Should we really be concerned about the violent crime rate suddenly soaring in the US? Probably not.

If you listen to the Associated Press, it surely sounds as if violent crime has suddenly increased in the US over the last two years.  From the AP:

The government says the rate of violent crime in America went up 15 percent last year and the property crime rate rose 12 percent, signals that the nation may be seeing the last of the substantial declines in crime of the past 20 years. 

Last year marked the second year in a row for increases in the crime victimization survey, a report that is based on household interviews. . . .

The thing to remember here is that the NCVS is a survey.  It is useful in the sense that not all crimes are reported to police, but it is still a survey (even if a very large one).  What the above figure indicates is that the recent increase in serious violent crime (rape, robbery and aggravated assault) as measure by the NCVS is a lot more variable over time.  The increase over the last few years may simply be because the previous drop was "too large."  You might get a survey that shows that there were fewer crimes than they actually were in one year (or even a couple of years).  But over time those errors will likely balance out.    Bottom line is that the decline is probably much closer to the smoother line shown by the FBI UCR data (the red line).  A somewhat similar pattern is also seen with property crimes.

The murder rate data is undoubtedly the most accurate data and it shows a general decline in recent years (though obviously the other crime categories don't need to have moved in lockstep with murder rates).