January 27, 2014
A useful reminder of the challenge of assessing crime rates, lies, damn lies and statistics
With apologies to Mark Twain for tweaking his famed comment about lies, damned lies, and statistics, the title of this post stems from this interesting recent story about the challenges of crime rate measurement and statistics. The story is from the Denver Post and it is headlined "Denver's top law enforcement officers disagree: Is crime up or down?". Here is how the piece starts:
Denver's top two law enforcement officials disagree on the answer to what ought to be a simple question: Is violent crime up or down?
Police Chief Robert White and District Attorney Mitch Morrissey aren't quibbling over minor details; they have a nearly 18 percentage-point difference in opinion about the way crime is trending. Experts say their disagreement underscores the complexities of measuring and interpreting crime trends in a major city.
White has repeatedly said violent crime fell 8.6 percent last year. Morrissey wonders how that can be true when felony cases submitted to his office rose 9 percent during the same time. "One of the things you can glean from it is that the crime rate is going up. It has to be. (The police) are presenting more cases to us," Morrissey told The Denver Post. "The trend is that our caseloads are getting bigger and bigger. How is that possible with the crime rate going down? I don't know."
White stood by the 8.6 percent decline he has boasted about at public gatherings and in police stations, saying it was the result of better police work, even with fewer officers on the street. "How about the fact that maybe the police department is doing a better job of arresting the right people?" White said. "His cases are going up because the police are out there working their butt off and doing a better job. That's not rocket science."
But there is some science to crime statistics, said Callie Rennison, an associate professor in the University of Colorado Denver's School of Public Affairs. "It's a hard, hard thing to measure, which of course makes it hard to say, 'Well is it really going up or really going down?' " she said. "Anyone who tells you, 'Here's my stat, it's a perfect one,' immediately don't trust them. No stat is perfect, but some are less perfect than others."
January 27, 2014 at 07:05 PM | Permalink
TrackBack URL for this entry:
Listed below are links to weblogs that reference A useful reminder of the challenge of assessing crime rates, lies, damn lies and statistics:
"The trend is that our caseloads are getting bigger and bigger. How is that possible with the crime rate going down?" I can think of several possible explanations:
Hypothesis #1: Maybe the DA's caseload increase doesn't come from violent crimes but from drug crimes or other victimless offenses. One may see reported crime drop but with the same number of police they still must do SOMETHING with their time and so make more drug arrests. That's what's been happening in Houston, for example. Wouldn't surprise me if it's also true for Denver.
Hypothesis #2: Actual crime goes down but policing resources stay the same. So police have more resources to investigate the crime does occur and therefore solve more cases, meaning DA caseloads increase.
Hypothesis 3: The PD is messing with their stats. In Dallas their property crime rate declined significantly in recent years but it turned out they changed their definitions to lump multiple reports into one and created new excuses not to include certain offenses in their reports. Similarly, the National Crime Victim Survey reported increases after they changed their methodology to be more inclusive, but reported UCR crime continued to drop. How one defines crime affects how it's measured.
There are any number of reasons the cops and the DA's numbers might be different.
Posted by: Gritsforbreakfast | Jan 28, 2014 10:18:02 AM
The problem, obviously, is that there is no direct way to measure crime rates. So they have to use proxies. It is easy to count how many people are arrested, how many crimes are reported, how many cases are referred to the DA, or how many offenses are charged. But those numbers are dependent on things other than the actual number of crimes committed. Therefore, changes in the data can reflect trends that have nothing to do with incidence of crime.
Posted by: Chris Jenkins | Jan 28, 2014 2:02:55 PM