Private Security and Closed-Circuit Television (CCTV) Surveillance: A Systematic Review of Function and Performance

Abstract

Private security personnel play an important but largely overlooked role in the operation of closed-circuit television (CCTV) surveillance to prevent crime in public and private areas. This role can take a number of forms, including active monitoring of cameras. Drawing upon a global database of CCTV evaluations (N = 165), this article examines the function and performance of private security personnel as related to the effectiveness of CCTV. Findings indicate that CCTV schemes operated by private security personnel generated larger crime prevention effects than those operated by police or those using a mix of police and security personnel. Policy and research implications are discussed.

Keywords

private security CCTV surveillance crime prevention systematic review

Introduction

Closed-circuit television (CCTV) surveillance cameras occupy a central role in contemporary policing and crime prevention (Goold, 2004; Weisburd & Majmundar, 2018; Welsh & Farrington, 2009). As the practical application of CCTV has increased in recent years, so has the evidence base on its crime prevention effect in public and private space. The cumulative evidence demonstrates that CCTV surveillance is associated with a significant yet modest reduction in crime. However, effects vary across a range of contextual factors, including geographical setting (e.g., city and town centers, car parks), crime type, camera monitoring strategy, use of complementary interventions, and country of origin (Piza et al., 2019).

Private security personnel play an important but largely overlooked role in the operation of CCTV surveillance to prevent crime in public and private areas. This role can take a number of forms, including on-site active monitoring of cameras and on the ground responses to crimes in progress captured on cameras (e.g., Gill & Spriggs, 2005; Waszkiewicz, 2013).

In general, research on private security personnel in the context of CCTV surveillance has focused on the operations of those who monitor the cameras, sometimes known as the “watchers,” as well as on security guards working alongside or as a complementary intervention to CCTV. Research on both fronts is limited, and neither has attempted to investigate the relative effectiveness of CCTV systems monitored by private security personnel or police. In the case of the latter research focus, this is distinguished from security personnel (i.e., security guards) serving as the primary intervention to prevent crime (see Welsh et al., 2010), and is really a matter of security guards exercising a formal surveillance function (Cornish & Clarke, 2003) and serving as a secondary or additional intervention to CCTV cameras. Unfortunately, too few examples of this preclude an analysis of security guards as a moderating variable of the effects of CCTV on crime (Piza et al., 2019). With respect to the other area, some qualitative research has examined the day-to-day operations of those who monitor the cameras, with some distinction among the different parties involved: private security, police, local government staff, or volunteers (Gill & Spriggs, 2005; Wilson, 2005).

The main aim of this article is to examine the function and performance of private security personnel as related to the effectiveness of CCTV surveillance. The chief question of interest is as follows: How effective is CCTV surveillance in preventing crime when it is operated by security personnel compared with other parties? Using systematic review methods and incorporating meta-analytic techniques, the article draws upon a recently updated database of CCTV evaluations (N = 165), covering 40 years of research (1978–2018) and drawn from the United States, United Kingdom, Sweden, Canada, South Korea, and other industrialized countries.

Method

The primary list of studies was compiled by Piza et al. (2019) as part of their updated systematic review and meta-analysis of the effects of CCTV on crime. Studies were identified and located following a comprehensive set of search strategies, and studies were included in the systematic review if they met the following criteria: (a) CCTV was the main focus of the intervention; (b) the evaluation used an outcome measure of crime; (c) the research design involved, at minimum, before and after measures of crime in treatment and comparable control areas; and (d) both the treatment and control areas experienced at least 20 crimes during the pre-intervention period. Building upon the prior systematic review conducted by Welsh and Farrington (2009), Piza et al. (2019) amassed a database of 161 CCTV studies (80 included and 81 excluded). In line with the scope of the present review, four additional studies were eligible for consideration, bringing the total number of CCTV studies to 165 (84 included and 81 excluded; see Appendix for all included studies). Of the 84 included studies, 76 could be used in the meta-analysis. Four did not provide the requisite data for an effect size to be calculated. The other four did not provide enough detailed information about the nature of the CCTV operation to allow for coding of the scheme operation variable (i.e., police, mixed-police, or security).

“Scheme operation” is the primary variable of interest in the present review. We reviewed each study to determine the personnel primarily in charge of carrying out surveillance functions and notifying the appropriate parties when an offense was observed on camera. CCTV schemes that exclusively incorporated sworn police officers in the surveillance function were coded as “police.” Thirty-seven studies fit this criterion. Twelve studies reported on schemes incorporating both police officers and civilian security personnel in the surveillance operation. These evaluations were coded as “mixed-police.” Twenty-seven studies reported that civilian security personnel were solely involved in surveillance functions, and were coded as “security.” Given the scope of this review, we pay particular attention to the effect of security schemes as compared with that of the police and mixed-police schemes. Authors of the primary studies were contacted via email when a determination could not be made from the study text.

It should be noted that we were unable to distinguish the nature of civilian security personnel used in the CCTV schemes beyond our typology: security, police, and mixed-police. This was owing to a general lack of detail reported in the studies. The majority of studies reported the use of police and/or civilian operators without discussing a number of related processes, such as the nature of operator training, the policies guiding monitoring practices, and whether civilian security personnel were in-house or contracted. From the available information there seems to be a high level of variance across civilian security operators. For example, the civilian security operators working alongside police officers in Newark were hired, trained, and supervised directly by the Newark Police Department. The Burnley CCTV scheme used retired British Legion personnel hired directly by the local authority who monitored cameras from a separate facility, with the police having no authority over the CCTV operators (see Appendix).

Meta-analytic techniques were used to compare the effect of CCTV across the three scheme operations (police, mixed-police, and security). The odds ratio (OR) is used as the measure of effect size. The OR indicates the proportional change in crime in the control area compared with the treatment area. An OR greater than 1.0 indicates a desirable effect of the intervention, and an OR less than 1.0 indicates an undesirable effect. The inverse of the OR communicates the crime difference within the treatment area, with a value of 1.25, for example, indicating that crime decreased by 20% (1/1.25 = 0.8) in the treatment area compared with the control area.

Analyses were conducted using BioStat’s Comprehensive Meta-Analysis software (Version 3.0). We conducted all analyses as random effects models under the assumption that effect sizes are heterogeneous across individual evaluations, as well as subpopulations of evaluations (Lipsey & Wilson, 2001). In each case, observed Q statistics and associated p values supported this assumption, demonstrating significantly heterogeneous effect sizes across categories of our “scheme operation” variable. Also, to account for the potential influence of outcome measures on observed effect sizes, we followed the analytic approach of recent systematic reviews of using three approaches to report meta-analytic results (e.g., Braga et al., 2018). In the first approach, all reported outcomes are summed to present an overall average effect size statistic. This is a conservative measure of the effect of each type of CCTV scheme operation. In the second approach, the largest reported effect size for each study is used, which presents a “best-case” estimate. In the third approach, the smallest reported effect size for each study is used, representing the lower bound estimate of effect.

Finally, we conducted meta-regression models to further explore how effects differ across the scheme operation categories. When combined with traditional meta-analysis methods, meta-regression provides the benefit of controlling for moderator variables that researchers believe may partially explain observed effect sizes. The scheme operation covariate is the independent variable of primary interest in the present review. “Police” was set as the reference category, generating covariates that measure whether the mixed-police and security schemes outperform police-led schemes. Five additional variables were included as control variables, because Piza et al. (2019) found each of them to be significantly related to CCTV effects on crime. Two binary variables identify whether the CCTV scheme was deployed in a car park or a residential setting. A binary variable identifies whether the study was conducted in the United Kingdom. The two final covariates are binary measures identifying whether the scheme incorporated active monitoring of surveillance cameras and whether multiple complementary interventions were deployed alongside CCTV. Similar to the meta-analyses, all meta-regressions were conducted as random effects models to account for the heterogeneity not explained by the covariates (Thompson & Higgins, 2002).

Results

All three scheme operations exhibited statistically significant crime reducing effects. However, the operations differed in terms of the proportion of evaluations reporting desirable effects, as well as the magnitude of the pooled effects. For the 37 police-led schemes, eight reported desirable effects, one reported undesirable effects, and 28 reported nonsignificant effects (see Figure 1). For the 12 mixed-police schemes, three reported desirable effects, whereas the remaining nine reported nonsignificant effects (see Figure 2). Eleven of the 27 security schemes reported desirable effects, a higher proportion (40.7%) than what was observed for both the police (21.6%) and mixed-police (25.0%) schemes. Two of the security schemes reported undesirable effects and 14 reported null effects (see Figure 3).

Figure 1.

Forest plot of police schemes (average effects).

Figure 2.

Forest plot of mixed-police schemes (average effects).

Figure 3.

Forest plot of security schemes (average effects).

Table 1 displays the findings of the meta-analysis comparing the pooled effect sizes across scheme operations. In the average effects meta-analysis security schemes exhibited the largest effect (OR = 1.225), indicating a crime reduction of approximately 18% in treatment compared with control areas. Pooled effect sizes for mixed-police (OR = 1.164) and police (OR = 1.081) indicated crime reductions of approximately 14% and 7%, respectively. ORs achieved statistical significance (p < .05) for each of the scheme operations. Security schemes once again demonstrated the strongest effects in the largest effects meta-analysis (OR = 1.208), indicating an approximately 17% reduction of crime in treatment compared with control areas. However, the effect of police schemes was nearly identical (OR = 1.206) when the outcomes exhibiting the largest effects are considered. The mixed-police schemes exhibited the smallest effect (OR = 1.186), which was not substantially smaller than either the security or police schemes, with a reduction of about 16% in treatment compared with control areas. As with the average effects models, all ORs achieved statistical significance (p < .05) for each of the scheme operations in the largest effects meta-analysis. The smallest-effects meta-analysis again indicated security schemes as having the largest effect size (OR = 1.136), followed by mixed-police (OR = 1.100) and police (OR = 1.026).

Table 1.

Effect on Crime by Scheme Operation.

Category	N	Odds ratio	95% confidence interval		p
Category	N	Odds ratio	Lower	Upper	p
Average effects
Police	37	1.081	1.007	1.160	.031
Mixed-Police	12	1.164	1.314	2.450	.014
Security	27	1.225	1.059	1.419	.006
Q = 24.898, df = 2, p < .001
Largest effects
Police	37	1.206	1.097	1.326	<.001
Mixed-police	12	1.186	1.022	1.378	.025
Security	27	1.208	1.072	1.361	.002
Q = 7.219, df = 2, p = .027
Smallest effects
Police	37	1.026	0.955	1.103	.483
Mixed-police	12	1.100	0.943	1.283	.226
Security	27	1.136	0.973	1.327	.107
Q = 14.087, df = 2, p = .001

Table 2 displays the meta-regression results. The independent variables of primary interest are mixed-police and security, which measure the crime prevention effect of these scheme types compared with police schemes. These variables achieved statistical significance in only the largest effects meta-regression model. In this model, both security schemes and mixed-police schemes were positively related to effect size. However, the coefficient for security schemes (0.363) was more than twice as large as the coefficient for mixed-police schemes (0.136). These variables were nonsignificant in both the average effects and smallest-effects meta-regression models. However, we should note that the use of multiple complementary interventions alongside CCTV was significantly related to larger effect sizes in both the average effects and smallest-effects regression models. This suggests that implementing multiple interventions alongside CCTV perhaps should be the primary consideration for practitioners and policymakers. This has important implications for security-led CCTV schemes and is discussed in the next section.

Table 2.

Meta-Regression in Predicting Crime.

Covariate	Coefficient	SE	95% confidence interval		p
Covariate	Coefficient	SE	Lower	Upper	p
Average effects
Mixed-police^a	0.056	0.095	−0.131	0.243	.557
Security^a	0.046	0.085	−0.121	0.213	.592
Car park	0.265	0.146	−0.021	0.550	.069
Residential	0.057	0.084	−0.109	0.222	.501
United Kingdom	0.032	0.080	−0.125	0.190	.687
Active monitoring	0.055	0.077	−0.095	0.205	.473
Multiple interventions	0.198	0.100	0.001	0.394	.049*
Largest effects
Mixed-police^a	0.136	0.112	−0.090	0.3618	.238
Security^a	0.363	0.124	0.121	0.606	.003*
Car park	−0.338	0.208	−0.746	0.070	.104
Residential	−0.069	0.096	−0.256	0.119	.474
United Kingdom	−0.424	0.125	−0.669	−0.180	.001*
Active monitoring	0.127	0.096	−0.060	0.315	.183
Multiple interventions	−0.046	0.081	−0.204	0.112	.567
Smallest effects
Mixed-police^a	0.057	0.099	−0.137	0.250	.565
Security^a	0.003	0.088	−0.170	0.176	.975
Car park	0.371	0.158	0.061	0.681	.020
Residential	0.052	0.085	−0.114	0.217	.540
United Kingdom	0.086	0.083	−0.077	0.249	.302
Active monitoring	−0.016	0.077	−0.168	0.135	.832
Multiple interventions	0.213	0.105	0.008	0.419	.042*

Note. Log odds ratio is the dependent variable for each model. Each regression ran as a random effects model.

“Police” used as the reference category.

p < .05.

The analysis concludes with a test of potential publication bias of the meta-analysis results. We used BioStat’s trim-and-fill procedure to estimate how reported effects would change if bias was discovered and addressed (Duval, 2005). This is based on the assumption that effect sizes should show symmetry around the mean when a representative collection of studies has been obtained. When there is asymmetry, the trim-and-fill procedure inputs the hypothesized missing studies and recomputes a mean effect size. The analysis showed that asymmetry is present (results not shown), and that nine studies should be added to create symmetry. When the effect size is recomputed to include these additional studies, the mean effect size increased from 1.140 to 1.187. However, the 95% confidence intervals of the observed and adjusted ORs overlap, suggesting that the effect sizes are not significantly different. The smallest- and largest-effect versions of the trim-and-fill procedure similarly produced estimates with overlapping confidence intervals (results not shown). From the results of these tests, we can conclude that publication bias did not influence the meta-analysis results.

Discussion and Conclusion

This review’s findings reflect positively on CCTV schemes operated by security personnel. In each of the meta-analyses, security-led CCTV schemes exhibited the largest reduction in crime. The differences in OR effect sizes were particularly magnified in the average effects meta-analysis. In this model, security schemes generated crime reductions of approximately 18% compared with approximately 16% for mixed-police schemes and approximately 7% for police schemes.

It is important to note that the effect of security-led CCTV schemes was less magnified in the meta-regression, which controlled for key factors related to the effects of CCTV on crime. Security schemes and mixed-police schemes were significantly more effective than police schemes in only the largest effects meta-regression model. On one hand, this suggests that these schemes may be preferable to police schemes when the maximum potential effect is achievable. On the other hand, some may give more emphasis to the average effects and smallest-effects meta-regression models given that they represent more conservative estimates.

In both of these models, the use of multiple complementary interventions alongside CCTV was significantly related (car parks also related, especially in smallest) to the CCTV effect. We feel that this finding tangentially supports the increased use of security personnel in CCTV operations.

Using security personnel in the CCTV monitoring function may help to achieve the “force multiplier” effect that policymakers have long sought from video surveillance cameras (Norris, 2003). Traditionally, policymakers have considered video surveillance cameras a force multiplier because they provide more “eyes on the street,” which theoretically increases police presence. However, research has shown that standard CCTV operations detect rather low levels of criminal activity (Piza et al., 2014), which calls into question CCTV’s role as a proactive place-based strategy for increasing guardianship (Weisburd & Majmundar, 2018). In this context, the presence of CCTV alone does not seem to do much to deliver the force multiplier effect that policymakers envision.

Conversely, having security personnel monitoring CCTV cameras may free up police officers to conduct proactive operations in support of surveillance functions, which would serve as a much stronger force multiplier than the conspicuous presence of CCTV cameras alone. In recognition of prior research analyzing the effect of integrating proactive policing units within CCTV operations (La Vigne et al., 2011; Piza et al., 2015), such a strategy would likely strengthen CCTV’s overall crime prevention effect.

This review has two limitations. One has to do with missing information in the included studies, resulting in eight studies that could not be included in the meta-analysis. We can take some comfort that only four of these eight studies were missing information about the scheme operation variable. This bears directly on poor descriptive validity in the reporting of primary studies, something that confronts other systematic reviews. A second limitation concerns the lack of details in some studies about the main roles that security personnel and police play in the deployment of CCTV surveillance and in what capacities security personnel interact with the other parties. Process evaluations could furnish some of this valuable information, and they should be carried out in concert with outcome evaluations.

The findings of this systematic review should provide policymakers with the beginning of an evidence base in considering security-led CCTV schemes as a viable option in deploying CCTV to prevent crime. Until now, little research has been available to help guide decision-making on the use of security personnel, police, or some combination of these two parties in the monitoring of surveillance cameras. Additional factors should also be considered, including financial costs, intervention context, and police-community relations.

There is also a need to better understand why security-led CCTV schemes are more effective in reducing crime. Is it the specialized training of security personnel? Is there something to do with the targeted duties assigned to security personnel? Does it have to do with the role of operating procedures and overall governance for security personnel? Unfortunately, these questions could not be investigated as part of this review. However, we feel there are opportunities for future research to rigorously explore these issues through randomized experiments. For example, Piza et al. (2015) randomly assigned an additional CCTV operator and proactive directed patrol units across preexisting CCTV sites. In a similar fashion, we feel that existing CCTV sites can be randomly assigned experimental factors related to operator functions. Such experiments can involve random assignment of procedural aspects (e.g., training, monitoring policies, hiring in-house vs. outside security operators) of the operator function to better isolate the effect such factors have on the crime prevention effect of CCTV. Furthermore, there is a real need for qualitative research to explore the day-to-day operations and behind-the-scenes activities that guide security personnel working in CCTV projects. Importantly, the findings of qualitative studies need to be integrated with evaluation research findings. We think this is long overdue and much needed to guide policy and practice on the use of CCTV surveillance.

Footnotes

Appendix

Acknowledgements

We wish to thank the co-editors of this special issue and the anonymous reviewers for insightful comments.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Author Biographies

Brandon C. Welsh, PhD, is a Professor in the School of Criminology and Criminal Justice at Northeastern University and the Director of the Cambridge-Somerville Youth Study.

Eric L. Piza, PhD, is an Associate Professor in the Department of Criminal Justice at John Jay College of Criminal Justice, New York.

Amanda L. Thomas is a doctoral student at John Jay College of Criminal Justice, City University of New York, and a retried sergeant of the New York City Police Department.

David P. Farrington, PhD, is Emeritus Professor of Psychological Criminology in the Institute of Criminology at the University of Cambridge and the Director of the Cambridge Study in Delinquent Development.

References

Braga

A. A.

Weisburd

D. L.

Turchan

(2018). Focused deterrence strategies and crime control: An updated systematic review and meta-analysis of the empirical evidence. Criminology & Public Policy, 17, 205–250.

Cornish

D. B.

Clarke

R. V.

(2003). Opportunities, precipitators, and criminal decisions: A reply to Wortley’s critique of situational crime prevention. In Smith

M. J.

Cornish

D. B.

(Eds.), Theory for practice in situational crime prevention (pp. 41–96). Criminal Justice Press.

Duval

(2005). The “trim and fill” method. In Rothstein

Sutton

Bornstein

(Eds.), Publication bias in meta-analysis (pp. 127–144). Wiley.

Gill

Spriggs

(2005). Assessing the impact of CCTV (Home Office Research Study, No. 292). Home Office.

Goold

B. J.

(2004). CCTV and policing. Oxford University Press.

La Vigne

N. G.

Lowry

S. S.

Markman

J. A.

Dwyer

A. M

. (2011). Evaluating the use of public surveillance cameras for crime control and prevention. Urban Institute.

Lipsey

M. W.

Wilson

D. B.

(2001). Practical meta-analysis. Sage.

Norris

(2003). From personal to digital: CCTV, the panopticon, and the technological mediation of suspicion and social control. In Lyon

(Ed.), Surveillance as social sorting (pp. 249–281). Routledge.

Piza

E. L.

Caplan

J. M.

Kennedy

L. W.

(2014). Is the punishment more certain? An analysis of CCTV detections and enforcement. Justice Quarterly, 31, 1015–1043.

10.

Piza

E. L.

Caplan

J. M.

Kennedy

L. W.

Gilchrist

A. M.

(2015). The effects of merging proactive CCTV monitoring with directed police patrol: A randomized control trial. Journal of Experimental Criminology, 11, 43–69.

11.

Piza

E. L.

Welsh

B. C.

Farrington

D. P.

Thomas

A. L.

(2019). CCTV surveillance for crime prevention: A 40-year systematic review with meta-analysis. Criminology & Public Policy, 18, 135–159.

12.

Thompson

S. G.

Higgins

P. T.

(2002). How should meta-regression analyses be undertaken and interpreted? Statistics in Medicine, 21, 1559–1573.

13.

Waszkiewicz

(2013). How effective is the public video surveillance system in Warsaw? In Bjorklund

Svenonius

(Eds.), Video surveillance and social control in a comparative perspective (pp. 153–170). Routledge.

14.

Weisburd

Majmundar

M. K.

(Eds.). (2018). Proactive policing. National Academies Press.

15.

Welsh

B. C.

Farrington

D. P.

(2009). Public area CCTV and crime prevention: An updated systematic review and meta-analysis. Justice Quarterly, 26, 716–745.

16.

Welsh

B. C.

Mudge

M. E.

Farrington

D. P.

(2010). Reconceptualizing public area surveillance and crime prevention: Security guards, place managers, and defensible space. Security Journal, 23, 299–319.

17.

Wilson

(2005). Behind the cameras: Monitoring and open-street CCTV surveillance in Australia. Security Journal, 18, 43–54.