Issues in Educational Research, 4(2), 1994, 103-108.

The mechanics of qualitative analysis

Jettie Mulder
Master of Arts candidate
Curtin University of Technology

This paper describes the benefits I have found by using the Outline option within the View Menu in my word processor as a tool to analyse qualitative data. The essence of the analysis was that I prepared the data with Outline headings, examined the labels in isolation and then moved similar labels together. As I work at the analysis I am in constant touch with the data, and I have full control over interpretation. This tool is sufficiently flexible to allow me to continue to use it to examine increasingly abstract theory as this emerges. The ease of operation keeps my attention on the research rather than on the mechanics of the analysis.

The problem

In my research project I am interviewing 40 people and from the 37 completed data files I have already amassed a large quantity of qualitative data. It was my intention to discover and describe similar viewpoints and to examine anomalous ones, and to write a report which was based on the data. I knew the theory behind what I wished to do, but the mechanics of doing it presented me with many challenges.

I initially used the Ethnograph, one of the computer programs which assists in analysis of qualitative data. Coding the interviews was largely a mindless task of systematically entering numbers in response to a prompt in the program, but I was able to print out all instances of a code for each interview. For the analysis of my first code I chose one which recurred within nearly every interview but did not contain exceptionally large quantities of data. Yet the print-out was seventy-four pages long and within that I had the viewpoints of only my first twenty interviewees. It was difficult to gain an understanding of the content of this amount of material. I did not know how I could come to terms with so much complex detail or how to draw out common themes in an efficient manner.

Using the Ethnograph further was not an option I favoured because the program separated me from the words and experiences of the interviewees. I could not maintain an understanding of the context, compare incidents or consider emerging patterns while I was doing the analysis with the Ethnograph. When theories began to take shape and I wanted a print-out of a new grouping of data, I had to suspend my development of the emerging ideas while I typed in line numbers. This was so unsatisfactory to me that I experimented with different approaches to analysis for several months. This paper describes the benefits I have found by using the Outline option within the View Menu in my word processor as an analysis tool.

Outline view

The Outline View was designed to produce different levels of headings such as those used in a standard report format. I use Word for Windows which accepts nine levels of headings. Once any headings are created the text under a heading becomes linked to that heading, and this gives new options to the user. Three of these options made this a valuable analysis tool:

Viewing only headings

It is possible to hide all the text of the document and view only the headings. The user can specify which headings are to be viewed: you may view only Level One headings, or two levels of headings and so on. This allows the user to see an overview of the report, and to move quickly to any part of the report.

Viewing the text

The user can switch instantly from viewing the headings to viewing the entire text. The position of the insertion point denotes which part of the document will be on the screen. For example if the insertion point is within the third heading when the user switches to viewing all text, then all text will be available for viewing, but the text immediately on the screen is the text under the third heading. This speeds up the movement to any part of a report. Alternatively the text under any one heading may be displayed. This is useful when the detail of any particular point needs to be checked.

Moving text

It also is possible to move any heading to a new position. When you click and drag with the mouse, you can move any heading to any new position and the text associated with that heading moves with it. This can be accomplished while viewing only headings or when all the text is on the screen.

Each of these three options is permanently available while the document is in the Outline View. The user can alternate between any of them continuously by the click of a button.

Analysis method

The essence of the analysis was that I prepared the data by creating labels which were outline headings, I examined the labels in isolation, and then I moved similar labels together. The mechanics of this operation were simple to effect but critical reflection was necessary to accomplish both the labelling and the moving. Many times I scrutinised the data for meaning before deciding on a label or considering where to move a piece of data. Yet these actions were worthwhile in themselves as they enabled me to get to know the data. By the time I had finished the labelling I knew the data very thoroughly, and as I moved similar themes together with continuous comparisons and criticisms, embryonic theories were being formulated.

Preparing the labels

1. I used the Ethnograph to define and print to disk all instances of a designated code in my research cases.

2. I changed the view of the document into the Outline View.

3. At the beginning of each interview extract I wrote a synopsis of the main point that was being expressed; this was in effect a label on the data.

4. I made this label into a heading within the Outline View. I moved through all the data repeating steps three and four for each extract. In this initial preparation I kept the interviews together and made a Level One heading of every case number and Level Two headings for each label. When the document was in Outline View I could see a synopsis of the expressed opinions of each research case.

Examining the labels

The labelling was the first level of interpretation of the data. There were no constraints on the creation of the labels, though I found that it was more efficient to keep them brief. While labelling the data I could distinguish between instances where one person repeated an illustration, or gave different illustrations of the same point. By examining the labels without the subtext I readily gained an understanding of the range of viewpoints expressed by all the research participants.

I found four additions to the process of labelling were useful:

• I broke a long extract into shorter pieces if there were different statements being made, and then represented each statement with a separate label. In these instances I needed to ensure that each part contained the case number because this was lost when it was moved from its original context.
• Where more than one viewpoint was expressed within an example I duplicated the extract and marked each one as a duplicate.
• When it was an advantage to obtain an overview of the spread of the viewpoint across the research cases I added the case number to the label. This allowed me to see the range of cases while examining the common themes.
• Where the code did not occur in a case I made a heading which stated that there was no instance of the code recorded.

Moving similar viewpoints together

I separated the data into groups of common themes by moving similar labels together. I was then able to examine the detail of each cluster of information. For example I could examine how many people spoke of a similar incident, what were the circumstances in which it occurred, and what the consequences were. Once I understood the content of these clusters of information I could compare different clusters for any relationships between them. Anomalous views also became apparent because they were isolated once the grouping was complete.

Further analysis

The Outline View as an analysis tool has proved sufficiently flexible to allow me to continue to use it in further levels of analysis. In some instances it was practical to delineate further levels of analysis within any large group. A worthwhile application has been the ability to reorganise the data according to emerging theory. For example when I discover an idea which explains three groupings of data, I demote the levels of the groups and impose the new theory as a Level One heading. The structure of the developing analysis could be said to resemble a growing pyramid: the lower levels of headings contain original labels and interview data, and the emergent analytical constructs are represented by increasingly higher levels of headings. I found different files were an advantage in keeping a record of this process as it occurred because this reorganisation has often been a tentative conclusion which may or may not endure.

Continuous change

Although I have described the stages of analysis as being discrete steps of labelling, moving into groups, and constructing theoretical interpretations, in practice the stages overlapped. I often reorganised the groups in accordance with changing ideas. Also as theoretical ideas became clearly defined I returned to the original complete interviews, and examined them for examples which had not been coded as part of the category in the first instance. Where I discovered these I copied them, added labels, and included them in the analysis. So the reality was that I engaged in any of these steps as the need arose, at all stages of the analysis.

Writing up the analysis

The threads of information which run from the data to the theory are well organised by the time the analysis is complete. I hope that the report can be written directly from this structure. The groups within the code are clearly delineated with ready reference to the spread of this view across the research cases and to individual viewpoints. Moreover examples and quotes which may become illustrations are already in place awaiting a final choice of which ones to use.

Benefits

As a researcher I have been able to maintain control of the interpretation of the data. The development of the theory is dependent on my own skill, and this analysis tool assists my exploration of meanings within the paradigm of qualitative research. Throughout the process I am in constant touch with the data, and this is valuable to me because one of my major aims is to present the experiences and concerns of the research participants faithfully. I could hear their voices as I worked with the data and I could take immediate account of factors such as the lifestyle or gender of the person, the context in which the incident happened, or the gestures and emphasis given as they told their stories.

My research benefited from the use of the Ethnograph to print out my initial coding because I was dealing with a large quantity of data. I consider that in smaller projects it would be feasible to examine the interviews directly for instances of a code, and use the routine cutting and pasting procedures of word processors to obtain the initial examples of a code ready for analysis.

Conclusion

It is enjoyable to work on my research project. At times when I probe for explanations of why and examples of how, the grouping of data takes on a meaning of its own. This is exciting. This is a new perspective on knowledge which has its origin in the data, and has been shaped by my insights. I would be well pleased if my method of doing the analysis allows others to enjoy the research process also.

Author: Jettie Mulder has a versatile background as a teacher. She has worked in all years K-7 of the primary school as a classroom teacher and also has experiences as a School of the Air teacher. Currently she is working on a causual basis lecturing and developing programs in access and bridging courses at TAFE. She is a candidate for a Master of Arts Degree in Social Welfare in the School of Social Work at Curtin University and is using qualitative methodology to write a thesis on the consumers' viewpoint of Family Law. Please cite as: Mulder, J. (1994). The mechanics of qualitative analysis. Issues In Educational Research, 4(2), 103-108. http://www.iier.org.au/iier4/mulder.html

© 1994 Issues in Educational Research
Last revision: 6 Nov 2013. This URL: http://www.iier.org.au/iier4/mulder.html
Previous URL: http://education.curtin.edu.au/iier/iier4/mulder.html
Previous URL from 24 July 1997 to 7 Aug 2001: http://cleo.murdoch.edu.edu.au/gen/iier/iier4/942p103.htm
HTML: Clare McBeath [c.mcbeath@bigpond.com] and Roger Atkinson [rjatkinson@bigpond.com]