I have written a few articles on the misrepresentation of scientific theory. I have pointed out that theories are not “fact.” They are not known to be true. In the past, I have used the word “consistent” rather than “true” in order to address a theory, but consistency is not really enough to describe how “good” a theory is. For this, “robustness” is needed. While this short discussion does not generate an actual metric for robustness, I use the term enough that I should at least explain what I mean, in general, when I say that a theory is or is not robust.
The core issue with calling a theory “true” is that we really have no idea if a theory is true or not. We do not know how many different models exist which are consistent with a given body of theory and evidence, and we therefore cannot use induction unless we make a large number of limiting assumptions, which in most cases would be unjustified. However, it is still useful to compare different theories in order to make a selection. First and foremost, a theory must be falsifiable. If it is not falsifiable, then it is not useful to us. A robust theory is one which is consistent with existing evidence and other theories, is able to explain a large number of observations, has survived extensive attempts to falsify, and has a large number of other theories which would become false if the theory in question were false.
Evolution is a prime example. The theory of evolution, or really the collection of theories on evolution, are consistent with various other theories in biology and paleontology. It is consistent with the current body of evidence available to us. Part of epidemiology is dependent on our theories of evolution, as is a lot of psychological theory. If evolution were untrue, then it would throw much of our body of scientific theory on its head. It has also been tested numerous times over the last few hundred years. For this reason, evolution is robust.
One final point is that due to the idea that falsifiability is at the core of our theory selection process, it is not necessarily the “simplest” explanation that is the one we select, but rather the one which is most easily falsified. It does not make sense to select a theory which would require a large amount of time and money to produce experiments capable of falsifying it when there are theories which require fewer resources. This however is a little different from “robustness.”