The Bureau of Justice Statistics has released a revision of its 2016 National Crime Victimization Survey results. Here are links to the press release and full report.
The NCVS is different from the FBI's Uniform Crime Reports, better in some ways and not as good in others. As a survey, it does not depend on crimes being reported to police, but it is subject to sampling error. Because respondents are only asked about crime committed against themselves, it does not measure homicides or crimes against children. The indexes are also put together somewhat differently.
The revision shows a smaller bump in the violent crime rate between 2015 and 2016, falling beneath the threshold of "statistical significance." (There's that troublesome sampling error.) The UCR shows a 3.4% increase for the same interval.
The revised NCVS does show a significant increase in violent crimes against males and persons aged 25-34. It's not immediately obvious why there would be a bump in that particular bracket.
"The revised NCVS does show a significant increase in violent crimes against males and persons aged 25-34. It's not immediately obvious why there would be a bump in that particular bracket.”
One of the ways to avoid becoming a crime victim is to avoid places where crimes occur, especially at those times of night when they are most likely to occur.
Late night restaurants and bars for example. Males age 25-34 are more likely to be in those places at those hours. The gentrified neighborhood in Denver known as “LoDo” is peaceful and safe until around midnight. Then it becomes a place to avoid. Those over age 34 mostly do.
Young men have long been the most likely demographic to become victims of violence.
What is not obvious is why there was a larger increase in violent crimes in this particular demographic segment from 2015 to 2016 than for other segments.
It could be a glitch. The "crosstabs" in surveys necessarily have considerably smaller sample sizes than the overall numbers, and hence larger margins of error.