Properties of Theories
What constitutes a good theory?
What constitutes a good theory?
- Inherent quality criteria - before the theory gets in contact with empirical data.
„experimental tests are often superfluous and we should instead focus more on nonempirical evaluations of the quality of our theories” (Szollosi & Donkin, 2021)
- Empirical evidence for the theory (“Empirische Bewährung”), usually expressed as relative evidence compared to a competing theory.
Inherent quality criteria
A good theory …
- defines the variables
- specifies the domain
- builds internally consistent relationships
- makes specific predictions about phenomena that have not yet been observed (only specific predictions, which can also be false, allow falsifications)
- unifies (seemingly) disparate phenomena
Wenn der Hahn kräht auf dem Mist, ändert sich’s Wetter oder bleibt wie es ist.
More properties of good theories
Slide from Karolin Salmen’s talk @ ZPID
Useful (for a specific purpose), not true
His design for a London underground railway was initially rejected as too revolutionary. Today, the technical draughtsman’s idea has gained worldwide acceptance - Beck simplified the route (only horizontal, vertical and diagonal lines) and ignored exact distances.
One of my favorite theories:
\(rB>C\)
A good theory is falsifiable?
Very difficult topic …
- Quine-Duhem problem: A failed test always can either be attributed to (a) a core hypothesis (which would falsify / weaken the theory) or (b) an auxiliary hypothesis (“The manipulation did not work; the measurement was unreliable”)
- Lakatosian defense (blaming the auxiliary, creating new ad-hoc auxiliary hypotheses)
(see Von Nordenflycht, A. (2023). Clean up Your Theory! Invest in Theoretical Clarity and Consistency for Higher-Impact Research. Organization Science, orsc.2022.16122. https://doi.org/10.1287/orsc.2022.16122)