I have seen frame analysis been used in the comparison of qualitative data successfully. It might also be useful in you case to compare whether different groups of people e.g. construct the same issues as problematic in the same way, believe in the same solutions for the same reasons etc. I have only been starting to work with this tool very recently myself and would therefore be interested in any experiences someone might want to share.

What type of data are you planning to use?

I cannot remember a tem as "comparative analysis" in qualitative methodology; it seems rather vague: actually you always compare themes (or "nodes") because you have to code what you hear (or read from transcripts) so grouping what is similar and gouping elsewhere what is not. So you can analyze what is siad by two different groups of subjects to see what is similar and different.

I give you a link for a useful e-book.

Here follows the file

Hi Norman,

I woudl recommand you the technique called Qualitative Content Analysis (QCA). It is a specific quali method that has been developped in order to conduct systematic comparative analyses when a coding scheme is already existing or when it will be developped in order to be apply rigorously to compare different sub-groups within a dataset. This method seems appropriate based on what you described. I would recommand the book 'Qualitative Content Analysis In Practice' by Margrit Schreier (2012) to start with.

Best,

Jonathan

Grounded Theory was originally developed as comparative sociology. In the original method you took social units of analysis and compared them on all characteristics in order to develop a theory that explained what was happening in the action area.

But both comparative analysis and grounded theory subsequently went in different directions....

Thanks for the responses, I shall check out the references as soon as I can.

In order to clarify what I am looking to compare, the research is into organisational culture and its impact on practice in care settings. As part of the fieldwork, I will be holding focus groups of care staff to elicit their views of management, operational prioirities, training provision and the experienced ethos of the setting. I will also be interviewing the owners/trustees of the setting, who may have a view of waht the theos and culture should be and want to have a structured analytical approach to compare the two views. Further to this will be interviews with the managers of the care settings who have to manage the interface between the owners/trustees and the staff. As a consequence of their role there may (will?) be a corruption or distortion of the preferred ethos and culture. I want to do a structured analysis comparing all three views.

I know the coding and analysis will be unique to the project but I was wondering if there is an existing methodology or technique for such a comparative analysis. From the replies so far it seems that there may be but I need to check the references first.

Thanks again for the responses.

Vincenzo, many thanks for the ebook link - it should be really helpful

Several qualitative methodologies share the "constant comparative method" of data analysis, including grounded theory developed by Glaser and Strauss and expanded on by Kathy Charmaz. Also, Miles and Huberman have a very useful text on qualitative analysis also based on the constant comparative method. NVIVO software allows you to code data into nodes, assign attributes to your interview transcripts, and then compare data by attributes (e.g., group, gender, level of organization) etc.

I completely agree with Marjorie on the use of the constant comparative approach, and can confidently recommend the Miles and Huberman text for what you want to do (they go through specific examples of different types of displays, matrices, etc. that are very well suited for cross-site comparisons. Personally, however, I favor Atlas.ti over nVIVO for computer-aided qualitative data analysis -- it makes fewer assumptions about the structure of your data so you can set it up however you need to make the comparisons you want. And you can use it for coding, querying those codes and quotations, visualizing your data in multiple ways (e.g., networks of codes), and can also export to HTML, XML, and SPSS if needed..

Have a look at the method of

Treleaven, L., Cecez-Kecmanovic, D., and Moodie, D. (1999) 'Generating a Consultative Discourse: a Decade of Communication Change', Australian Journal of Communication , 26, 3, pp. 67-82

Cecez-Kecmanovic, D., Moodie, D., Busuttil, A. and Plesman, F., (1999) 'Organisational Change Mediated by E-mail and Intranet - An Ethnographic Study', Information Technology and People , 12, 1, pp. 9-26

Phenomenology will allow you to find what is common, the essence.

Phenomenography will allow you to find what varies

Saldana, The Coding Manual for Qualitative Researchers will help you do the analysis and decide how you will analyse.

The most important issue, before comparing is that your predefined categories of analysis are completely builded. If not, you can´t apply any single analysis or comparation (e.g.: one or your categories must be organisational hierarchy and its respective levels, and another one must be common values, within it's own value definitions, to compare values between or among levels or hierarchy, even among companies or economy sectors). I hope it could help you.

Javier many thanks for your comments. The framing you have outlined gives me a really good start alongside the codings I am preparing.

Thanks Marjorie - I have picked up a copy of Miles and Huberman. It looks really useful, thanks. I have access to NVIVO so will look at setting that up for the comparisons.

That's the point, JAVIER GÓMEZ-DÍAZ. To be practical: better to invest time at the begining, defining dimensions of analysis (values, attitudes and communication seem to be the three wide concepts to "brake" in coherent pieces). Those piecese should give you your codes for analysis, in different levels of abstraction. Your main codifying process should be "top-down", but you will be able to adapt the code system, to fit your case study.

Thanks J Andres Dominguez - I have now begun following the points outlined by Javier earlier but your comments are very timely. The initial concepts I am framing my analysis around will be hierarchy, values, attitudes, motivation and communication. This should support some of the analysis I wish to develop so that, for example, I can compare attitudes to providing care or to older people and map these across the organisations by hierarchical position and within an organisation via the hierarchy. The methodology would then be combining (possibly) content analysis and discourse analysis to provide the data.

The only concern I then have is - how much time is this going to take? :-)

I totally agree with this iterative approach. It is fully connected to the grounded theory, it is inductive in that the findings are derived from the data and deductive in that the concepts and established links are interpreted in the light of that already exists in theory. Thus, it is possible to develop analytical frameworks, frameworks and even theories based on empirical research, which consists of a process of interaction between theory and concepts, and on the other, between observation and interpretation.

I also understand that the comparative study, especially in situations of multiple cases, It has gained a lot of strength with the use of the QCA, especially because it provides a systematic way to study phenomena that bring together aspects such as asymmetry, equifinality and causal complexity.

I have begun plotting out the first codings for the analysis. Having had a good read through of Miles, Huberman and Saldana I am aware that, while they discuss a first and second coding I may have to consider a third coding as well. J Andres Dominguez has given me a good starting point with hierarchy and my initial thoughts on the related concepts for my study should make the comparison easier at the start. The second coding would enable me to gain a more detailed analysis as is usual. The third coding I am considering would be of features in the language used relating to the differences between the organisational bases. Data collection begins in April - we'll see how the analysis stands up then!

Thanks for all the helpful comments.

I recommend an ethnographic observation and interview-based process in which you search for patterns of similarity or dissimilarity from both the interpersonal conversation of the interview as well as the often more informative practice of discreetly (creepy, I know) watching people's body language, behaviors, and habits they engage in alone or with a small group of peers.

Hmm interesting Phil. I have been considering some form of annotative record of body language and behaviours during the interviews but without some way of attaching it to the timings I am not sure how helpful it would be. I am also looking at passive non-participant observation of the care staff as an additional data capture exercise, again with a coding annotation to record it.

I will have to give some thought to the ethical considerations of observations of staff interacting with people who may not have the mental capacity to consent to an observation...

You could review some of the studies here and see if there are any comparable research questions: http://www.compasss.org/. QCA works very well with medium-N populations (8-25) and qualitative, categorical (or categoricalizable, if that's a word) data.

Hi Amy, thanks for the link. I have decided to use NVivo for the analysis using the codings to facilitate between cases analysis as well as within case analysis. I will post (probably much later) when I have made progress.

Immediately following an interview, you must write / record your thoughts, impressions etc of body language and your inferences. In constructivist ground theory methodology these journals become part of the primary data and will help to refresh your memory as you validate the accuracy of the transcripts. Be careful of becoming locked into your initial coding categories, since they might blind you to a more interesting and profound relationship. Again, keep journals about your thoughts as they develop, including citations if your new thought is triggered by something that you read. This will simplify the writing process when you turn to completing your work.

Your theoretical framework will lead directly to the questions that you ask which, in turn, will dictate the methodology for collecting & analyzing your data.

Your description makes it sound like you are interested to learn of the effectiveness of the communication process between hierarchical levels, so, grounded theory methodology is appropriate. A theoretical framework might come from the discipline of communication. But, you must be clear what your research question(s) are before selecting the methodology.

Hi Charlotte

Thanks for these pointers. I have been making notes and keeping journals all the way through so far - keeping them in some form of order is proving tricky so I am trying to order them into an outline structure of my thesis (e.g literature search , methodology, fieldwork, etc.).

While communication is a part of the cultural transmission I am trying to identify how it is expressed in the physical attributes of the care setting, the attitudes of the interviewees and practice in the care setting. I have looked at grounded theory but as the overall theme of my research relates to 'wicked problems' I am using a mixed method approach for within and between case analysis.

Thanks for the pointers.

Very nice discussion! Thanks all!

As a final comment on this discussion, earlier I had commented on having a two-level coding process.  After all data capture I eventually settled on a single coding related to the theory base of the research and then used Thematic Network Analysis as the tool for extracting the larger themes for the discussion.  Thank you all for your helpful comments which have been guiding my thinking all the way through.