Dennis Jay Kenney, Crime, Fear, and the New York City Subways. Crime and Citizen Action. Praeger, New York. 1987.

P.1: "Having identified several law enforcement problems within their city, the Kansas City, Missouri Police Department began, in 1971, searching for workable remedial strategies. As they considered their options, it became evident that any concentration by the department on these specific problems would significantly interfere with their normal preventive patrol operations. Because patrol is considered by many, perhaps most, police officials to be the "backbone" of modern police technology, this was no insignificant problem.

Building upon the work of Albert Reiss and others who had suggested that police patrol was far less productive at preventing crime and deterring criminals than was previously believed, the Kansas City Patrol Division Task Force proposed to put their methods to the test. [There was a year-long experiment.] After analyzing and comparing crime and arrest statistics, augmented by victimization surveys, the researchers concluded that neither doubling nor virtually eliminating preventive patrol had produced any real effect on the number of crimes committed. Additionally, from residential and commercial surveys, they discovered that large variations in patrol activities had no effect on citizens' fear of crime or on the degree of satisfaction or dissatisfaction with the police."

P.2: In 1975 the Rand Corporation, at the request of the National Institute of Law Enforcement and Criminal Justice, did a 2-year nationwide study on the "nature and effectiveness of police investigative operations. "Not only were the often reported arrest and crime clearance rates shown to be unreliable measures of investigative success, they were actually deemed more representative of the crime-fighting abilities of a community's citizens than of its police. Of cases that the police reported as solved, approximately 30 percent of the clearances were the result of officers being summoned and arriving at the scene prior to the offender's departure. In roughly another 50 percent of cleared crimes, the identity of the suspect was known -- usually provided by a victim or a witness -- at the time the initial crime report was taken. In these cases, the primary investigative responsibility consisted of little more than locating the suspect, taking him or her into custody, and assembling the facts for a courtroom presentation. This left only 20 percent of cleared crimes whose solutions could be attributed to investigative work. Even with these cases, however, the study showed that most were solved either by patrol officers, by citizens spontaneously providing further information, or by routine procedures that could have been followed by clerical personnel."

P.3: "After pointing out that the police departments in the 58 largest U.S. cities vary in size from 1.7 to 7.0 officers per 100,000 citizens, at least one observer has noted that there is no observable correlation between the size of a community's police organization and either the number of crimes that are committed or the proportion that are solved. Additional support for this observation has appeared in recent research showing that crime rates do not increase appreciably during police strikes and that increases in police employment cannot be shown to have reduced the levels of crime in a community."

This book, studying the work of the Guardian Angels, shows that there was so little subway crime in the two-month period of the study that they couldn't attribute any success or failure to the Guardian Angels. Whether the Angels were patrolling or not, very little happened on the subway.