Crime declines too big to attribute to police falsifying stats

Whenever it's mentioned on this blog that crime rates have declined for the past two decades, a slew of anonymous commenters show up who claim that crime is just as high as ever but police are intentionally underreporting their data, or else victims have been scared away by a phantom "stop snitching campaign." Grits certainly doesn't deny that occurs - in fact instances have been well documented in places like New York City and Dallas. (The HBO drama The Wire famously portrayed this phenomenon as a recurring theme at the Baltimore PD.) But some data - like murders - are difficult to fudge. And surveys of crime victims - which are frequently compared to reported crime data to estimate the extent of underreported crime - consistently show crime drops as substantial as the reported numbers.

So I was interested to see via CrimProf blog this New York Times piece titled, "Crime report manipulation common among New York police, study finds" (June 28), which reported that "An anonymous survey of nearly 2,000 retired officers found that the manipulation of crime reports — downgrading crimes to lesser offenses and discouraging victims from filing complaints to make crime statistics look better — has long been part of the culture of the New York Police Department." One survey respondent put it this way: “Assault becomes harassment, robbery becomes grand larceny, grand larceny becomes petit larceny, burglary becomes criminal trespass.”
 
NYPD responded by attacking the study's credibility: “The latest report from Eterno and Silverman appears designed to bolster the authors’ repeated but unsupported claims. ... The document provides no explanation of how the survey sample was constructed.” NYPD says the survey contains a sampling bias because those surveyed are self-selected. The surveyors didn't attempt to construct a valid sample group, the way pollsters conduct surveys of voters, for example, but instead calculated data from respondents who self-selected and may represent the views of an outspoken, disgruntled subset. Basically NYPD is saying the survey is invalid for the same reason one can't compare internet polls to those conducted by pollsters using statistically valid sampling techniques.

New York City has reported an 80% drop in major crimes, according to the Times,  so if the decrease is really due to fraud by officers downgrading charges, that's a massive conspiracy, and IMO an unlikely one. I don't doubt that virtually every large department succumbs to pressure to fudge crime statistics to some extent, but that would only affect the numbers at the margins. The massive crime reductions recently witnessed IMO can't be explained by manipulating crime stats. The Times story concluded giving voice to a view much closer to Grits' own:

research conducted by Franklin E. Zimring, a criminologist at Berkeley Law School, that compared the department’s crime data for homicide, robbery, auto theft and burglary to insurance claims, health statistics and victim surveys and found a near-exact correlation.

In an interview Wednesday, Mr. Zimring said his research found that the 80 percent decrease in those four crimes reported by the department from 1990 to 2009 was “real.”

He said that there was always “some underreporting, and there is some downgrading in every police force that I know of,” but that his research showed that any manipulation was too minuscule to significantly affect the department’s crime statistics.

There are a number of powerful political constituencies who consider falling crime rates more of a problem than good news: Police unions, elected DAs and career prosecutors, prison systems and their employees - all of them see job security in rising crime rates and risk to their budgets in acknowledging crime has fallen. None of them want to admit such a motive, which is why on this blog such claims are only ever made by anonymous trolls unwilling to attach their names to their opinions and unable to support them with anything but supposition and anecdote. Even if police downgrade thefts on police reports, for example, the number of insurance claims filed wouldn't show a corresponding drop unless there were actually fewer thefts. And while police surely have incentives to downplay crime, respondents to crime victimization surveys (which do use statistically valid sampling techniques) have no reason to falsely downplay crime.

That said, all crime data is imprecise because of both victim over- and under-reporting as well as differences in reporting methodology, accuracy and completeness among departments. So it's easy (and frustratingly common) to misinterpret short-term trends and unwise to draw conclusions without several years of data for comparison. Grits often thinks many folks' expectations of reported crime data are simply too high: They can tell us broad trends but will never capture every jot and tittle of crime that occurs. Taken together with other sources, though, they paint a convincing picture of unprecedented and IMO undeniable long-term crime reductions over the past 20 years.