Answer:
There are many definitions out there, but there is some common ground to them. They all tend to agree that a theory needs to be (a) substantiated, (b) explanatory, (c) predictive, and (d) testable. That is: (a) substantiated - a theory cannot be independent of prior work and evidence, there needs to be some justification of it, within previous work in the field (other currently accepted theories) and in the sum of available evidence. For (b) is needs to actually explain something about the science it is in. The explanation covers causality. So, for example, the laws of thermodynamics are laws rather than theories because they describe rather than explain what happens. (c) and (d) are linked. A theory needs to make predictions that can be tested, so that the theory itself can, in principle, be rejected. And for the theory to be sound there needs to be a genuine commitment to reject the theory of the tests fail to support it. It isn't really a theory if, for example, there is either the intent or the logical possibility of interpreting evidence both ways. There is often a fifth criterion, which is essentially a coherence or elegance - does the theory "feel" right. This can include Occam's razor, ruling out excessively or unnecessarily complicated theories, which can easily be devised. Whether all of these are truly "essential" is probably a matter of debate, defining features of any category can be a problem, but together they describe the common characteristics of typical theories, and that's a good place to start.