Sunday, October 2, 2011

Integrating Numerical and Non-numerical Data: Qualitative Quandries

I have been struggling with the issue of getting out of the quantitative/qualitative conundrum...figuring out how qualitative research fits in a world of big data, micro-applications and new forms of data collection, etc., etc.  This morning I came up with a diagram that begin to help me 'see' what I have been thinking about. 

In this diagram:  above the mid-line represents Numerical Data, and below the mid-line represents Non-Numerical Data.

Both numerical and non-numerical data are divided by forms of data created by the researcher and forms of data created by others (the background/context/literature of a study). 

There are four inter-linked quandrants:
Numerical Arena
1.  Upper left:  Numerical data created by the research dominates.  The furthest point out to the left is "Big Data"  where millions of data points are generated by cell phones, telephone bills, tweets, etc.  As you move closer toward the middle you have larger surveys...scaling down into smaller surveys...At the center point as you cross over you have descriptive statistics created by simple counting of the researcher.

2.  Upper right:  This area represents numerical data created by others.  This is numerical data that serves as background or context for the work of non-numerical researchers.  Increasingly we have access to a broad range of statistics that we employ to describe the context of our work.  

Non-Numerical Arena

3.  Lower left:  This is non-numerical data created by others:  This is "THE LITERATURE".  It's not on the chart, but probably pure theoretical work is someplace to the left of meta-studies...coming down to studies on specific projects. Numerical and non-numerical researchers turn to literature in their field as a kind of data.  Using qualitative data analysis software has made me more and more aware that this is a living, breathing form of data on a par with an interview or an observation. 

4.  Lower Right:  This is non-numerical data created by a specific researcher.  This is where qualitative researchers and their studies live and breath.  Close to the mid-point are non-numerical studies that have structure and variables much akin to numerical studies...and one moves out to the right from there to places that draw more and more on emic/insiders perspectives.  Observations...take you outside of the words you have chosen in the organization 'they' have created.  Non-numerical studies range from those that are short-term and semi-embedded in the environment of 'the other' those that are totally embedded and long term in duration.

One critical question that this diagram raises for me (actually re-raises for me because I have been coming back to it a lot in my head) and that is--What does non-numerical analysis really consist of?  How are we making those patterns or theories we think are at the heart of it?  As I work with coding in the Sexting project in which I am now involved, to be perfectly honest, a lot of my pattern making deals with...what do I have the most of...what do I have the least of...what category is heavy/rich/saturated?  what tags/codes should be there but I don't see it?  Now, when you are asking questions about more and less/absent or present--doesn't that imply a numerical perspective? 

OK--now some of you qualitative researchers may be saying--I don't do it that way.  I make my patterns in other ways...OK--how?  Can you describe what you are doing?  I am going to venture that more/less-present/absent is a starting point in which we are employing simplistic numerical concepts to launch us into meaningful places where we can find useful theoretical work.  This is why I had to draw each of the quadrants as a line that is shading/decreasing into the area of the nearby quadrants.

Disclosure:  I have been working this out in my head without paying attention to the growing literature on mixed methods.  Maybe this has been all hashed out already, maybe it is irrelevant to the emerging arguments about this issue.  Mixed methods people, tend to be closer to numerical discussions than I am.  From what I know of the lower right hand quadrant...and those who inhabit it...this has not be widely discussed. 

The discussion I am having here with myself draws a lot on what I have learned through working with qualitative data analysis software.  It is also related to the issue of inclusive and exclusive research approaches that I blogged about in September 2011.  Challenges to my secure vision of qualitative research are raising up from all around me.  

Enhanced by Zemanta


Norma said...

Hi Judith. Thinking about this: nearly all of my scientific research has been of the qualitative sort. Nevertheless, my conclusions, of course, were based upon "patterns" I saw in the qualitative data. Most mathematicians I know would say that all patterns are mathematical in nature. Being a not-so-great mathematician, but a really good mental synthesizer of patterns, I expressed my knowledge in a way that worked for me. No one complained.
I am far from understanding the formal methods of qualitative research; in fact, I had trouble understanding a recent book on the topic, yet I was able to make sense of my own project in a way that worked for the materials and methods used. I wonder how much the gathering and expression of data in qualitative vs. quantitative ways depends on the peculiarities of the researcher? (Einstein quote, paraphrased: if you can't explain it simply, you don't understand it well enough. Me: then you will explain it in the way you understand best) Anyways, I enjoyed reading your blog :-)

Judith Davidson said...

Norma: Thank you for your response and for sharing about pattern making. I guess I am wondering if I really understand qualitative research at this point...How do we make patterns out of different kinds of texts? Thanks. Judy