In economics they call type [1] positive and type [2] normative - the implication being that economics only deals with positive questions and tells us nothing a

In economics they call type [1] positive and type [2] normative - the implication being that economics only deals with positive questions and tells us nothing about normative questions (how should the world be). When I studied a bit of sociology they took an approach borrowed from philosophy in categorising methods (and effectively split up type [1]): they were interested in how the theory (substitute argument here) addressed the questions of [a] what is the nature of the problem (ie ontology or how the world is), and [b] what can we know about it (ie epistemology). How you answer questions of type [b] determines whether type [1] statements can "be tested by experiment" - ie not everything that is in the world can "in principle be tested by experiment". For example, often whilst we can say that A is generally followed by B, and so infer that A at least partially causes B (whilst causation does not imply correlation we usually infer it does if we have some plausible mechanism in mind for why it should), but that in itself can not reveal precisely to us exactly why A causes B - ie how the mechanism works exactly. Most writers tend not to explicitly reveal their views on [b] and so we are left to infer them. As to identifying type [2] statements, the verbs "want", "should", "prefer" are dead give-aways but the problem lies when they are dropped from the sentence - eg cabalamat's example becomes "Everyone likes cats". Using [b], we can ask whether it is possible to know this, which might help identify it as a type [2].