Each January, Charlotteans receive news about the previous years crime statistics from the Charlotte-Mecklenburg Police Department. The news coverage typically emphasizes recent trends in the citys crime rates. Consider the Charlotte Observers article last year describing the change in crime in Charlotte from the previous year: the number of crimes dropped 7.1 percent last year. This years news coverage reported that the citys overall crime rose by 2.8 percent in 2012, fueled in part by an increase in robberies and aggravated assaults. Are we to believe that crime actually fell by 7 percent one year and increased by 3 percent the very next year?
What actually dropped by 7.1 percent in 2011 and increased by 2.8 percent in 2012 were crimes known and recorded by CMPD, not necessarily crimes that actually occurred. This distinction suggests an important point: actual or real crime is comprised of both crimes known to the police and crimes not known to the police not known because they were never reported.
If police-based crime statistics show a drop of 7 percent or an increase of 3 percent in one years time, those changes might be due to shifts in real crime levels. On the other hand, suppose real crime levels stayed the same and the likelihood that victims report their victimizations to the police was what actually changed. Maybe both changed. In fact, after 40 years of rigorous nationally representative survey research based on the National Crime Victimization Survey, we know that the majority of crime victimizations reported in that survey are not reported to the police. We also know that crime reporting behavior changes over time.
Consider the following example. The NCVS data indicate that an estimated 58.8 percent of households victimized by residential burglary in the United States reported the victimization to the police in 2010 (plus or minus 3.7 percent margin of error). Then, in 2011, this same reporting rate dropped to 52 percent (3.5 percent margin of error). If one combines the CMPD residential burglary statistics over this time period (7,305 in 2010, and 6,352 in 2011), with the NCVS reporting ranges, we conclude that the real number of residential burglaries in Charlotte could plausibly lie anywhere in the range of 11,688-13,258 for 2010 and 11,445-13,097 for 2011. Since these two ranges overlap, we cannot with any confidence say whether residential burglaries increased, decreased, or stayed about the same in Charlotte from 2010 to 2011.
When our students learn about these kinds of ambiguities, they sometimes ask why the police-based crime statistics receive so much attention. Our answer is that they are the only numbers readily available to measure crime incidence at the local level. At the same time, it doesnt follow that the police numbers are well suited for measuring the real incidence of crime at the local level or any other level.
Numbers still useful
So, what information value should we attach to police-based crime statistics? We think these numbers serve two useful purposes. First, they tell us about the workload of the police. The police respond to and count the crimes that come to their attention. Thus, bringing a crime to the attention of the police creates workload that triggers time and money costs for the police.
Second, they tell us about long-term trends in crime. When we look at police-based crime statistics over the long run, we often see changes that are larger than what could be explained by the margin of error in the NCVS reporting rates. Such changes can provide strong evidence that crime is truly changing over long periods of time even though the changes in any single one- or two-year period are less clear.
Here is the basic problem. When we use police-based crime statistics to say that crime increased by 3 percent from one year to the next, we are automatically making the assumption that all of this change is due to changes in crime and none is caused by changes in crime reporting.
It is too easy for people to fall into the trap of thinking that crime known to the police is the same as crime. Our view is that greater attention to this distinction will lead to a better understanding of what we know and what we dont know about crime trends in U.S. cities.