It is possible to start working on your data really fast if you want to – just type up, save, and import as a source and you’re ready to go. But you gain significant advantages from careful preparation, especially if your data con- tain any regularities or structure that can be exploited to facilitate management or coding.
What often works best, especially if you have mixed types of data and/or data that require special formatting or organization, is that you devise a plan and start by preparing only a small subset of your data. Play with that subset through all of the primary tools you plan to use such as coding, classifying and querying (this could take several weeks) and by the time you are done with this pilot set of data, you may realize things about preparing the data that you’d like
to change.4 Of course, it may not be possible or practical to conduct
this pilot with a small portion of the data, but if you are able to do so it could save you significant time and frustration as the project gets bigger.
9 Sources may be added to a project throughout the life of the project. Coding and classification systems remain flexible throughout a project.
Preparing text-based data
Text in an NVivo document can include most of the familiar richness of appear- ance that word processors provide, such as changes in font type, size and style, colour, spacing, justification, indents and tabs. So when you’re making notes or transcribing your interview, focus group or field observations, make use of this to help shape your data, express emphasis, convey the subtleties of what is
4 For many qualitative researchers, analysing some initial data before
collecting more is an important principle in any case – often leading to changes being made in ques- tions asked and/or theoretical sampling. This initial data should be retained within the project, however.
designing an nvivo database 57
happening, clarify how your respondents were expressing themselves,
or draw attention to critical statements.5
What can be imported as part of a text-based document?
Any formatting that helps enrich or clarify presentation of text, including headings, bold, different font types, and coloured fonts (but not highlighting).
Text in different languages can be included within the one document.
Embedded items such as images, text boxes, and illustrations (although you can code the entire embedded image, text box, or illustration only, not portions of these items).
Tables (the text coded into a node from a cell will appear in the context of its entire row when you open a node).
Ignored items include: headers, footers, page numbers, line numbering, shapes, high-
lighting, and comments.
9 Preparing data records is one of those areas where ‘fools rush in’. Careful editing and thoughtful structuring each reap rewards. As an analyst, you are dependent on the representation of reality that is contained within those records.
9 If you have any unusual formatting features, check out how well they import, and check how the coded text and other items look in a node, by importing and coding one or two prepared sources before you spend time preparing all the others.
Making transcriptions …
What appears at first sight as a purely mechanical task is, in fact, plagued with interpretive difficulties.
Transcribing involves translating from an oral language, with its own set of rules, to a written language with another set of rules. Transcripts are not copies or representa- tion of some original reality, they are interpretative constructions that are useful tools for given purposes. (Kvale, 1996: 165)
When you transcribe, you discover the value of using a high- quality recorder for your interviews and, even more so, for group discussions (using two record- ers is recommended for the latter). It is also valuable to do your own transcrib- ing, if at all possible – building familiarity with your data. At the very least, if another person typed the transcripts, it is absolutely essential for the person who did the interview to review and edit the transcript while listening care- fully to the recording. A typist who unintentionally reorders or omits words can reverse the intended meaning in some sentences – typists routinely miss n’t on the end of verbs, for example.
5 Although these features can be added to the text once it is imported into
58 qualitative data analysis with nvivo
The flat form of the written words loses the emotional overtones and nuances of the spoken text, and so it is beneficial for the interviewer to for- mat or annotate the text to assist in communicating what actually occurred with a view to the purpose and the intended audience for the transcription. ‘Transcription from tape to text involves a series of technical and interpreta- tional issues for which, again, there are few standard rules but rather a series of choices to be made’ (Kvale, 1996: 169). The goal in transcribing is to be as true to the conversation as possible, yet pragmatic in dealing with the data.
Kvale (1996) and Mishler (1991) provide useful discussions and examples of issues involved in transcription, and Bazeley (2013) lists general guidelines to follow when transcribing. Whatever procedural decisions are made, they should be recorded as clear decisions and formatting guidelines for the typist(s) to ensure consistency in the transcription process, in your project journal to aid interpretation from the data, and (if you are a student) in the methods chapter of your thesis.
… or not transcribing
Interviews and other sources for sociolinguistic, phenomenological or psycho- logical analysis generally should be fully transcribed. When nuances of expression are not needed for the analytic purpose of the research, verbatim transcriptions may not be needed; notes may be adequate for the task. Using a computer program to assist analysis does not automatically mean you are required to use full transcripts. As with transcripts, however, keep the com- ments in their spoken order and context rather than rearranging them into topic areas.
When Pat had assistants interviewing researchers about the impact of receiving a small financial grant on their development as researchers, the conversations were recorded. Much of the recorded conversation was not directly relevant to the topic of her research (researchers have an irrepressible urge always to tell what their research is about), and so the interviewers made notes from the recordings, supple- mented by occasional verbatim quotes.
Again, in a different setting, when Australia’s corporate watchdog set out to investigate ‘boiler room’ share-selling scams, the researchers worked entirely from notes of their telephone conversations with those who had been approached by the scammers. These were sufficient to trace the chain of events and the interviewees’ responses to them (Australian Securities and Investments Commission, 2002) – deep emotional responses or phenomenological essences were not the concern of the sponsoring body!
Then there are always times when you discover the audio recorder wasn’t working. And times when your participant opens up just as you have your hand on the door to leave …
Formatting text-based documents with heading styles
Where different sections of a source identify contributions from different people or responses to different questions or notes from different days, the use of headings formatted with heading styles will be critical to your being able to easily differentiate and create separate cases or coding categories from those sections of your documents.
Headings are usually less than one line, for example ‘Q01 History’ or ‘Danny’, but in Word they are treated as being a whole paragraph. The heading describes what is contained in the (normal text) paragraph(s) immediately following it.
Heading styles in Word are always applied to whole paragraphs. Thus the use of styles for headings means that a built-in style in Word has been selected and applied to the paragraph that comprises the heading. NVivo will not see text as a heading just because it is bold or in a different font. Figure 3.1 shows part of one of the interviews in the Environmental Change project with a map of the heading styles and levels used in that document.
Interview with Margaret from Davis, North Carolina; Margaret works at a museum.
Connection to Down East Heading 1
Nancy Heading 2
How long has your family lived Down East?
Margaret Normal
We have lived Down East since about 1735 in Davis. Both of my parents are Davis’s from Davis. I am Davis. So basically I am a Davis Shoreman from way, way back. This is where my roots are. This is to me the most beautiful place on earth. I have experienced other places, and they were incomparable to Down East, Core Sound. We have our own sort of language and values systems and just the beauty of the area that I think is not found in a whole lot of other places.
Connection to Down East natural environment Heading 1
Nancy Etc.
Tell me about your connection to the natural environment Down East.
Margaret
I do love to go clamming and stuff but I never have the chance anymore I am sorry to say. The last time I went fishing was with NCCAT–the North Carolina Center for the Advancement of Teaching.
As in Word, heading styles in NVivo are hierarchical, breaking a document into parts, with subparts. A heading level allows NVivo to select all of the text between it and the next point where there is
another heading of the same level.6 NVivo recognizes three features
within the heading:
1 the heading level that has been applied;
2 the exact string of characters where the heading level has been applied; and 3 the associated text that follows the heading.
These features allow NVivo to differentiate between Nancy’s and Margaret’s texts, since they are separated by and identified with headings with a style of Heading 2. (The questions are given a Heading 1, which we will talk about later, but for now let’s focus on being able to sort the data by person.)
Using headings to separate and identify cases
When speaker names are set out as headings so that turns in conversation are separated in the document, this presents several opportunities.
If you are analysing a focus group or a source with multiple speakers, you can apply auto coding to efficiently separate out all of the contributions from each individual and store them as part of the case information for each of those individuals (see Chapters 5 and 6).
In field notes, the heading might be used to identify the case being observed, or perhaps the date or site or circumstances of the events being described. This might be simply to clarify the structure of the notes, or so you can auto code sections of them.
If the document is an interview with two speakers (the interviewer and the participant), you will be able to auto code for who is speaking, and then run a text query to look for the word sustainability, for example, and limit finds to occurrences within the par- ticipant’s text (see Chapter 5).
9 We usually recommend using Heading 2 for speaker names, so Heading 1 is available should you choose to insert topic headings (the assumption is that the topic covers more than one speaking turn). Consistency in the level of heading chosen for a particular kind of item (such as speakers, or questions, or anything else you are using headings for) across all your documents is important, as this will facilitate use of auto coding and query tools in NVivo.
6 In this regard, NVivo is slightly different from the way heading levels work
in Word. In Word, selecting text on the basis of a heading style will select all text to the next heading of that level or higher; that is, Word recognizes the hierarchy of levels, so that if you select on the basis of a Heading 2, it will select down to the next Heading 1 or 2. Because NVivo has to work with names of the styles from across all languages, it works on the basis of finding the next example of the same style only, without considering the
hierarchy of levels (hence the reference to paragraph styles rather than heading styles in its dialogue boxes).
9 Have an observer at your focus group to record a speaker identifier and the first word or two said for each speaking turn. If the typist is having trouble identifying different voices from the recording, this can be used to verify who was speaking.
9 Add heading styles as you transcribe using a keyboard shortcut,7 or after you have
typed the transcript by using Replace (click on the More button in the Replace
dialogue).
Using headings for separating questions or other sections
Heading levels in NVivo can be useful also to efficiently separate data in a file based on question asked or topic covered, again using auto coding. This strategy is often applied when researchers want to collect all the answers for Question 1 and separate these out from all the answers to Questions 2, 3, 4, etc., or when several topics are covered in a focus group discussion or in a set of field notes. By using a heading level for each question or section (usually Heading 1), you will be able to use auto coding to create nodes, each of which will contain all the answers to a particular question or comments or notes on a particular topic.