We've done this a little using coarse natural language processing to go through focus group transcripts, and even medical journal papers. See attached working papers for the ideas. The software we used was free, as per cites in there.

Although some claim the tools are quantitative, I've always used them under the humble assumption that they're qualitative in nature and that they'll help surface new theory or meekly question existing theory.

Natasja - if you don't mind me saying - your research approach seems somewhat misplaced. It's highly unconventional to conduct research, collect data - and then ask 'how can I analyse the data?'. Your data analysis approach should be guided, from the beginning, through your literature search and review and then your data collection technique.

You are correct that theories are usually tested quantitatively - but that is on the basis of existing theory. The main 'theoretical' approach to qualitative research is Grounded Theory (and Strauss's approach is mainly adductive) - but you would not be asking this question if you have originally adopted this methodology. You can't conduct qualitative interviews and then say 'I'm now going to adopt a Strauss-inspired GT approach which I hadn't known that I was going to use previously'. 

Most qualitative analysis techniques are set-up to be inductive rather than deductive, but one alternative would be certain types of content analysis. To do this, you would set up a code book in advance, based on your theorizing. You would then predict the patterns that you would expect to see in your coding.

As a general reference on content analysis, I would recommend Krippendorf's book, titled Content Analysis.

Marco, thanks a lot for your input - and the documents attached. I have read them both with great interest. To me, natural language processing is a new approach I have yet not familiarised with - thus, reading your draft as well as a search on the internet gave me some food for thought. The idea of indexing documents by individual terms corresponding to words or word stems - as I understood it - seems highly interesting, especially compared to traditional coding tools. However, it strikes me as complex tool where the trustworthiness of results produced is - such as every other methods- highly dependent on its thorough application. Would you, based upon our experience in working with this, gather it a comprehensible tool for first time users?

Dean, thank you, too, for your critical remarks. It makes clear to me, the importance to communicate matters clearly and provide relevant background information when approaching any public audience that is not familiar with the project.  I do agree with you in -  and dare to take it for granted - that any methodological procedure should be guided and well-informed throughout the research process. I have dismissed GT as an option, as my data collection procedure did not allude to GT principles. However, I do believe that research should be undertaken in the best interest of the informants and with tools chosen that are most suitable to produce data that provide nuanced insight into the topic of investigation. Unconventional choices in research may hence not be inappropriate ones - even if that implies adjustments in the initial methodological procedure during the process - as long as the amendments do cohere with the ontological and epistemological pillars upon the research has been designed and data have been collected. 

I would - if I may- like to clarify the matter described above, in order to avoid further confusion, and would appreciate your point of view on the following:

I collected qualitative data on nurses who have been employed in a coordinating function, in order to support cancer patients in their treatment and rehabilitation process. It is a pioneer project initiated by the Cancer Association. Being employed in highly varying municipalities and the freedom to implement their position in different ways, little is known about in what way they work. The interviews conducted are semi-structured. The aim is to get an understanding of how these nurses engage in community work and how they regard their own role in supporting patients.

While conducting and reading the interviews, it emerged from data that these nurses, despite great variations in municipalities, engage in their work with a strong focus on health promoting practices, assets based approaches and salutogenic orientations.

Applying inductive analysis does require bracketing of all prior knowledge. That leaves me with the choice to dismiss highly valuable material to exemplify health promotion and salutogenic orientation applied to clinical practice and in political work at local level - which could potentially add empirical data to a field whose literature, in my eyes, is mostly conceptual to the current point of time.

Alternatively,  I could attempt generate empirical based knowledge on health promotion practices and salutogenic orientation of community nurses in coordinating functions, using a deductive or abductive approach. This consideration is informed by articles by e.g. Reichertz, or MacFarlane&O’Reilly-de Brún, 2012, that argue that use of orienting concepts highlights the scope for qualitative research to test or further explicate preexisting theory as long as their use is resonant with the iterative processes inherent in qualitative research.The idea to proceed abductive or deductive rather than inductive is inspired by work from Starks&Trinidad, 2008, who apply different analytical approaches to the same data set to demonstrate different results produced, from which I gathered possible to apply an alternative approach to my data set than initially planned.

The problem I am facing now is mostly related to coding. The key constructs in conceptualisations of both health promotion and salutogenesis  do incorporate specific labels such as - in the latter case, Sense of coherence, general resistance resources, a.s.o. While the data embrace these key constructs as described by nurses, they have - amongst others due to lacking acquaintance to the terminology - not explicitly applied the terms.

The question is then more particular: Does anybody have experience in using orienting concepts in qualitative research to test or further explicate preexisting theory, and if so, in coding qualitative data for preexisting theory, where key concepts lie implicitly in the material rather than being named directly? (E.g.: "I work on creating walking groups" => HP; "I want them to make sense of their situation" => SOC/salutogenesis.) Is there any recommended approach to it, or methodological article from which I could gain deeper insight into such a procedure?

And of course, critical remarks and pinpoints to potential pitfalls are highly appreciated as resource for reflection an contemplation as well.

David, I'd like to thank you as well for your input, this is appreciated. I will make sure to get access to the book you have recommended. It is encouraging to read that there are certain ways to code for deductive content analysis. This is what strikes me as the most suitable approach in this case. As I described in a more exhaustive comment above, I am facing the challenge that the constructs of the theory - the the "pillars" of the conceptualisation - do have technical labels which the informants do not use in their day-to-day language. Anyhow, the the informants provide rich description of the theoretical constructs. Would you - from your point of view - regard it possible to code for concepts without having their labels outspoken in the text? (E.g. "I focus on generating municipal resources to improve health" = "assets approach"). Would you further think that coding is equally appropriate for theoretical constructs that also have been tested using quantitative methods? (I am here referring to the General resistance resources and the Sense of Coherence - both theoretical concepts that incorporate key terms which I consider coding qualitative data for, but which in prior quantitative investigations also have been tested using a scale/ questionnaires.) I would acknowledge if you would share your thoughts about the two considerations described above.

Natasjke, Everything you have described is a straightforward qualitative analysis, where you have inductively derived four concepts. In particular, you haven't mentioned testing any hypotheses, which would be the defining feature of a deductive analysis. So, I'm not sure what your concern is.

FYI content analysis can be either more qualitative (inductive) or more quantitative (deductive), depending on the extent to which you use pre-determined theory.

Hi Natasjke, David's comments are quite relevant and helpful.  One more you can look at is the approach that Michael Agar suggested in his small but informative book, "Speaking of Ethnography" published by SAGE.  In it he describes the formulation/modification of explanations (and "theories" as well) from a qualitative approach using "strip analysis".  You may want to take a look at that.

Thank you both for your comments, David and Jack. David, I think my concern was that I come from a school from phenomenology, where bracketing and the exclusion of pre-defined concepts is of great importance, thus the analysis from a pre-defined perspective is a new method for me. Thus, I needed to clarify some ideas to ensure I am doing this properly. Your input and clarification is appreciated! 

Jack, this looks really interesting! I think it might help me clarify the last details I need to know. Thanks a lot! 

Thank you both for your comments, David and Jack. David, I think my concern was that I come from a school from phenomenology, where bracketing and the exclusion of pre-defined concepts is of great importance, thus the analysis from a pre-defined perspective is a new method for me. Thus, I needed to clarify some ideas to ensure I am doing this properly. Your input and clarification is appreciated! 

Jack, this looks really interesting! I think it might help me clarify the last details I need to know. Thanks a lot!