The concept of mixed methods has begun to take root around the world. There are arguments on both the qualitative and the quantitative side of the aisle when it comes to mixed methods, so we recommend that you check it out for yourself by looking to incorporate a few mixed method approaches into your project.
At Dedoose, we are unabashedly enthusiastic supporters of mixed methods. But, we know it might not be right for everyone. So, we have highlighted three things you can do in your project today that will help you explore if mixed methods can be valuable to you. These ideas include: adding a scale system to one of your codes, tracking demographic information for your participants, and using graphs to focus on code applications across sub-groups to discover new and, perhaps, hidden patterns.
Code Weighting: Add a scale to one of your codes
Let’s pretend that you are interviewing school principals for your research study and one code that you want to use is labeled ‘Leadership.’ You can start by excerpting all the pertinent areas in which each participant mentions or alludes to leadership. In this way, you are simply employing pure qualitative techniques. But, here is where it gets interesting.
Now that you have coded each instance, create a code weighting system for this code. For example, you could develop and activate a 5-point rating scale indexing ‘importance’—where a ‘1’ indicates that leadership is ‘not at all important’ and a ‘5’ indicates that leadership is ‘very important.’ You then have to think a bit more about how you will use the points on the scale. For example, what does moderate ‘importance’ look like in the context of the qualitative content. How is a ‘3’ different than a ‘4.’ An interactive way to approach this is to go through each excerpt tagged with the code ‘leadership’ once more and this time you can decide if the excerpt should be rated a 1 or a 5—really high and really low can be an easy place to start. Once you have gone through identifying your 1s and 5s you can move on to your 2s, 3s, and 4s. You can treat them like a separate group, simply identifying the remaining highs and lows and, PRESTO, you’ve developed a 5-point rating scale.
Essentially, it could look like this:
Principal #1 says, ‘I value true leadership in the classroom. That is why we have leadership training for all teachers and administrative staff. We are always looking for ways to improve. I am constantly learning about how I can be a more effective leader to my team.’ You would probably want to give this excerpt a 5 score. Another principal may say, ‘Leadership is not that important for my teachers. They can spend time on creating better lesson plans, but to spend time or money to grow as a leader is a waste.’ You would probably want to rate this excerpt a 1. Now that you have the extremes accounted for, you can perform the same function for the middle pile of excerpts. Ask yourself if each excerpt is a 2 or a 4… and so on.
Remember, in all the excerpts the word ‘leadership’ was explicitly mentioned or the concept of ‘leadership’ was alluded to. Yet, if you only looked at code frequency, you might assume that leadership is quite important to the principals in your study. However, the truth is that everyone brings sentiment to their statements this fact can color your study in exciting ways. Code weighting may help you find new and important findings as related to your research questions—the difference between really strong evidence and, eh, evidence that is good enough to tell a story.
Descriptors: Dive deeper into what ‘he said’ versus what ‘she said’
If you have been following our blog, you will now that we LOVE descriptors, but they can also be a bit confusing at the onset for qualitative researchers. If you are working on a purely qualitative study, you do not need to use descriptors. Yet, we always recommend that all users consider giving descriptors a try as they help organize information about the sources of your data (i.e. demographic information about participants).
When you are just starting out with descriptors, try pulling the basic demographics of your participants together and adding them in as what we call a ‘descriptor set.’ You first create fields that essentially ask the questions (i.e. gender, ethnicity, etc.). Next, you create descriptor value points that answer those questions for each case (i.e. female, Hispanic, etc.). We recommend using option lists or categorical variables/fields wherever possible. When you have descriptors set up in your project, you can go through an interview and reference these key demographics quickly and build great analytics into your project when you come to the analysis phase of your study.
At first, working with descriptors can be tough. That is why we created our descriptor article series to help you get started. We also recommend that you check out the video recording from our recent webinar on working with descriptors. And, there is always the user guide and the video tutorials. These resources should help you get started in no time!
Explore Mixed Methods Graphs, Plots and Tables: Bring those Data to Life!
So, your data is in. You even used descriptors! Great. Now what?
Now you analyze your data. Dedoose offers a variety of mixed methods charts, tables, and plots that will help you bring your data to life. You go into your research study with certain assumptions. Diving deeper with a mixed methods approach allows you to validate or invalidate those assumptions easily and quickly.
When you go to the Analyze Workspace in Dedoose you will see file folders on the left hand side of the screen. One folder is labeled ‘Mixed Methods Charts.’ This is where you can go to explore different ways to look at your data.
You can start looking at how codes have been applied, broken down by descriptors. Perhaps it is interesting to note, for example, how female principals discuss leadership versus how male principals talk about leadership. And, because Dedoose has ‘hot’ charts, you can click on any bar or plot point and open up the coded excerpts behind the visualization.
We love the bubble plots as well. These plots quickly expose complex multi-dimensional relations between the applications of codes across sub-groups and facilitate the identification of important patterns in respondent opinions regarding the research questions.
More Mixed Methods?
Of course there are more ways to bring mixed methods to life – from importing surveys with open-ended and close-ended questions to running advanced data set filters looking at how participants in certain ethnicities that also talked about a certain code with a 4 or 5 rating discussed other themes – but we thought we would keep it as simple as possible to start. Resources GALORE:
For more general information on getting started with mixed methods we recommend that you visit the WT Grant Foundation website.
Ask the Audience Time…
What are your biggest challenges when it comes to implementing mixed methods? Fears? Excitement? Let us know in the comment field below! We are always listening to our awesome users, so the more you share, the more we learn, and the better help we can offer.;