Methods as theory

At least in linguistics, there is a general tendency to contrast theory with methods. This dichotomy is also reflected in the classification of academic papers as theoretical or methodological.

In this brief post I will argue that the theory/methods divide is epistemologically incoherent, and I will instead propose a taxonomy in which methods are subsumed under theory and are part of it.

1 What is theory?

The term theory is used with different senses in different domains. Here I will cover only three (see National Research Council 2008, Ch. 2).

In the popular usage, it is often just a synonym of hypothesis, idea, or hunch. For example, one might say:

The socialist party could win the next elections, but it’s only a theory.

Note that the use of only stresses the fact that the socialist party’s victory is not a determined outcome, but rather “only” a hypothesis.

In mathematics, a theory is a set of axioms (statements taken to be true) and theorems (statements mathematically proven to be true, usually derived from axioms and/or other theorems).

Finally, a scientific theory can be defined as a generally-recognised set of well-corroborated hypotheses that form an integrated model of one or more natural phenomena (National Research Council 2008: 32–36).¹

In the remainder of the post I will use theory in this last sense (for an in-depth overview of the structure of scientific theories see Winther 2021).

2 Are methods not theory?

At first sight, one might think that methods are not part of theory, but rather that they are just the means with which one builds a theory by gathering evidence. The following schematics illustrates this relation.

Under this methods-for-theory view, a set of methods (from experimental design, through recruitment, data processing, to statistical modelling) is secondary to the theory. Methods are seen as accessory to theory building, but not part of the resulting theory.

Rather, I argue that methods are an integral part of theory, and dissociating them from theory can lead (a) to a defective formulation of the latter, and (b) to the detrimental disincentivisation of resources dedicated to the refinement and popularisation of methodological advances.

3 Methods as theory

A different conceptualisation of theory sees it as the interaction between concepts and methods. Both components are theoretical, in the sense that they are the building blocks necessary to build a theory in the first place.

The methods-as-theory view is represented in the following graph.

With concept, I mean any statement, hypothesis, law, rule, constraint, principle, observation that is part of a particular theory.

For example, Heisenberg’s uncertainty principle is a fundamental concept of quantum mechanics.

The use of a quantum computer that manipulates photons to answer a research question is, unambiguously, a method of quantum mechanics. How the experiment is brought about by researchers and how the results are used for inference, together with many other factors related to conscious decisions and less conscious biases, all influence the resulting scientific contribution.² This “subjective” aspect of science has been brilliantly labelled the garden of forking paths and the researcher’s degrees of freedom (Gelman & Loken 2014).

In other words, the research methods end up being encapsulated in the theory by way of the numerous decisions that take place during the investigation and that shape the conceptual conclusions communicated at the end of it.

4 Conceptual and methodological theory

In sum, in the methods-as-theory view, theory is not opposed to methods, but rather a theory comprises both concepts and methods.

Moreover, concepts and methods are not meant as categorically distinct entities. On the contrary, they are just two perspectives from which one can look at the constituents of theory, but in practice it is pointless to try and clearly segregate one from the other.

To conclude, as I see it, a theory encompasses both concepts and methods. Concepts and methods are reciprocally entangled, and dismissing the latter for the former would be equivalent to scientific hand-waving.³

References

Bird, Alexander. 2018. Thomas kuhn. In Edward N. Zalta (ed.), The Stanford Encyclopedia of Philosophy. Winter 2018. https://plato.stanford.edu/archives/win2018/entries/thomas-kuhn/; Metaphysics Research Lab, Stanford University.

Gelman, Andrew & Eric Loken. 2014. The statistical crisis in science: Data-dependent analysis. A “garden of forking paths”–explains why many statistically significant comparisons don’t hold up. American scientist 102(6). 460–466.

Kuhn, Thomas Samuel. 1962. The structure of scientific revolutions. The University of Chicago Press.

National Research Council. 2008. The role of theory in advancing 21st-century biology: Catalyzing transformative research. National Academies Press.

Winther, Rasmus Grønfeldt. 2021. The Structure of Scientific Theories. In Edward N. Zalta (ed.), The Stanford encyclopedia of philosophy. Spring 2021. https://plato.stanford.edu/archives/spr2021/entries/structure-scientific-theories/; Metaphysics Research Lab, Stanford University.

Footnotes

1. The tacit role of the “community” in the process of accepting theories is brilliantly explored and systematised in the classic work by Kuhn (1962). See Bird (2018) for an overview of Kuhn’s ideas.

2. I use science here in the most inclusive sense, as opposed to pseudo-science.

3. In future posts I would like to expand on the idea that methods become embedded within theory by focussing on statistical modelling, and on the helpful but often neglected concepts of basic (pure) and applied science and how these can further illuminate the epistemology of theory I sketched here.