The research is using mixed methods, framed by a pragmatic theoretical framework, with a convergent design using triangulation. In using this design, it was noted that some researchers code qualitative data in order that there is some 'dialogue' between the two methods used in the design (Hesse-Biber, 2012). I believe that there will be some convergence between the two sets of data but not all (which I expect). In order that I am able to explain these findings, and not risk my paper being thrown out, how can I interpret this?
Triangulation is an ambiguous term. The current trend is to use it for strategies of analysis in which you investigate your material from different angles and with different methods. That would be an argument for not "quantitizing" your qualitative data but rather do the integration of quan and qual during the final stage of interpretation. In this case, I think that some divergence is rather a strength than a weakness - it's like an ex-post argument for using multiple methods in the first place.
If you want an information that is more tailored to your specific situation you'd have to provide more information: what kind of research question, what kind of material, what kind of audience are you addressing ...
I am not one for following trends but rather a design that suits the research question really, as otherwise I may end up with flawed research in terms of methodology.
The research is centred on a sub-group of the population focusing on fear of crime, social and psychological factors and how this impacts on social intergration (actually is a bit more complex than that). The material is a generic questionnaire using likert scales and, running in parallel to this, are in-depth interviews. While one method tests individual factors, the other looks at experiences of the individual within a social context, so to speak. Therefore, in order to bring the findings from each method, I guess I'm asking the question as to what ways will I be able to allow this to 'speak', even if they converge or diverge. This is new experience to me but I have searched for answers to this question but they are quite varied; they are dependent on a number of factors...time, experience, methods, the question(s) and so on.
But terminology follows trends - and the term triangulation at the moment is understood in a specific way. That's what I was trying to say.
Since you are using two methods for two different aspects of the same phenomenon I think it is completely reasonable to have to separate strands of analysis and discussing them in relation to each other afterwards. After all, there are reasons to believe that either method might not really be able to achieve what it is meant to, so the separate analyses can also be useful for validation (which is more difficult if you quantitize your qualitative data). Divergence is not necessarily a bad thing, especially if it means you have a richer picture.
What would be the purpose of quantitizing? If you have a clear idea about this, there is nothing wrong with adding a few variables from coding of your interviews to your questionnaire dataset. You could then check, for example, if certain kinds of narration lead to certain response patterns in the questionnaire. But from what you write I would see that only as an additional option and keep the two strands of analysis apart.
Many thanks for your answer to my question. Could I possibly be so rude as to ask another one? Who, in your opinion, do you suggest I should read?
You mean on mixed methods? There is a recent thread on this:
https://www.researchgate.net/post/What_are_the_best_books_authors_to_consult_in_order_to_get_more_than_a_fair_understanding_of_mixed_method_research
I am not aware of more specific literature dealing with the very question you have in mind - the more programmatic texts by Ongwuebuzie, Tashakkori, Teddlie, Creswell etc. are "state of the art".
I agree with Kenneth Horvath that the term "triangulation" has become so blurry that people prefer not to use it. Instead, convergence is currently the preferred term for a direct comparison of whether both a qualitative and a quantitative method produce equivalent results.
If this is indeed the goal you are pursuing, then the most common way to do it is indeed to quantify the qualitative data. One source on this topic is:
Sandelwoski et al. 2009 On Quantitizing. Journal of Mixed Methods Research, 3:208-222.
I personally feel that searching for convergence is a rather limited goal in mixed methods, and that it is particularly problematic when the two sets of results "diverge." I also would say that creating a "dialogue" between methods is something that only the researcher can do -- the data won't speak for themselves.
I found the following article which may be helpful to you. It is published "online first" at the Journal of Mixed Methods Research:
Kate R. Fitzpatrick. Points of Convergence in Music Education: The Use of Data Labels as a Strategy for Mixed Methods Integration.
Basically, she determined a set of quantitative results and then examined the qualitative data to find confirming, contradicting, and mixed results there.
You might also look at:
Cline et al. (2013) Factors Influencing RNs’ Perceptions of Quality Geriatric Care in Rural Hospitals. Western Journal of Nursing Research, 36(6):748-768.
They created a matrix to cross-tabulate the findings from the quantitative study with the themes from the qualitative.
- Badges
- Science topic
- Similar topics
- Phytochemistry
- Extracts