Hello,

Great subject! I have done different types of analyses and coding. I find that the best option is to use data coding Software such as Atlas ti. If you want to identify cultural constructs, iI would suggest you start by coding without pre-established concepts, this way you will be able to capture more details and more variances in the discourses of your respondents. Culture is such a fuzzy concept - identifying how it is perceived within the organization as compared with what the organization supports will certainly be of interest.

A question, will you be collecting data from multiple cases?

Hope this helps.

Hi.

To add to Chantal's comment, using a grounded theory approach could help with coding. But yes, the fuzziness of culture could pose a challenge. I would suggest that you use different theoretical lenses through which to conceptualise culture. Each will require attention to difference variables, but together they create layers for the analysis. To be more specific, this should help you to analyse variables such as gender and race. Each organisation will have professed values for gender and racial equity, as well as other dimensions of diversity. How these values are articulated may differ starkly with the experiences of staff within organisations. To theorise this, works such as Baxter's Feminist Post-structuralist Discourse Analysis could be useful. Other levels could of course include community engagement and conflict resolution. All these factors may loosely be classed under culture. As such, you may need to define your focus (at least your initial focus) more specifically.

It depends on the subject of your research. If you are interested in the functional or problem-solving aspects of culture--what are the problems, what would constitute a solution, who is responsible for what, etc.--then code for that. If you are studying workplace justice, that's a whole different set of issues. In any case I would consider trying to isolate critical incidents where matters were widely perceived to go awry and see how different actors diagnose the underlying issues.

These are all really useful answers - thanks everyone.

Chantal, I will be using NVivo for the analysis of each data set I obtain. I should have 3 sets from 3 different settings and it will be analysis within and between the settings that I am after.

Marthinus, I wold love to have had a grounded theory approach but it is not feasible due to time constraints and the model I am working with. I will be analysing the data through lenses based on a particular model of culture and also the concepts from a theoretical appraisal of corruption of care. The layers and concepts may take some work to coordinate!

Steven, thanks for the feedback - I am not going to be looking at any specific incidents although some of the data may relate to a hypothetical incident.

Hello Norman, these precisions on the setting and approach are quite helpful. Here then, I understand that you are using a particular theoretical lens for coding and will be applying deductive coding. I would still suggest you have a phase of inductive coding of the data so you can allow soace to capture some aspects of culture that are not acounted for in theory. I quite like the idea of a hypothetical incident for application, the type of data would lend itself well for inductive coding since there would be sone unpredictable responses.

As for coding, I am familiar wth Atlas ti not NVivo. A feature I like on Atlas ti is the networks and types of links. I don't know if NVivo has such a feature, but if it does, it could help with keeping track of how the consepts are connected. Another thing you can do is to group codes according to "famillies" of codes, it makes keeping track of the layers easier.

Hi Chantal. Thanks for the thoughts - I had considered inductive coding but wasn't sure about the value, so your comments here have helped to reconsider that aspect of the analysis.

The hypothetical incident is also aimed at getting comparative answers between degrees of seriousness. It will be a non-serious incident in care provision (an error of some sort) but asking what the consequences would have been if it had turned out to be serious or even fatal.

I have only had a brief overview of NVivo so far - I will see how the coding facilities shape up.

Hello Norman,

I die my PhD on language/discourse differences between managers, consultants and service engineers participating in change projects. I used transcripts from interviews, formal and informal conversations between these groups and coded this using four discourse perspectives; speed acts, negotiation strategy, communicative support and conversation phases. On top of that I counted words in the strategic / change project documents. This PhD study was a case study accross three cases in the Netherlands. It was explorative and a mid range grounded theory appraoch. Perhaps this discourse framework might be helpful to you, without having the coding required from a pure grounded theory appraoch. I will upload the digital version of my PhD later.

All the best,

Jos

Sounds interesting Jos. I think the four broad categories you refer to would be useful in an inductive analysis a the start. The part that is really interesting for me is the coding across the interviews/conversations and the project documentation as that is part of the work I am going to undertake. I would greatly appreciate a look at the discourse framework.

Hello Norman. I have just uploaded the file of my disertation, but I am not convinced it is available, perhaps due to th file size (8MB).

Hi Norman! You might take a look at contrapuntal analysis, a discourse-analytic tool that is specifically designed to examine how discourses compete, clash, or contradict. A full description of how it can be used and enacted is available in Chapter 6 ("Doing Contrapuntal Analysis") of this book:

Baxter, L. A. (2011). Voicing relationships: A dialogic perspective. Thousand Oaks, CA: Sage.

Hope this helps!

Hi Jimmie

Thanks for the reference - that sounds exactly like the approach I am after. And a great name for the author to have :-)

Hi Jos

I can't see your thesis so I guess the file size is blocking it as you say. Do you have the methodology section separately? - it might be a smaller file.