About methodology - part 1/many
(On my thesis construction) OR (why I'm creating a conceptual framework before data mining)
This post is mainly on the “why” behind the steps I’m taking in my PhD. Many (?) other posts on the matter will follow over time as it is keystone in scientific research. But pinky promise, I will try to keep it illustrated and not too theoretical.
I’ve been very awkward from the start of my PhD since my advisor’s indications seemed counterintuitive. Just recently I might have found their meaning, hopefully. Here, I wish to share this lil’ epiphany with you.
The “circularity” problem
I’m doing a PhD in management and in the lab, the scientists usually go with “qualitative” methodologies, although the director of the lab prefers to use the term “comprehensive” because you can also incorporate simple and robust statistic models, like the Qualitative Comparative Analysis.
In those approaches, one of the main problems is “circularity”: you can find any data that corroborates a theory and your theory will make your data look legit in return. However, in the process you might have (consciously or unconsciously) discarded data that was challenging the theory. It’s a very sensitive issue as there’s a lot of room for biais while gathering data. The “solution” is to collect your material independnetly of any theory. Easier said than done…
In a nutshell, most of my PhD peers are expected to go and observe the field (for instance a business unit in a firm) any theoretical construction in the beginning. Then, back to the lab, they will analyse and find different concepts helping the interpretation. Confronting them is part of the “fun”. Eventually, this work results in the evolution of said concepts and helps to push forward the boundaries of knowledge.
My crux
As my advisor expects, I’m doing a lot of theoretical research from the start, defining the notions and the scope of my study, advancing concepts and so on and so forth. Yet, I still haven’t collected any data! Am I jumping into the circularity pitfall? That’s been troubling me for a long time.
Plausible explanation
The more I’m deepening my topic (AI for managing customisation in the heath-care path), the more I discover intricacies: what’s exactly AI? Should I also include Data Science? What does “customisation” mean? Etc.
The tipping point has been when I become aware that the complexity of the subject could be shallowed whole by the fast evolving filed of AI, the many discussions and problems it generates and so on. What I’m trying to observe is still relatively new, uncertain in its boundaries and multi-dimensional; it’s very easy to be lead astray by all the glittering and be led to collect data that will be useless to my problem1.
Hence, the importance of creating a firmly and solid basis, not in order to interpret the data, but to pick the relevant information which will then be analysed through other conceptual frameworks.
In other words, my job now is to assimilate the different conceptual notions of my problem in order to “see” and differentiate what’s not part of it. Then, I will be able to gather the data and challenge other concepts in order to make sense out of it. But the theory I bring up right now will not intervene anymore. There shouldn’t be a circularity risk at this point since I don’t try to challenge or improve the initial concepts.
I think this approach is not generally used because, in most cases, the topic is very straightforward so there is no need to define it further.
Yet…
This explanation might be wrong; it might just be me trying to create any possible sense to justify my actions. Isn’t this what “circularity” is about? I just look at the things that can justify what I’m doing without taking in consideration what could prouve me wrong. Despite my earlier explanation, I still have lingering doubts.
« Attendre et espérer ! » (to wait and hope) is the last message left by Le Comte de Monte-Cristo in the eponym novel by Dumas. Incidentally, that’s how I would define the path of the researcher. It’s about advancing surrounded by uncertainty, without being sure you actually advance. And then, maybe, at some point you will find something. Or not!
Note: hope you enjoyed the lecture at least as much as I had fun writing it. I get it’s not very easy and I might edit it back to make it more accessible, if I can improve my lacking skills meanwhile. Thanks for helping with your feedbacks :)
Yes, it could answer another question, but trying to find a question to your answer is a slippery slope to circularity.