I received an e-mail this morning from a US client who we are working one-to-one with. He is very competent with NVivo and has completed his coding using the tool. The thrust of his e-mail was that he had completed his coding and where to from here? I’ve posted my entire reply below with only his name changed for obvious reasons:

“Hi Noel,

You certainly don’t sound like you need any training from me. Did you follow a pre-planned cycles of analysis strategy document?

There are two reasons you could cite when creating a rationale for using NVivo in your methodology chapter. These are efficiency and transparency. Queries like matrices allow you to view the data in a way that would be impossible (certainly within any kind of reasonable time frame) in a manual system. The coding process (at least the manual interpretive coding process) in NVivo is no faster than in a manual system. You select a segment of text or media by highlighting it and you drag it to a code as opposed to using a highlighter and writing a code beside it as you would manually. However, it’s in the  retrieval and reporting of data that you gain far more efficiencies in a database environment over a manual system. The scope of the analysis has much greater depth and breadth when one can move the data around so much faster and get different views and perspectives on it (Read Michael Bergin – NVivo 8 and consistency in data – click here– as one example of using a matrix to reveal a pattern in his data that he was not aware of).

However, efficiency is quite obvious when it comes to creating a case for computer aided methodologies. For me, transparency is the real benefit. Qualitative researchers are sometimes accused of being ad-hoc, subjective, undisciplined, touchy feely for example. There is a perception amongst some observers and even some academics that the process merely involves listening to a few tapes, getting out the highlighters and “identifying key themes” and that’s it. Off you go now and write a report and back it up with a few well selected quotes

You, on the other hand, have to produce an audit trail for your methodology chapter. It’s here that you establish your credibility as a researcher by demonstrating rigour in the analysis. That is why I asked you about cycles of analysis (not cycles of coding as coding is only part of the analysis). You should produce a well defined analytical strategy for this study that is entirely consistent with the philosophical underpinnings that support your methodology. Each cycle should be backed up with clear appendices that give evidence to the work. Once you establish rigour in the ‘methods’ section of your methodology chapter you should move to your findings. Not every finding, but those results that will end up engaging with the literature and philosophy in your discussion chapter; these findings should be supported with clear appendices that shows that they came out of the data. They came as a result of the data having been put through a series of processes. They are not merely an opinion based on some loose “identification of themes”

So you need to create the story of the analysis and cohesive and coherent well supported findings for all of the reasons cited above. I doubt you need any help from me in this regard Noel

It sounds like the ‘end is nigh’ which I am sure you must be relieved about.

Kind regards,

Ben Meehan

Consultant/Trainer for Computer-Aided

Qualitative Data Analysis Projects

Tel: 0035387 4172259

e-mail: benm@qdatraining.eu

WEB:http://www.qdatraining.com

One thought on “Using NVivo for your Audit Trail – demonstrating rigour in your analysis and supporting your findings”
  1. I've recently been working through Miles & Huberman' classic text on how to move on from coding to generate findings and (even) theory. Although it was first published in 1986, and the second edition in 1994, it seems to me far superior to any text written since. They centre their discussion around real research projects that they themselves worked on. Their method is to build successively abstract displays and models (trees and networks). Although much of what they show is based on "pen and paper" methods, they're supportive of QDA and make many suggestions for how to adapt their methods to it (NVIvo was called NUDIST, back then – a sad loss IMHO!).  You can certainly make at least a start on some of their matrix displays via Matrix Queries (altough most of what they do is closer to Framework), and you could use the Modeling tool to draw causal networks (I've not tried that seriously yet). Pat Bazeley's paper, that Ben mentioned, cites M&H heavily, and is also very helpful, especially for NVivo users.

    I'd recommend the book  highly –

    Miles, M.B. & Huberman, M., 1994. Qualitative Data Analysis: An Expanded Sourcebook Second ed., Sage Publications, Inc.
     
    Although it's still in print, there doesn't seem to be an e-book, but second-hand copies are available on Amazon.
     
    Alan.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.