Program Profile: Police-Monitored CCTV Cameras in Newark, N.J.

This program is designed to deter street-level crime in Newark using closed-circuit cameras, hard-wired to physical structures, which are monitored in real time by CCTV operators. The program is rated Ineffective. Results from one study showed no statistically significant differences in shootings, auto thefts, or thefts from autos. Results from a second study showed a statistically significant decrease in auto thefts but no statically significant difference in shootings or thefts from auto.

An Ineffective rating implies that implementing the program is unlikely to result in the intended outcome(s).

An Ineffective rating implies that implementing the program is unlikely to result in the intended outcome(s).

Program Goals

The goal of Police-Monitored Closed-Circuit Television (CCTV) Cameras in Newark, New Jersey, is to reduce street-level personal and property crime through the installation and monitoring of public surveillance cameras in high-crime areas. Monitored footage in real time is meant to reduce police response time and increase the likelihood of apprehending someone who commits a crime.

Target Site

Newark is the largest city in New Jersey, with a population of approximately 280,000 in 26 square miles. The percentage of residents living below the poverty line in Newark (28 percent) is nearly three times that of New Jersey as a whole (9 percent).  Ethnic minorities make up most of the population of Newark: 52.4 percent of residents identify themselves as Black, and 33.8 percent identify themselves as Hispanic or Latino.

Program Components

CCTV sites are largely permanent fixtures, with cameras hard-wired to physical structures. Camera footage is transmitted to a communications center where it is monitored and recorded by CCTV operators under supervision of a police sergeant. Crimes viewed by the operators are documented, and police are dispatched to the incident locations. Network connectivity is a prerequisite for the camera installation sites.

The installation in Newark occurred in two waves. It started in the city’s urban enterprise zone and continued to other neighborhoods as the capacity to expand the network increased. Several types of cameras were installed, including dome, bullet-resistant, and rooftop cameras. Dome and bullet-resistant cameras are usually mounted on telephone or light poles at street intersections and are in plain view. These cameras can zoom, pan 360 degrees, and tilt 180 degrees. Dome cameras are designed to be less visible from the ground. Bullet-resistant cameras, encased in glass and steel housing to protect them from projectiles, are easily visible to those on the street. Lastly, rooftop cameras are mounted on building roofs and are mostly inconspicuous to those on the ground.

Program Theory

Surveillance methods, such as the use of police-monitored CCTV cameras, are rooted in rational choice theory, which posits that people decide to commit an offense on a case-by-case basis, weighing the specific benefits with the likelihood of being caught. For this method to deter criminal activity effectively, a person must believe police will respond and that there is a high likelihood that he or she will be caught (Piza et al. 2014b).

Study

Caplan and colleagues (2011) used a quasi-experimental design to evaluate the deterrent effect of police-monitored, closed-circuit television (CCTV) cameras on crime in Newark, New Jersey.

The study examined the effects of cameras installed before and after two installation time points (March and July 2008) on measures of crime. Two different placement strategies were applied. The placement of cameras installed in March was dictated by sponsors who paid for the cameras and required that they be placed in Newark’s Business District. The cameras installed in July were placed in known high-crime areas in consultation with Newark Police Department (NPD) personnel. Only dome cameras were examined in this study (n=73); they have the greatest range of motion and their aim is not easily visible to those on the ground.

For the unit of measurement, the authors calibrated the public area visible from the location of the camera, called a viewshed, to view a 582-foot radius (or buffer zone) around the camera. They used Google Earth and ArcGIS to digitize the viewshed area, in addition to tools in the NPD’s CCTV control room, to validate the aerial data with what was occurring in real time.

In addition to the 73 experimental viewsheds, 73 comparison viewsheds were created using two criteria: 1) the area had to be patrolled by the NPD, and 2) NPD administrative data had to be available for the area. Comparison viewsheds were randomly placed in hypothetical camera locations by using the same technical system as the CCTV treatment viewsheds.

NPD administrative data were examined for experimental and comparison viewsheds for the 13 months before (pretest) and 13 months after (posttest) the camera installations. Crime was measured by number of shootings, automobile thefts, and thefts from automobiles that occurred within the areas. T-tests were used to compare the effects of CCTV in the experimental viewsheds relative to the comparison viewsheds.

Two subgroup analyses were conducted. The first compared the deterrent effect of strategically placed cameras versus  randomly placed cameras. The second compared the deterrent effect on crime of cameras in low-crime areas, versus cameras in high-crime areas.

Study

Piza (2016) conducted a quasi-experimental study using propensity score matching (PSM) to evaluate the effectiveness of police-monitored CCTV cameras in Newark, New Jersey, on deterring crime. Viewsheds were developed, using the same methods as detailed in Study 1 (above), for all CCTV cameras. However, the size of the viewsheds was reduced from 582-foot to 423-foot buffer zones.

Cameras were installed in four phases from March 2008 to April 2010. Comparison viewsheds, placed in hypothetical camera locations, were selected via PSM, using the following variables: 1) the number of crime incidents during the 1-year pre-installation period; 2) number of arrest incidents during the 1-year installation period; 3) sociodemographic variables, including concentration of social disadvantage, racial heterogeneity, residential population, and ambient population (those who frequent the area but do not live there) in the surrounding 2010 U.S. Census block; and, 4) whether the viewshed fell into one of four Newark police precincts.

After the propensity scores were calculated, a matching process was conducted that accounted for the phase in which a camera was installed. The author used a final sample size of 228 (114 treated, 114 untreated) viewsheds for automobile theft, 224 (112 treated, 112 untreated) viewsheds for theft from automobiles, and 226 (113 treated, 113 untreated) viewsheds for the violent crime analysis.

Pretest and posttest data included NPD administrative data for 1 year before (pretest) and after (posttest) installation for automobile theft, theft from automobiles, as well as a composite violent crime variable that included murder, non-fatal shootings, and robbery.

Odds ratios and average treatment effects were calculated to measure the overall change in automobile theft, theft from automobiles, and violent crime in treatment versus comparison viewsheds. Micro-level effects were also calculated to determine whether individual treatment viewsheds experienced a statistically significant treatment effect relative to their matched control units. Subgroup analyses were not conducted.

Caplan and colleagues (2011) used an ANOVA to conduct a subgroup analysis to examine whether the two sets of strategically placed cameras (from March and July installations) in Newark had a differential deterrent effect, compared with randomly placed cameras. Results suggested that there was a statistically significant deterrent effect on shootings for groups of strategically placed cameras, compared with the group of randomly placed cameras. There was no statistically significant difference between groups on auto theft or theft from auto. There was a statistically significant effect on shootings, however the authors caution that the effect is trivial and not substantively meaningful.

A second subgroup analysis was conducted based on degree (low or high) of criminal activity in the area. The study authors calculated the “location quotient” (LQ), which is a measure of the relative significance of crime incidents within experimental view sheds relative to their significance in the larger City of Newark. Therefore, an area with an LQ<1 is considered a low- crime area, and an area with an LQ>1 is considered a high-crime area. Caplan and colleagues (2011), using an independent samples t test, found that areas with LQs<1 were associated with a statistically significant decrease in the number of shootings and auto thefts, but not in the number of thefts from auto. Areas with LQs>1 yielded no statistically significant results for shootings, auto thefts, or thefts from auto.

These sources were used in the development of the program profile:

These sources were used in the development of the program profile:

Piza, Eric L., Joel M. Caplan, and Leslie W. Kennedy. 2014a. “Analyzing the Influence of Micro-Level Factors on CCTV Camera Effect.” Journal of Quantitative Criminology 30(2):237–64.

Piza, Eric L., Joel M. Caplan, and Leslie W. Kennedy. 2014b. “Is the Punishment More Certain? An Analysis of CCTV Detections and Enforcement.” Justice Quarterly 31(6):1015–43. (This study was reviewed but did not meet Crime Solutions' criteria for inclusion in the overall program rating.)