Longitudinal comparability of the NCVS data is also a moot point, because self-reporting is subject to changing perception of what is a crime. That pertains especially to rape or marital rape but also to other crimes as well. Vigilante hype may cause Rs reporting as crimes incidents they did not previously considered as such (e.g. an argument with an aggressive panhandler may be interpreted as an "assault"). However, I understand that major surveys try to minimise the subjective factors by probing Rs for specific events.
In any case, this possible discrepancy strengthens your point re. decilining crime - the rates are declining even with the possibility of increased overreporting due to the effects of war on crime hype.
wojtek