There is no credible doubt that crime has fallen dramatically in the last generation, because of increased use of incarceration, more police, more aggressive and proactive policing, the aging of the most crime-prone component of the population, and probably several other factors. But even given this welcome fact, we have vastly more crime than gets reported.
The crime figures I have been using (and almost all other bloggers and academics use) are taken from the Uniform Crime Reports compiled by the FBI. That source, however, gives only reported crimes. The number of unreported crimes is staggering. As you can see from the chart below derived from the BJS National Crime Victimization Survey (see here), almost all types of crime are massively under-reported. The chart shows that theft -- perhaps the most common crime -- is under-reported by over 70%; rape and sexual assault by about 65%; simple assault by 60%; burglary by just over 40%; and robbery by just under 40%. Indeed, it would seem that the only crimes that routinely get reported are auto theft and (although the chart does not show this) murder.
I have taken this chart from Vox, a liberal source.
Tell me again that now is the time to become complacent about crime.
The reason that NCVS numbers do not show the underreport rate for murder is that NCVS's methodology is to directly interview the victim of the offense, obviously not possible in homicide.
Even so, we can be reasonably sure that the underreport rate for homicide is far less than for most crimes, as found bodies and unexplained disappearances are regularly reported.
Why does motor vehicle theft have the lowest nonreport rate, by far, of the crimes listed? One reason for not reporting is the belief it won't do any good. The victim of motor vehicle theft has strong non-law-enforcement incentive to report -- insurance reimbursement and not being liable for what the thief does with the car.
Valuable post, Bill, especially because I think it very valuable to (try to) consider unreported crime in any broader crime mertics. (As I have suggested before, I also think many crimes committed while in prison (both by guards and prisoners) ought also be part of our data sets about crime rates and recidivism.
On the unreported front, I have long wondered whether any/many of the significant reported crime spikes for various crimes (including even homicides) that started in the 1970s and 1980s involved a lot more crimes moving from the unreported to the reported category. Specifically, there is good social reason to believe/fear before modern civil rights movements (1) that lots and lots of crimes against women and minorities went unreported nationwide (and perhaps especially in certain regions), and perhaps also (2) that lots of theft/economic crimes went unreported because there was limited insurance/economic reason to justify the reporting effort(s).
Of particular importance, especially in light of Bill's suggestion crime is high and should remain a top concern for voters and policy-makers, is whether in 2015 likely to have a higher percentage of crimes going unreportd than in decades past. Not sure this VOX data show any clear patterns, but all of these data highlight further why all crime statistics need to be unpacked carefully for devising an effective policy response.