However, I would define it as the level of rigor used in obtaining data and drawing conclusions from it, including the use of well-designed experiments to test hypotheses.
Physics is a 'hard' science because of the clarity of such experiments, economics is a soft science because such experiments are almost impossible to do, especially in the field of macroeconomics. Earth sciences are hard or soft depending on how 'macro' the activity is -- classification of rocks may be 'hard', but earthquake science is 'soft', because theory-testing experiments are kind of hard to carry out.
That's my view.
Yeah, that's a good start. I've always appreciated studies that had a good experimental design, and it seems, to me, that has been neglected in some sciences recently.
But, to the chagrin, of many "hard" scientists, biology and experimental psychology would qualify as very hard by these criteria.