The tenth and final workshop for FIT6021 looked at Mixed Methods, an approach in which researchers use a combination of qualitative and quantitative data.
Tashakkori and Newman (2010) describe the seven main reasons for using mixed methods research: complementarity, completeness, development, expansion, corroboration/confirmation, compensation and diversity. There are several advantages to using a mixed methods approach; it can overcome or compensate for any defects or weaknesses in either method, ie. having the best of both, covering any gaps, as well as just providing richer or more detailed analysis (to paraphrase the lecture notes, the quantitative data can tell you what is happening whereas the qualitative data can tell you why it is happening). A researcher could use the second type of data to confirm what has been discovered with their first set of data.
Though using mixed methods research can be beneficial, it does have some drawbacks. Firstly, it can be difficult to decide if using mixed methods is really necessary. If the researcher does choose to use it, they must take into account the added time and resources that will be needed (especially if they are conducting the different sets of research sequentially). This will not only slow down the research itself, but it could prove problematic when it comes to publishing their findings (eg. do they just publish one set at a time, keeping their work 'up to date', or do they wait til it is complete and risk having its timeliness reduced?). Also, to conduct mixed methods research, the researcher must be skilled in the use/analysis of both qualitative and quantitative data or, if they are not, must work with other researchers (which can cause further problems if they come from conflicting paradigms; although sometimes it is good to have both a positivitist and an interpretivist etc looking at the data, as one may see something the other has missed) (Hesse-Biber, 2010). Integrating the two sets of data (into matching formats) for analysis could also be a challenge depending on the way the experiments/observational exercises were set up (Cresswell, Clark and Garrett, 2008). There is also the chance that both of the methods used could produce conflicting results (this may not necessarily be a bad thing, but it would be a problem if one set of data was meant to 'confirm' or support the other).
A lot of published articles we looked at (both in class and while we were researching our group presentation) that use mixed methods seem to use it in an almost 'cursory' manner, ie. the majority of their experiments and analysis will be conducted using one method, followed by a briefer set of research conducted in the other method (often also described in less detail). While I don't necessarily see this as a problem (some topics naturally lend themselves more to one method than another), it is still important that both types of research be conducted rigorously and justified and reported accordingly, otherwise there is little value in conducting the 'supporting' research at all.
Though it is too early to know for sure, it is possible I will use a mixed methods approach in my research. However it will probably focus more on the qualitative side rather than quantitative; most of my research will be based on user testing and and observations and probably talking to potential users before the research/design stages begin, though interviews and questionnaires will need to be used as well. At first I thought that using mixed methods would be a useful way to confirm my data and make it more plausible, however the more I learned about it, the more I realised it had the potential to become very messy if not handled correctly, as discussed below.
I would have to consider is whether I would conduct the research sequentially or concurrently. Concurrently would be difficult simply because I will be conducting the research on my own, and managing my time between the two and also shifting my 'mindset' back and forth would be challenging (not to mention my general dislike of statistical analysis, which I would have to learn how to do). Sequentially could prove to be troublesome as I would have to collect and analyse the data from one lot first (which could take months) and then collect and analyse the data from the second lot (again, potentially taking months). Depending on when I get to the point where I am ready to begin conducting the tests etc, there is a good chance I would run out of time to do both, or at least run out of time to properly write up the results at the end. Therefore, though it could be a useful method, it may not be practical for me in terms of time constraints and resources. I will have to wait and see.
Mixed Methods Online Resources
<resources are listed on Wiki set up by Danny, requires login to Monash Google account>