency. Attempting to use it for more than that, and particularly as a primary tool for interpretive coding, is bound to disap- point. As Lyn Richards (2009) noted, searching for words in the text is a mechan- ical process, so why should one expect it to help in interpretive research?
The primary difference between a word frequency query and text search is that text search allows you to look for any word, even if it is not in the top 1,000, and it allows you to look for alternative words and for phrases instead of just solitary words. As with the word frequency query, you can also look for stemmed words and synonyms. Like most other queries in NVivo, text search allows you to choose between viewing a Preview Only of your results or placing them in a Results node in the database (see Chapter 11). You can specify how much context from the original document is returned along with
Some years ago Pat led a team under pressure to complete a report for a large time- limited project in which some of the interview data had not been coded (Bazeley et al., 1996). They needed to write about mentoring of early career researchers, and they knew it had been specifically asked about in the later interviews. A search for the word ‘mentor’ pointed to all those passages where the question was asked or that otherwise dealt with this topic, and so it was easy then to review the original sources, to find what was said, and to confirm and elaborate what they already sensed were the issues around mentoring. However they found it was impossible to do the same thing with respect to the importance of building a niche area in research – there were no keywords to unlock that aspect of the texts. Similarly, if you try to run a search for time management, or even just ‘time’, in data from researchers (and probably from most people), you will find everything you didn’t want to know about next time, first time, full time, at the time, whole time – and virtually nothing at all relevant to managing time.
the search term, or you can view the context of your finds after the search is complete (Figure 5.7). It is then up to you to review the passages found, and to determine which are actually useful and which are not. Those that you want to retain are best coded on into a new or existing node. You can then delete the original (temporary) results node.
Figure 5.7 Results of a text query, showing keyword in context and alternative display options. Additional context is available from the right-click menu