Hello Maha, 

It seems that you are talking about two different types of design within MMR (mixed methods research). Usually, having few qualitative questions within quantitative questionnaire does not constitute a mixed methods study of the embedded design; rather it is a quantitative survey with qualitative elements.

To be able calling the study MMR, you need to have both qualitative and quantitative parts of the study being of equal weight. That is one of the basic criteria for quality mixed methods studies. In the convergent parallel design, you collect in parallel both quantitative and qualitative data (say, you do qualitative interviews and an online quantitative survey) at the same period of time. Perhaps, qualitative data from your non-participant observations and the quantitative questionnaire that you mentioned, together, could better fit the definition of the convergent MM design, if both parts have equal weight. This type of design has its certain methodological limitations, though.

I would recommend the following reading for quick reference of MMR designs and their pros and cons:

Creswell, J. W., & Plano Clark, V. L. (2011). Designing and conducting mixed methods research, (2nd ed.) Thousand Oaks, CA: Sage.

Hope it helps. 

Anna 

Hello Maha, 

It seems that you are talking about two different types of design within MMR (mixed methods research). Usually, having few qualitative questions within quantitative questionnaire does not constitute a mixed methods study of the embedded design; rather it is a quantitative survey with qualitative elements.

To be able calling the study MMR, you need to have both qualitative and quantitative parts of the study being of equal weight. That is one of the basic criteria for quality mixed methods studies. In the convergent parallel design, you collect in parallel both quantitative and qualitative data (say, you do qualitative interviews and an online quantitative survey) at the same period of time. Perhaps, qualitative data from your non-participant observations and the quantitative questionnaire that you mentioned, together, could better fit the definition of the convergent MM design, if both parts have equal weight. This type of design has its certain methodological limitations, though.

I would recommend the following reading for quick reference of MMR designs and their pros and cons:

Creswell, J. W., & Plano Clark, V. L. (2011). Designing and conducting mixed methods research, (2nd ed.) Thousand Oaks, CA: Sage.

Hope it helps. 

Anna 

Dear Anna,

Thanks Anna for your response.  What about the documents I collected afterwords.  I mean When I finished collecting the qualitative observation and the quantitative questionnaire, I collected some documents from the three settings I collected my data from.  Does convergent design cover this as well? I will read more about it?

Regards,

Maha

I think Anna has made some valuable comments.  I often call " mixed methods" "mix-upped methods" because  qua research design, each method is totally different; one is inductive, the other is deductive.  We can't pretend we're doing "mixed" methods: it's like oil and water: one sits on top of the other.  However, there is no reason not to collect data using both approaches, side by side or sequentially.  The term "mixed methods" is used either by quantitative researchers to colonize qualitative researchers, or by qualitative researchers to ingratiate themselves in the eyes of the colonizer. 

Thank you indeed for your response.

Regards,

Maha

Maha - what is your sample for the questionnaire, the observations, and the documentation? Have you analyzed the data from the questionnaires and the observations before collecting other data from documentation?

Will - thank you for your comment.  I certainly appreciate your perspective (and many scholars would concur with your statements). I do believe, however, that mixed methods research can produce insightful findings that either quantitative or qualitative methods cannot provide when used on their own. This could work on three conditions, in my opinion: an effort is put to design and implement a quality MMR (meaning, it is important to follow the guidelines of quant research for the quant part, and qual research for the qual part), while simply mixing whatever does not really work well; second - research question should guide the method - MMR could work best for exploratory studies, when little preliminary data is available about the phenomenon to build upon, but one is looking into both depth and breadth of information; and three - the most challenging: one should invest an effort into integration and corroboration of the findings from both qualitative and quantitative streams. Integration techniques have been developed and continue to improve within the research community, based on studies reports. Journal of Mixed Methods Research provides an ample of useful how-to examples.

As to the different approaches to knowledge creation - deductive vs. inductive - we do this in our day-to-day life without paying attention and do not divide. We collect information about the world around us sometimes using numbers, other times using words, still other times using both numbers and words - depends, on what works best to answer our queries. We do it naturally, we process the information, arrive to conclusions and proceed further in our knowledge quest. So, why cannot we do the same in academic or community  research? Perhaps, it is just a matter of our own perception?   

Dear Anna,

Thank you so much for your comments.

Regarding your questions:

1. What is your sample for the questionnaire, the observations?

The sample is 117 patients collected from three hospitals.  So I attended the consultation with the female patient and her caregiver as a non participant observer and at the same time I recorded the medical consultation .  After the consultation finished, the patient filed a self-administered questionnaire and handed it to the receptionist. 

2. What is your sample for the documentation?

For the documents, I collected some documents concerning the phenomenon I  am working on.  This happened within the data collection phase and after I collected the data.  These documents were collected from one hospital, another institution in the same city and from Ministry of Health in another city.

3.Have you analyzed the data from the questionnaires and the observations before collecting other data from documentation?

In terms of the analysis:

I analyzed the observation data first, then the questionnaire data afterwords.  Up till now I have not analyzed the documents; I might use them in the first chapter to describe the context of the study.

I have been reading the mixed methods designs in Creswell and Clark's book and I have been confused between Convergent parallel and embedded design (I have 3 open-ended questions in the quantitative questionnaire) or just use convergent parallel as a mixed method design for my study?

Best regards,

Maha

Hello Maha,

In the convergent design, you should have two separate sets of data (let's say, observations and consultation records as a qualitative strand, and quantitative self-report questionnaires as a quantitative strand) of equal importance, collected independently, analyzed independently, and then results from both strands are merged during the interpretation phase.  These two strands should not be connected or influence each other at the time of data collection or analyses.

In the embedded design, you would use one method as a leading one such as qualitative method (say, observations and consultations), and then another method as supplemental (quantitative questionnaires with open-ended questions) to enhance the overall (qualitative) design of the study in some way. You would put more emphasis on the leading method and use the other only to supplement.

In relation to your questionnaire, how many questions do you have? Is it a quantitative structured tool with three open-ended questions? Or, is it a brief questionnaire that contains mainly open-ended questions? If open ended questions constitute a small chunk of the tool, it will be a quantitative questionnaire. Having some open-ended questions in the quant questionnaire by itself does not meet criteria for mixed methods.    

Hope it clarifies.

Regards,

Anna

For the sake of simplification,  in mixed methods design, two most common designs are sequential and concurrent. The former refers to the design in which one paradigm (e.g. QUAN) is followed by another (e.g. (QUAL). Specific examples are: questionnaire survey with follow-up interview or the other way round, or questionnaire is facilitated by preceding interviews. This type of design is chracterized by the dominance of one methodology which is accompanied by the other. The latter refers to the design in which  QUAN  and Qual are used separately or in parallel manner. The common principle is one methodology should be informed by the other. From your description, I agree that it is NOT mixed methods but one methodology with some element of the other methodology.

Good luck.

Canh

As far as I know, documentation are treated as secondary data.

I agree with Anna that for the classic convergent design, the goal is to compare the results from two independent sources. This is amounts to studying the same thing twice, just to see if the two methods produce equivalent results. This is essentially what Donald Campbell et al. meant by "triangulation."

I don't think Creswell and Pland-Clarke  gave us a helpful label when they called something a "convergent parallel" design, because those two words mean the opposite of each other. In essence, they are describing a design where you keep the two methods separate (parallel) until you bring them together (convergence). This is basically a concurrent approach, because you are collecting both types of data at the same time.

In my opinion, however, the point of a concurrent design is more about how you use the data, rather than when you collect it. For example, you may have collected your surveys and observations at more or less the same time, but you chose to analyze the observations first. So, the real question here is whether what you learned from the observations affected how you used your survey data.

For example, if you basically want to do a one-to-one comparison to find out if they show they same results, then that is convergence. As a second option, if the questionnaires are supposed to add something separate and new to what you learned from the observations, then that is what Creswell and Plano-Clarke would call parallel-convergent. And finally, if the results from analyzing the observations influence your goals for analyzing the surveys, that is a sequential design.

So, don't worry so much about when the data was collected, and think instead about how you want to use the different things you learned from each method.

As for the term embedded design, this also comes from Creswell and Plano-Clarke and in practice it usually refers to a large program intervention that also includes a smaller qualitative component. I personally do not find this a particularly useful piece of jargon because it does not fit within the concurrent versus sequential distinction that Canh mentioned. In particular, things that get called embedded are nearly always QUANT --> qual sequential studies, where the needs of the dominant intervention lead to the us of a further supplementary qualitative study.

Maha Al-ayyash your posted question was well answered by Anna Azulai. Thank you Anna for sharing your knowledge and expertise in MMR

Best wishes Maha.·

Dear David,

Thank you so much for your reply.  I am personally  confusing between Convergent Parallel design and triangulation.  In my thesis, I have different research questions and thus I have used observation to answer 3 research questions and questionnaire to answer 4 questions.  My question here, do think that using the term triangulation is more focused than using convergent parallel design since my thesis is trying to answer different research questions?

Regards,

Maha

The term triangulation was originally introduced by Donald Campbell and his various coauthors to mean what we would now call something like convergent findings, i.e., producing highly similar results from different methods that have different strengths. Unfortunately, that specific meaning became blurred over time because people said they were doing triangulation whenever they were doing any kind of mixed methods. The classic citation on this problem is: Greene, J. C., Caracelli, V. J., & Graham, W. F. (1989). Toward a conceptual framework for mixed-method evaluation designs. Educational Evaluation and Policy Analysis, 11(3), 255-274.

That kind of triangulation requires using two (or more) different methods to answer the same research question, which is quite different from the design you are using.

The problem is that there is no agreed-upon terminology for what you are doing, even though assigning different methods to different aspects of the overall research question is a very common design. Probably the most frequent label you will hear is "concurrent" design, but that simply refers to when the data are collected (i.e., at the same time), not how they are used. 

In my own book on mixed methods, I refer to this design as "additional coverage," which matches the notation of QUAL + QUANT etc., while also capturing the underlying motivation of a division of labor that assigns different methods to different aspects of the research question.

D. Morgan, 2013, Integrating Qualitative and Quantitative Methods, SAGE.

Maha:

An excelent source to understand all these conceptual differences and procedures to present your  data as a MMR is the book:

Onwuegbuzie, A., Slate, J. R.,  Leech, N. & Collins, K. M. T.  (2012). Mixed reearch: A step by step guide.  London & New York: Routledge.

I hope this helps,

Yolanda

Hi Maha, 

Mixed methods conceptualization is still being developed, which is why the terminology can be quite confusing at times. In addition to great responses and various articles suggested above, I could offer a relatively recent published literature review on triangulation and its utilization by emerging scholars, which was written in an attempt to get ahead of this concept and its application in graduate research:

Azulai, A., & Rankin, J.A. (2012). Triangulation in Canadian doctoral dissertations on aging. International Journal of Multiple Research Approaches, 6(2), 125-140. doi: 10.5172/mra.2012.6.2.125.

http://pubs.e-contentmanagement.com/doi/abs/10.5172/mra.2012.6.2.125?journalCode=mra 

Hope it helps. 

Anna 

Dear Azulai,

Thank you so much for your article. 

Besr regards,

Maha

Hi Maha,

I am also doing a mixed methods research project and using a Convergent Parallel Design. I found the book - "Research Design" by John W. Creswell 4th Ed. 2014 very helpful and clear in how to recognise the differences between the approaches. Also I came across this document where John Creswell takes you through a step by step process to work out whether you are doing mixed methods research and if so what type - http://digitalcommons.unl.edu/cgi/viewcontent.cgi?article=1047&context=dberspeakers

I wondered if it would be helpful for you / for you to work through this with your supervisors to see if it helps you decide what fits your study the best?

Kind regards

Melissa

Melissa:

Thank you for your time and excellent advice. I truly appreciate it.

Best regards,

Yolanda

 Melissa

Thank you so much for your message.  

All the best,

Maha

I m using convergent parallel design in my thesis. How to write research questions and research objectives that denote the both sides qualitative and quantitative? 

how to overcome the unequal sample size problem in convergent parallel design? e.g. using snowballing technique quantitative samples reached 160, and only 25 of 160 sample agreed to participate in qualitative interviews. (sample is HIV patients)