Share to Google Plus

Today we are featuring a post from guest blogger Hal Pepinsky, who is known as the co-founder of ‘Peacemaking Criminology.’ Hal retired in 2009, but blogs regularly about crime and criminal justice at where this piece originally appeared.

My recent critique of NYPD-based, CompStat enabled, “broken windows” police (“’Broken Windows’ Unjustified,” May 6) rests on two decades of trying to understand what “crime” and “criminality” statistics mean, from a law-school class in 1967, through the study “Explaining Police-Recorded Crime Trends in Sheffield (UK)” (Contemporary Crises, 11, pp. 59-73, Jan. 1987; pdf’s of this and other articles cited here available on request to those who can’t get them online).

The class on “crime and society” was taught by the Executive Director of the President’s Commission on Law Enforcement and Administration of Justice, James Vorenberg, and the Director of its Task Force on Assessment of Crime, just before the Commission issued its reports in 1967. There we were introduced to national victim surveys, the first of their kind. Our primary question on the final exam was to imagine oneself to be a congressional intern advising on the significance of rising police-reported crime rates. I wrote that the connection between what police reported and what actually happened in communities they policed was unknown. I got one of my two worst law-school grades in that course, which inspired my choice of sociology dissertation topic several years later.

For my dissertation, in a “high-crime” area of Minneapolis, I gathered over 80 items of information, starting with the date, time, and code given in the call for service, describing and detailing any interactions with complainants/witnesses, ending with what offense, if any was reported. In 373 calls for service during 500 hours of patrol, on all shifts, over a period of about one year, 97 offenses of all kinds were reported. Ninety-three of those reports occurred after the dispatcher had named an offense in the call, where the police founded a complainant’s naming of any offense, while in at least 22 calls where the dispatcher had not named an offense but complainants had described one, no offense reports were filed.

In sum, when dispatchers sent police to check on offenses and they found corroboration of any offense, they created crime statistics. Without that signal from dispatchers, police never reported offenses, regardless of what complainants reported. The police, meanwhile, believed that their decisions were simply based on the evidence at hand, and were surprised by the results. The study was published as “Police Patrolmen’s Offense-Reporting Behavior,” Journal of Research in Crime and Delinquency, 13 (Jan. 1976, pp. 33-47). And so I learned that police discretion whether to report create crime statistics can rest heavily on factors independent of “the true” incidence or rate of crime.

I am grateful to my dissertation adviser, Marvin Wolfgang, for the access he gave me to his copious files of reprints of articles on the history of crime measurement from Europe to the United States. I supplemented my Minneapolis findings with a critique of “The Growth of Crime [Measurement] in the United States” (Annals of the American Academy of Political and Social Sciences, Jan. 1976, pp. 22-30). I traced how the size and political weight of “the crime problem” had grown, from judicial data beginning early in the 19th century, to newly touted victim and self-report survey data. This comparison of measures of crime evolved into a book, Crime Control Strategies: An Introduction to the Study of Crime (New York: Oxford U. Press, 1980), where a chapter at a time I “evaluated chances of controlling rates” of convictions, arrests, police offense reports,” victimization reports, self-reports, and recidivism, plus a chapter on cost-benefit analyses. In each category, compared choices of rate numerators and denominators, type I and II error factors in applying statistics using the measures, political considerations in implementation, and side effects of policies based on those data, where I traced trends in victimization and differences in self-reporting to changes in interviewing and in respondents’ relationships with data collectors. Fundamentally, crime and criminality trends reflected trends and differences in the behavior of the data gatherers and in how the data providers perceived themselves in the eyes of their interrogators.

This was followed by two longitudinal studies of police crime reporting. The first was an analysis of police offense reports and arrests, with William Selke: “The Politics of Police Reporting in Indianapolis, 1948-1978,” Law and Human Behavior, 6 (Dec. 1982: 329-342). Police data were supplemented by news reports from the Indianapolis Star and News. When two-way radios in patrol cars were introduced in the latter 1950s, “index” crime reports shot up. Criticized for failure to curb mounting crime, the Indianapolis police turned their focus to making arrests, which rose as reported offenses declined. Then public complaints that police weren’t taking offense reports and weren’t clearing offenses by arrest. In the uproar, offense reports swung upward while arrests declined, until public alarm over police losing control of crime swung the trends back. I labeled these recurring cycles “the roller coaster effect” of law enforcement reporting. This brought home to me the inverse relationship between trends in index offense rates, and arrests, most of which are for public order offenses rather than “serious crime.”


In the Sheffield study cited above, I had great assistance of the police in gathering extraordinarily detailed monthly printouts of crime recording records, and in interpreting patterns there from 1974-1979. In this and in trends from earlier years, a variety of demographic changes failed to correspond to social and economic shifts. However, with a surge of more highly educated recruits, during their first two years of probationary status, reports of “notifiable” offences climbed dramatically. It was followed by a wave of criticism that the police “clear-up” (crime solving) rate had fallen below the normally acceptable fifty percent. Patrol constables were reportedly instructed to avoid taking nuisance reports for offenses that could not be cleared. AS IN INDIANAPOLIS, records in Sheffield show an ensuing rise in arrests on one hand, in offense reports being “no crimed” (wiped off the books) more, and in previously reported offenses being “cleared” as “taken into consideration” (TIC-ed) during suspect interrogations, while notifiable offenses dropped, and clear-up rates rose back to acceptable levels.

I concluded the Sheffield study by calling for a moratorium on using counts of crime and criminality to evaluate and implement criminal justice policy. I had also envisioned how police and community members together could formulate and apply performance evaluation criteria that included a variety of ways police could “protect and serve” communities (in “Better Living Through Police Discretion,” Law and Contemporary Problems, 47, no. 4, 1984, pp. 249-267), for police and civilians to build trust, respect and growing sense of safety and security together. Not only does “broken windows” policing endanger police and civilian lives alike; statistical evidence that it works is illusory.

It took me a long time to give up on finding any connection between reported crimes and the lives of people in the communities from which reports are taken, and has led me instead to focus on ways human relations deteriorate or grow safer and more dependable in parallel fashion at all social levels, in all social settings (a paradigm laid out in Peacemaking: Reflections of a Radical Criminologist, University of Ottawa Press, 2006; final page proofs for free download at

I am aware that how centrally grounded criminological knowledge, and criminal justice policy and practice, are grounded in the assumption that lower crime rates and more crime control activity are evidence that the criminal justice system works. I want colleagues especially to know that I have not arrived at my rejection of the validity of this research construct lightly.

See related posts:
May 6 – “Broken Windows Unjustified”
January 6 – “Good News from NYPD”

Hal Pepinsky
Professor Emeritus
Department of Criminal Justice
Indiana University

Be the first to comment

Leave a Reply

Your email address will not be published.