What is Mixed Methods? A brief introduction

Mixed Methods is a methodology that attempts to breach the qualitative-quantitative divide by integrating aspects of both approaches. However, both methods are not just juxtaposed, but rather used to create combined results. In this, Mixed Methods Research (MMR) often follows a pragmatic doctrine that puts the research question above epistemological or methodological considerations.

This semester I attended two conferences focusing on Mixed Methods research (MMR) which is the central methodological paradigm behind Qualitative Data Analysis. Here, I briefly present my personal grip on the subject. Most is built on two talks held by Pat Bazeley (compare: Bazeley 2017) and supplemented by a book by Udo Kuckartz (2014b). So let’s dive right in and ask ourselves:

What is Mixed Methods?

Typically Mixed Methods is seen as a methodology integrating both qualitative and quantitative approaches within one research project. The movement is around since the 1980s when sociologists tried to resolve the disagreement between the qualitative and  quantitative paradigm by combining both into a third way (Kuckartz 2014b, 27-28). Of course, there are disagreements on in details, for example Johnson et al. (2007) list 19 different definitions. Still the overall picture is clear enough to be useful.
A simplified juxtaposition of the three paradigms. Illustration created by Scrached (cc by-nc 3.0).

Qualitative methods are often constructivist and conduct exploratory research. Quantitative methods, on the contrary, are rather (post)positivist and focus more on hypothesis testing. Here, Mixed Methods takes a stand in between. In this MMR is often strongly pragmatic. Whatever is most useful for answering the research question should be done.

“Judge available data by its relevance rather than its form!”

said Pat Bazeley, a strong advocate of this “pragmatic perspective” in the keynote of the MQIC2019.  Commonly, this means to combine the strengths of qualitative and quantitative approach, although Bazeley – being a pure pragmatic – does not even strictly insist on that.
A less controversial stance might be the definition provided by Udo Kelle:
“Mixed methods means the combination of different qualitative and quantitative methods of data collection and data analysis in one empirical research project”. (Udo Kelle quoted in: Kuckartz 2014b, 31)
However, I would argue that a quantitative data set is not necessarily needed to begin with. Instead – as many researchers do – a quantitative data set can be created from qualitative data in the course of analysis as well. And after all, it is called Mixed Methods and not Mixed Data.

Validation and Triangulation

Another approach that is closely related to Mixed Methods is called (method) triangulation. The term comes from trigonometry and describes the process when the location of a point is determined by forming a triangle.
Illustration of triangulation in the 16th century. Wiki commons. CC0.

The basic idea  here is, that combining different methods (or also combining two theories, two data sets or two individual researchers) is useful in validating the results. Here, the methods combined do not necessarily need to cross the qual-quant divide. Such multi method research designs can, for example, include the combination of narrative interviews, biographical analysis of diaries and focus group discussion in one study.

Of course, combining different methods makes perfectly sense in a field with many formalised methods. However, when few explicit methods exist in your field, the idea has little appeal. In the humanities research methods are often modular, eclecticistic or self-made. One could argue that they contain an in-built method triangulation because they often use data from many different types of sources. However this eclecticistic methodlogy is too often limited to either picking qualitative or quantitative approaches.

Benefits and Drawbacks

Qualitative and quantitative methods have their specific strengths and weaknesses. Combining them will create a benefit that is more than the sum of its parts, this is the core assumption behind Mixed Methods research. With a mixed methodology we might gain insights into an extraordinary case within a quantitative study, or compare the results of our specific case to a more general picture. Unlike in triangulation, Mixed Methods research aims to arrive at a fuller, more complex picture, instead of mainly validating results.

Still, there are limitations to Mixed Method approaches. Most challenging – especially as a historian – is the data side of it. If we do not have quantitative and qualitative data on the same cases, our options will be strongly limited. Furthermore, few researchers are brought up in a truly Mixed Methods field. Instead, they are more often “Quants” that stray into “Qual” waters or vice-versa. The danger here is, that researchers integrate methods that they  poorly understand and create results that are not methodologically sound.

Study designs and data integration

There are many potential research designs within the Mixed Methods methodology that are too numerous to cover here (see for example Kuckartz 2014b, 57-97).  In any case, the goal is not a juxtaposition of two methods, but instead a truly mixed methodology that integrates quantitative and qualitative thinking. Both strands should be influencing each other throughout the process, instead of creating independent results.

The added value is commonly created at the topic level. Notes (memos) and also research chapters should not be split up into “qualitative” and “quantitative” parts, but instead the permanent strive to answer the research question and the growing knowledge about one subject, naturally integrates the results from different approaches. Ideally, in the end the results of a strongly mixed methodology can no longer be traced back singularly to either a qualitative or quantitative strand of the research conducted.

For the humanities in particular, several Mixed Methods scenarios are possible: For example, one could use iterative cycles of close and distant reading on the same (qualitative) corpus. Another research design might combine demographic and economic data with narrative accounts. A third might analyse a vast quantity of paintings with the help of an algorithm and combine them with qualitative inquiries into the most typical cases in each cluster.

Conclusion

Overall, the “pragmatic” look of MMR combines quite well with the strategies that many humanists (digital or not) choose. I found it illuminating to think less of methods, tools and data, but pragmatically about the research question. In the end, the research question should drive our methodology, not the other way around. Or as Pat Bazeley put it: “Questions are not qualitative or quantitative, they are just questions!” I think from this pragmatic look, we can learn to strive for more openness in combining different research strategies to gain a fuller picture on the questions we wish to answer.


Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.