I just ran across this response I wrote about a year ago to Rennie’s Grounded Theory Methodology as Methodical Hermeneutics. I think it is a nice little viewpoint that phil sci should take more seriously.

Pierce argues that abduction is a valid form of reasoning/logic in qualitative research. Abduction is the mode of inference in which a hypothesis is formed to explain a finding. The abduction can then be shown to be consistent through the use of induction. Rennie says, “…the qualitative researcher insists on a number of instances, in one form or another, supporting the abduction before concluding that it speaks to the text as a whole.” I claim that abduction is not a valid form of inference in qualitative research by using one of the standard examples from quantum mechanics: a particle randomly placed in two boxes.

Suppose you have two boxes, A and B, with nothing in them. Then you have a way of randomly putting a particle into one or the other. You have no knowledge of which box the particle has been put into, but you would like to find out. As a qualitative researcher, you decide the most direct way would be to look into A. If you don’t find the particle in A, then it must be in B (you could look since it doesn’t change the result at this point). You have a result that the particle is in B. Now, the experiment was to determine which box the particle was in, not which one it is in. As a qualitative researcher that believes in abduction, you abduct that the particle was in B the whole time.

Pierce does not grant that you have a valid conclusion yet. You must validate this abduction inductively. There are several ways of going about this. You decide to repeat the experiment 50,000 more times, each time you wait less and less time before you check the box until you are checking only nanoseconds after the particle is put in one of the boxes. Each time you still get the same result. Now you have 50,000 other people do the experiment to make sure you don’t have subjective influences on the result. They all come up with the same result. No matter how you continually validate the abduction, it is wrong precisely because the experiment is being done qualitatively. Thus, abduction is not valid while doing qualitative research.

This abduction is the most obvious and common sense oriented abduction one could make. I see the particle in box B, therefore the particle has been in box B the entire time. In reality, the particle was in neither box A nor box B until you checked. The particle was actually a probability wave that was distributed over both boxes. Without quantitatively analyzing what is going on, it is impossible to determine this result. This is a clear case in which, despite inductively validating the abduction (perfectly every time), the abduction is still incorrect.

One objection to this counterexample is that Rennie is using Pierce’s theory of abduction in the context of qualitative psychology and human sciences not physics (although Pierce does use it in the natural science sense). One may argue that this “quantum weirdness” would never happen at a macroscopic level. This is untrue by the macroscopic form of the experiment known as Schrödinger’s cat. This would also be to miss the point. This case provides a perfect extreme: no matter how many experiments you perform, you will always validate your abduction. This will never happen in the human sciences, so shouldn’t we be even more skeptical? Also, it shows the tendency to label some abductions as obvious and not in need of other inductive evidence. It is almost absurd to think about performing all those extra experiments just to validate that the particle was in one box or the other. This mistake would be very easy to make at the macroscopic level.

Lastly, I would like to point out that although qualitative research does not provide the type of research necessary to validate abduction, I make no claim about its use in quantitative research.