I would use theta/2theta since in that case you have definite conditions and you are talking about powder diffraction. In order to improve the accuracy you can even go to higher orders. However, it is called 002 and not (002) since the last only described Miller indices which have to be prime, and 002 are Laue indices describing the reciprocal lattice point. It is the second interference order of lattice plane (001).
On the other hand you should not get a different result for the rocking curve as well if you are rocking 2theta of course. It only means that you do not keep the "reflecting" plane parallel to the surface plane which is not that dramatic for a powder. A different case is a single crystal. There the conditions are not that simple since you have to adjust your crystal in a very accurate way. This is also the reason why lattice parameters determined on single crystals are commonly not that accurate as for a powder. In powder diffraction the beam and the detector directions automatically select the best matching crystals.