There has been extensive research on the value of closed-circuit television (CCTV) for preventing crime, but little on its value as an investigative tool. This study sought to establish how often CCTV provides useful evidence and how this is affected by circumstances, analysing 251,195 crimes recorded by British Transport Police that occurred on the British railway network between 2011 and 2015. CCTV was available to investigators in 45% of cases and judged to be useful in 29% (65% of cases in which it was available). Useful CCTV was associated with significantly increased chances of crimes being solved for all crime types except drugs/weapons possession and fraud. Images were more likely to be available for more-serious crimes, and less likely to be available for cases occurring at unknown times or in certain types of locations. Although this research was limited to offences on railways, it appears that CCTV is a powerful investigative tool for many types of crime. The usefulness of CCTV is limited by several factors, most notably the number of public areas not covered. Several recommendations for increasing the usefulness of CCTV are discussed.
Introduction
Closed-circuit television (CCTV) surveillance cameras are widely used in policing, but that use is controversial. The United Kingdom (UK) government has described CCTV as “vital” for detecting offenders (Porter ‘2016), while the Washington, DC, Metropolitan Police Department (2007, p 2) argued that it is often “invaluable to police investigations”. On the other side of the debate, the campaign group (Liberty 2016) argued that extensive use of CCTV “poses a threat to our way of life” and that “widespread visual surveillance may well have a chilling effect on free speech and activity”. Similarly, the American Civil Liberties Union claimed that public CCTV surveillance creates “an almost Orwellian potential for surveillance and virtually invite[s] abuse” (Steinhardt 1999).
In the academic literature, there has been discussion of how CCTV fits into broader conceptions of surveillance (Hier –2004; Koskela ”—–2003) and the extent to which it increases or changes the nature of state or corporate power over citizens (Fyfe and Bannister –1996; Norris and Armstrong –1998). ]
Concerns have been raised that CCTV surveillance may restrict the diversity and vibrancy of life in public spaces (Bannister et al. –1998), or contribute to the exclusion of some groups in society (Reeve –1998). There has also been political debate about the proper balance between ensuring the effectiveness of CCTV and protecting the privacy of citizens (Sheldon –2011).
Although the debate about CCTV has been both long lasting and wide ranging, empirical evidence on the topic has so-far not covered all of its aspects. This article will attempt to provide evidence to inform one area of this debate about which evidence is currently limited: the extent to which CCTV is valuable for criminal investigations. The next section contains a review of the existing literature, followed by an explanation of the mechanisms that may influence the effectiveness of surveillance cameras in investigations. The following section will describe the data used in this study, derived from police reports of crimes on the railway network of Great Britain. The results section will describe how often CCTV has been useful in crime investigations, and in what circumstances. Finally, the implications of these results for policy makers and practitioners will be discussed.
Existing Literature
Given the controversial nature of CCTV, surprisingly little is known about how it is used and how effective it is in achieving many stated aims. CCTV has several potential applications for public safety, and has been deployed with the intention variously of preventing crime, detecting offences, improving the response to emergencies, assisting in the management of places and reducing public fear of crime (Ratcliffe 2011, p 15). CCTV can also be used for purposes not related to public safety, such as monitoring transport-passenger flows and investigating complaints against facility staff (National Rail CCTV Steering Group 2010, p 7).
Of these potential applications, almost all research attention to date has concentrated on the use of CCTV to prevent crime (Honovich 2008). Early studies by Mayhew et al. (1979) and Webb and Laycock (1992) suggested that CCTV was effective at reducing robberies at London Underground stations, although the evaluation methods used had some limitations. Since then, the subject has received substantial research attention with mixed empirical results. For example, several evaluations have found CCTV to be effective at reducing thefts in car parks (Poyner and Webb 1987; Tilley 1993) but others have shown it to have little or no impact on crime in residential areas (Gill and Spriggs 2005). A systematic review by Welsh and Farrington (2008) of 41 studies concluded that CCTV is effective at preventing some types of crime in some circumstances, but that the evidence suggests it has a more-limited impact than its widespread deployment may suggest.
In contrast to the extensive literature on the value of CCTV for crime prevention, there is little research on how useful cameras are for other purposes. Ditton and Short (‘–1998) found that in the 2 years after the installation of a CCTV scheme in a Scottish town, the proportion of crimes that were solved by police increased from 50 to 58%, with some offences showing larger increases than others. However, no information was given about whether these changes were statistically significant, and rates were only provided for some types of crime (the primary focus of the study was on crime prevention). In Australia, Wells et al. (2006) found that monitored CCTV in two suburbs led to the early arrest of a small number of offenders at the scenes of crimes, but did not look at whether recordings were useful in the subsequent investigations.
Limited evidence can be found in research on solvability factors: the features of an offence that determine the likelihood of the case being solved. Paine (‘2012) found CCTV to not be associated with higher detection rates for residential burglary. For non-residential burglary, Coupe and Kaur (–2005) found that CCTV being installed in a building was associated with double the rate of detections compared to other buildings, driven by the increased availability of suspect descriptions. Since this study used data from the year 2000, it is possible that subsequent developments in technology may have influenced the effectiveness of CCTV in solving this type of crime. For example, modern cameras are likely to provide higher-resolution images, and digital (as compared to tape-based) storage allows images to be retained for longer (Taylor and Gill –2014). Existing research on solvability factors is limited because it is largely focused on the investigation of a single crime type (burglary).
In the context of this limited academic evidence, several organisations have produced reports on the topic of the value of CCTV for investigation, some of dubious quality. For example, Davenport (2007) summarised an unpublished report by the Liberal Democrat political party which concluded that CCTV cameras were ineffective simply because London boroughs with more cameras did not have a higher all-crime detection rate. The group appeared to have made no attempt to control for confounding variables or for different types of crime. Despite the poor quality of the analysis, this report has subsequently been cited in the media (e.g. by Bates ”2008) as proof that CCTV is ineffective in investigations.
Journalists have also carried out their own analyses. Staff from The Scotsman (2008) newspaper reported that in a 4-year period CCTV cameras in Scotland had observed more than 200,000 incidents, with responding police officers making arrests in 14% of cases. However, no details were given on whether those arrests led to charges, whether further suspects were identified later or how the headline statistic varied in different circumstances or for different types of crime. In San Francisco, journalists found that cameras had given detectives new avenues of investigation in seven of 33 violent felonies committed in a crime hotspot over a 2-year period (Bulwa and Stannard 2007). Meanwhile the London Borough of Hackney (‘2016) reported that over a 12-year period the use of CCTV had been associated with more than 27,000 arrests, although it gave no further details. Edwards (2009) reported that CCTV evidence was gathered in 86 of 90 murder investigations and was judged by senior police officers to have been valuable in 65 of those cases.
There appears to be some disagreement within the police service as to how effective CCTV cameras are in criminal investigations. Several news outlets summarised a report from the London Metropolitan Police Service (MPS) that appeared to be highly critical of its usefulness. Bowcott (2008) reported that “only 3% of street robberies in London were solved using CCTV images”, although other articles reported that the 3% statistic applied to all crime (e.g. Johnson ‘2008). In another article based on the same report, Edwards (”2008) wrote that “up to 80 per cent of CCTV footage seized by police is of such poor quality that it is almost worthless for detecting crimes”. Hickley (2009) quoted a police spokesperson as saying that “in 2008 less than 1,000 crimes were solved using CCTV despite there being in excess of one million cameras in London”. However, to the present author’s knowledge, the report itself remains unpublished and no information is available on the methods used, nor any more details of the conclusions.
In contrast, the majority of British officers surveyed by Levesley and Martin (2005) believed that CCTV was a useful investigative tool. A report on the value of CCTV commissioned by Dyfed-Powys Police in Wales argued that cameras were valuable in the detection of crime, citing the opinions of police investigators and local prosecutors. However, the report also recommended that live-monitoring of CCTV cease because it was ineffective at preventing crime or improving the initial response to incidents (Instrom Security Consultants 2014). Several municipalities in Britain have decreased their investment in CCTV in response to recent budget cuts (Merrick and Duggan —2013).
Overall, little appears to be known about how the usefulness of CCTV for investigation varies across crime types or circumstances, which is likely to be important in any attempts to make CCTV more useful. The present exploratory study attempted to provide some evidence in these areas.
How Might CCTV Help Crime Investigations?
Before turning to the research questions addressed in this study, it is necessary to consider exactly how CCTV might provide useful evidence in a criminal investigation.
A criminal investigation can be thought of as a series of questions: who was involved in an incident, where did it happen, what happened, when did it happen, why did it happen and how were any offences committed, known as the ‘5WH’ investigation model (Cook et al. ”2016; Stelfox 2009). CCTV may be useful in answering at least two of these questions: what happened and who was involved (La Vigne et al. 2011).
A good-quality recording could potentially allow investigators to watch an entire incident unfold in detail, providing information about the sequence of events, the methods used and the entry and exit routes taken by the offender. Even if this is not possible, CCTV may be useful in corroborating or refuting other evidence of what happened, such as witness testimony (College of Policing 2014). Recordings may also provide information that investigators can use to contextualise other evidence (Levesley and Martin 2005).
CCTV may assist in identifying who was involved in a crime either directly, as when a suspect is recognised by someone viewing the recording, or indirectly, such as when the recording shows a suspect touching a surface from which police are then able to recover forensic evidence (Association of Chief Police Officers 2011). Images can also be used to identify potential witnesses (La Vigne et al. 2011, p 27). CCTV may be less useful in answering some of the other 5WH questions. For example, even a good-quality recording may shed little light on why a crime was committed.
In order for CCTV to be useful in answering investigative questions, certain circumstances are required. There are few legal restrictions on the ability of police officers to use CCTV recordings of public places during investigations. In the UK, for example, operators of camera systems can provide recordings to the police without a warrant (Information Commissioner’s Office ”2015). In the United States (US) a similar system operates, as long as the recording is of a place in which people do not have a reasonable expectation of privacy (Chace 2001). As such, the limiting factors on the use of CCTV are likely to take other forms.
Figure 1 summarises the process by which these circumstances may come about, broken down into three stages. In the first stage, CCTV evidence is not available, either because the police have not taken steps to obtain it or because no recording exists for technical reasons.
In the second stage, a CCTV recording is available but—perhaps because of the recording quality—is not useful to the investigation. In the final stage, a recording is both available and useful to the investigation.
The process by which CCTV can be available and useful to an investigation
The present study aimed to explore the usefulness of CCTV for crime investigation by examining the stages in this process. Specifically, two research questions were addressed:
- How often do CCTV cameras provide useful evidence in criminal investigations, and how does this vary for different types of crime?
- In what circumstances is CCTV most likely to be useful in criminal investigations?
It might be thought that CCTV is useful in a case only if it leads to a suspect being convicted. However, this approach presents at least two problems. Firstly, it would require a potentially unreliable counterfactual test of the weight of evidence that would have been available in a particular case in the absence of CCTV recordings. Secondly, it ignores the potential value of video recordings in allowing officers to eliminate a suspect from their enquiries, or to identify where a person has made a false report of crime.
Data and Methods
Answering the research questions addressed in this study required a source of data on the investigative value of CCTV. Although police forces collect a wide range of information about investigations for administrative and legal purposes, that information is not always stored in a way that makes it accessible for secondary analysis for research purposes.
For example, case files may contain information on whether CCTV recordings were used in a particular investigation, but such files are often held either on paper or in electronic formats that are not easily searchable. As a result, research based on case files (e.g. Jordan ‘–2004; Cockbain et al. –2011) tends to use small samples, potentially limiting its power to detect differences between crime types, circumstances and so on. The present research takes advantage of one police force having collected data on the value of CCTV over a long period of time and stored it in an accessible format. In 2010, British Transport Police (BTP)—the specialist police force for railways in Great Britain—added two questions to the electronic form that first-line supervisors are required to complete at the end of a criminal investigation conducted by an officer under their command. The first question asks the supervisor whether CCTV has been useful in the investigation and the second (required only if the answer to the first question is ‘no’) asks the supervisor to choose from a list of reasons why CCTV has not been useful in that case.
From these two questions it is possible to construct a third variable showing whether a CCTV recording existed in a case, by combining those cases where CCTV was useful with those where a recording of the location at the relevant time did exist but was not useful (for example because the images were of insufficient quality or because the wrong images were extracted from the system).
The present study used the answers to these questions together with other data about each offence, such as whether a case had been ‘solved’. In common with standard practice among UK police forces, crimes were deemed to be solved if a prosecutor (or, in minor cases, a specially trained police officer) decided that there was sufficient evidence against a suspect to send the case to court.
Again in common with police practice, such cases are referred to below as having been ‘detected’ (for further details, see Home Office 2016).
In the 5 years between January 2011 and December 2015, 251,195 notifiable crimes were recorded by BTP, or about 138 crimes each day.Footnote1 Since the data on CCTV usefulness is gathered only at the conclusion of an investigation, data for this study were extracted from the BTP crime-recording system at the beginning of March 2016 to maximise the chance of the relevant questions having been completed. Data on the usefulness of CCTV were missing in 4,768 cases (1.9% of the total).
The use of data from BTP was valuable for two reasons. Firstly, that force had for several years been collecting data that were not widely recorded elsewhere, providing a large sample of offences to study. Secondly, the railway network policed by BTP has a very large number of CCTV cameras. The total number of cameras on the railway network is not known, since the 25 train operating companies each operate their own camera networks and there is no central registry of cameras. However, the BTP CCTV Hub is able to monitor around 30,000 cameras at railway stations around the country (British Transport Police 2016). There are also an unknown number of cameras installed on-board trains.
The large number of cameras means that CCTV will cover the sites of a substantial proportion of recorded crimes. This helps to deal with a potential limitation of the studies discussed above, which is that only a few of the offences reported to those forces are likely to have occurred close to a CCTV camera. This in turn would make it more difficult to answer the second research question.
While the availability of data from BTP made this study possible, the use of data from a specialist police force may limit the generalisability of the results. The mix of crimes investigated by BTP is likely to be different from that seen by local police forces.
Another limitation of this data source is that in cases where supervisors identified that CCTV had been useful in an investigation, it was not possible to identify in what way it had been useful. This is a limitation of the data collection, which in turn is likely to reflect a focus within the police on improving CCTV usefulness by focusing on the reasons why recordings were sometimes not useful. It is likely that supervisors classified cases with reasonable consistency, since the second question described above asked them to pick from a list of reasons (shown in Table 1) why recordings were not useful. Consistency is also likely to have been maintained by the questions asked of supervisors being constant over time. Nevertheless, it is not possible to be sure that supervisors categorised cases with complete consistency. Neither can the possibility of bias be excluded, although the author is aware of no obvious sources of bias.
The data did not distinguish between different types of CCTV. Systems can vary from extensive high-definition networks to single cameras that produce grainy images (Gill and Spriggs 2005; Taylor and Gill –2014). It is therefore likely that the effectiveness of systems will vary. Information about system type was not recorded in the BTP data and so it was not possible to explore these distinctions. However, there may be less variation in systems on railways compared to systems in other environments, because there are industry standards for rail CCTV networks (National Rail CCTV Steering Group 2010).
Despite these limitations, to the author’s knowledge the sample used here represents the best-available large-scale dataset on the value of CCTV for investigating crime. More-detailed data could have been collected for a smaller sample of crimes (for example by interviewing investigators or reviewing case files), but it would have been impractical to do this for a large-enough sample to allow analysis of different crime types and crimes occurring in different circumstances.
Results
How often is CCTV Useful?
Using the distinction between availability and usefulness shown in Fig. 1, CCTV was available in the investigation of 111,608 offences in the 5 years between 2011 and 2015—45.3% of all crimes recorded by BTP. CCTV was classified as being useful in 72,390 investigations—29.4% of all recorded crimes and 64.9% of crimes for which CCTV was available. Camera recordings were, for example, useful in the investigation of 1,223 assaults causing serious injury, 4,120 assaults causing minor injury, 1,365 personal robberies and 2,810 sexual offences. Table 1 provides further details.
