A hypotheses should be empirically testable. It should be so stated that it is possible to deduce logically certain inferences close to the level of concrete observation so that they can be tested by observation in the field. That is, the hypotheses should have empirical referents.
The concepts embodied in the hypothesis must have clear empirical correspondence and should be explicitly defined. For example, ‘Bad parents beget bad children’ is hardly a statement that can qualify as a usable hypothesis, since ‘bad’ cannot be explicitly defined.