Body‑worn camera activation in prisons: understanding correctional officers’ decision‑making and use of discretion

(0123456789


Introduction
Correctional agencies globally are adopting body-worn cameras (BWCs) as a tool for officers working with prisoner populations (e.g. Beales and Marsh 2016;Dodd et al. 2020;Hong Kong Correctional Services 2018;McLennan 2019;Ministry of Justice 2017;Polley and Smith 2020). This decision is often motivated by concerns about corruption and excessive use of force by correctional officers (Crime and Corruption Commission Queensland (CCCQ) 2018). BWCs have also been touted as a tool to safeguard officers, by reducing the opportunity for assaults on staff (Polley and Smith 2020) and aiding correctional agencies to manage and resolve incidents of prison violence (Ministry of Justice 2019; Roberts 2020). They can be used to gather evidence in the event of critical incidents, including riots, assaults on officers, use of force incidents, and contraband raids (Roberts 2020). The primary aim of introducing BWCs in prisons, then, is to make the prison environment safer for all parties (Polley and Smith 2020;Sydes et al. 2020).
BWCs have also been introduced in other public-facing roles with the aim of reducing rates of workplace violence, with some success. Ariel et al. (2019) reported a 47% decrease in the odds of assault against railway staff who were equipped with a BWC following their introduction in train stations in England and Wales. BWCs have also been employed in mental health wards, with Ellis et al. (2019) reporting a reduction in the overall seriousness of aggression and violence in reported incidents when nurses were equipped with a BWC. However, research on the use and impact of BWCs within prisons is still in its infancy. Dodd et al. (2020) explored correctional officers' attitudes towards BWCs, finding widespread support for their use in prisons. There have been mixed findings, however, as to whether BWCs increase officer safety (both in terms of physical and professional safety) . Beales and Marsh's (2016) evaluation showed that BWCs can be an effective de-escalation tool and equipping correctional officers with cameras increased their actual and perceived physical safety. Similarly, most correctional officers in Polley and Smith's (2020) research described feeling physically safer when equipped with a BWC. However, many officers in that study were also fearful of 'negative accountability', namely the perception that BWC footage could be used to discipline or reprimand them. Other research shows that while correctional officers did not perceive BWCs as increasing their physical safety or having a civilizing effect on prisoner behaviour, they did feel the presence of BWCs improved their 'professional' safety, by reducing the likelihood of prisoners making false complaints against them .
To date, no studies have explored correctional officers' decision-making or use of discretion regarding BWC use in prisons. This is an important area for research since the utility of this technology depends upon how it is used (Young and Ready 2018). The amount of discretion provided to BWC users and how this discretion is exercised are also key considerations to realizing the benefits of this technology (Newell and Greidanus 2017), like increasing officer safety. Further, we argue that correctional officers' use of discretion when using BWCs represents a significant and potentially high-stakes decision because the use of these cameras with prisoner populations has the potential to raise significant and serious privacy concerns (Bülow 2014). The presence of BWCs, particularly in circumstances where a 'discretionary' recording policy is implemented, may suggest an "implied power dynamic that staff have over prisoners, whereby the decision to record is an overt intervention, intended to resolve issues of violence and unruly behaviour" (Polley and Smith 2020, p. 6). The unintended consequence of this power dynamic, however, may be to exacerbate prison violence rather than alleviate it (Polley and Smith 2020). For these reasons, there is a pressing need to better understand the exercise of discretion by correctional officers when using BWCs in a carceral environment.
This study, thus, explores correctional officers' BWC decision-making while managing prisoner populations. Using a mixed-methods approach, we draw on a state-wide survey of officers from prisons where a BWC programme was recently introduced (N = 510) and in-depth interviews with officers and other staff involved in the BWC programme (N = 34). From these data, we consider what predicts correctional officers' willingness to turn their BWCs on in different scenarios. We also highlight the cues used by officers to guide their BWC activation and explore how they use their discretion to navigate competing considerations of privacy and security. By doing so, we provide the first empirical examination of correctional officers' decisions when it comes to using BWCs in prison.

Police officer discretion when using BWCs
Due to the lack of research examining the exercise of discretion by correctional officers when using BWCs, we begin our discussion with the available policing literature. Issues surrounding the discretion given to police officers regarding BWC activation are recognized as one of the most contentious issues for police agencies to contend with (Edmonton Police Service 2015; Newell and Greidanus 2017). There is broad disagreement regarding whether police officers should have their BWCs recording throughout their entire shift (e.g. Ariel et al. 2016;Young and Ready 2018) or whether they ought to be empowered to decide if and when to activate their cameras (e.g. McClure et al. 2017). Those who favour the former position argue that "BWCs will have a minimal deterrent effect when officers have broad discretion around activating their cameras" (Peterson and Lawrence 2019, p. 8). Indeed, it has been suggested that when a lack of positive outcomes from BWCs has been observed in evaluation research, officer discretion in activating (or not activating) the camera may be to blame (Adams et al. 2021). Others argue that the right to privacy for officers, members of the public, and victims, together with practical considerations like the amount of footage to be stored, may preclude a continual recording policy (Newell and Greidanus 2017).
Researchers have examined the rate at which police officers working within a discretionary BWC policy framework activate their cameras. Katz et al. (2015), for example, observed relatively low activation rates amongst officers, with an average of under 30 per cent. Comparing activation rates under a mandatory use policy and a discretionary policy, Young and Ready's (2018) research showed that officers who volunteered to wear a BWC were significantly more likely to activate their cameras during incidents (67.4%) compared to officers whose use of BWCs was compulsory (51.4%). Further, when officers were provided more discretion (rather than operating under a mandatory policy), activation rates decreased by 27% (Young and Ready 2018). Various reasons have been given to explain why officers may not activate their BWC, including that they forgot or deliberately failed to activate the device, there was insufficient time for activation, the BWC malfunctioned, or the officer did not know how to operate the camera (Ariel 2016;Jennings et al. 2014;Lawrence et al. 2019).

The importance of correctional officer decision-making around BWC use in prisons
There are unique privacy and ethical considerations for correctional officers who use BWCs in custodial settings. The extant literature on the use of BWCs (predominantly by police agencies) focuses primarily on the potential benefits offered by this technology, with much less attention given to the possible ramifications of this surveillance tool (Lin 2016). As Adams and Mastracci (2017) argue, technology like BWCs is often adopted in haste. Consequently, the introduction of BWCs "often outpaces the laws and regulations that would ensure their appropriate use, and the negative consequences are rarely anticipated, particularly as they relate to privacy concerns" (Adams and Mastracci 2017, p. 313).
Correctional officers have intensive and continuing contact with the prisoners under their management (Liebling 2000). Although closed-circuit television cameras have been used in prisons for many years, BWCs offer a more dynamic and intrusive form of supervision due to their ability to capture close-range recordings involving both audio and video (Miller and Tolliver 2014). While reduced privacy rights and some level of prisoner supervision are necessary for the safe operation and security of prisons, there is evidence that "strong and extensive surveillance may be directly or indirectly harmful and morally problematic" for prisoners (Bülow 2014, p. 12). Intensive intrusions of prisoners' privacy may result in their experience of shame, strong and enduring feelings of being constantly suspected, and perceptions they are untrustworthy and incapable of reform (Bülow 2014). However, this is a contentious area, with debate about whether prisoners have a right to any level of privacy and whether this is even possible within a prison environment (Moran et al. 2013). Regardless, when activating their BWC, correctional officers must balance considerations of prisoner privacy with other important values, such as the safety and security of the prison.
Finally, as Polley and Smith (2020) argue, correctional officers' decision-making around BWC use is particularly important when a discretionary BWC policy is adopted, whereby individual officers must decide if, when, and where to activate their BWC. In these circumstances, officers' use of BWCs may be viewed by prisoners as an "overt intervention" against them, which reflects the underlying power dynamics between prisoners and officers (Polley and Smith 2020, p. 6). Thus, rather than improving prisoner behaviour, the use of BWCs may worsen or escalate situations (Polley and Smith 2020). This is of particular concern when considering the disproportionately high rates of mental illness amongst prisoners (Haney 2017;Mulvey and Schubert 2017), which, when combined with BWCs, has the potential to inflame acts of violence (Taylor 2016).

Introduction of BWCs in Queensland and policies governing use
In 2017, BWCs were introduced on a trial basis by Queensland Corrective Services (QCS) (CCCQ 2018;Queensland Government 2017). Following a successful trial, approximately 150 BWCs were in use across 12 high-security prisons across the state in 2018 (CCCQ 2018). 1 Their use is governed by the Deputy Commissioner Instruction (DCI) titled "Body Worn Camera, Deployment and Use" (Queensland Government 2017). The DCI prescribes the circumstances in which correctional officers ought to activate their BWC. It states that officers should record "whilst responding to operational incidents, use of force incidents, or at times where a corrective services officer reasonably considers there to be a need to record the interaction" (Queensland Government 2017, 4). It further sets out circumstances in which a BWC recording must not be made, including searches of prisoners requiring the removal of clothing (ROC) or in places where a reasonable expectation of privacy exists. Importantly, while the policy is designed to ensure that an officer's camera is activated in situations where conflict or use of force is likely to occur, it is recognized that "some latitude for [officer] discretion is allowed" (CCCQ 2018, p. 2).

Study aims
This study explores correctional officers' decision-making around BWC activation during encounters with prisoners. Understanding how officers exercise their discretion, including how and in what situations they would turn their camera on or off, and the ways they navigate the use of this technology amidst prisoner privacy and security concerns can provide invaluable insights for correctional agencies. This information may also highlight possible issues raised by the adoption of a discretionary BWC policy, where there are "mandatory and prohibited times for activation", but a "messy middle where officers are granted broad discretion" (Adams et al. 2021, p. 693).
Drawing from a mixed-methods study of correctional officers in Queensland, Australia, this study answers the following research questions: 1. What predicts correctional officers' willingness to turn their BWCs on in different scenarios?
2. What cues do correctional officers rely upon to guide their BWC activation? 3. How do correctional officers navigate privacy and security concerns when using a BWC within prison?
The first question is answered through a quantitative analysis of survey data, while questions two and three draw from qualitative interview data.

Research phase 1: surveys
A state-wide survey of correctional officers was conducted between December 2018 and March 2019. This survey (which had a completion time of roughly 15 min) was distributed to all currently employed correctional officers in Queensland (approximately 2500 officers). An invitation to complete the online Qualtrics survey was distributed via email. 2 Survey questions were derived from prior research (e.g. Gaub et al. 2016;Smykla et al. 2016;Tankebe and Ariel 2016) or developed by the research team. In addition to collecting demographic information, officers were asked questions relating to their support of BWCs and the circumstances requiring activation (participant characteristics are presented in Table 1).

BWC decision-making scenarios
Participants were presented a list of 15 scenarios and asked to indicate the scenarios in which they believed an officer should activate their BWC. 3 There were eight scenarios where participants showed almost universal support for BWC activation. These scenarios were as follows: a riot (98%; n = 498); a prisoner attacking (97%; n = 494) or threatening a correctional officer (96%; n = 490); a prisoner's escalating behaviour towards an officer (95%; n = 486); violence between prisoners in the yard (95%; n = 482); a fight in a medical unit (93%; n = 476); a prisoner doing drugs in the bathroom (90%; n = 457); and an officer physically restraining a prisoner (88%; n = 446). Alternatively, there were three scenarios where only a minority of participants indicated that BWCs should be activated-prisoners making complaints about officers in private (38%; n = 193); during a ROC search (12%; n = 61); and in close vicinity of a prisoner's medical consultation (8%; n = 39). The remaining four scenarios revealed mixed decisions, albeit a general tendency for participants to indicate that activating their BWC would be appropriate. From most to least support, the responses were as follows: the discovery of contraband in a prisoner's cell (84%; n = 427), a verbal disagreement between prisoners in the gym (74%; n = 378), a correctional officer threatening a prisoner (70%; n = 359), and a correctional officer addressing a prisoner's behaviour (70%; n = 358). Within these responses some trends appear, prompting two areas for further exploration. The scenarios with almost universal support from participants represent scenarios involving imminent safety concerns. On the other hand, in the scenarios where only a minority of participants indicated a BWC should be activated, many of these scenarios represent some element of privacy or breach of privacy. For example, while most participants agree a BWC should be turned on for an incident involving a fight in a medical unit (a scenario with safety concerns), few participants support recording in close vicinity of a prisoner's medical consultation. A question requiring analysis, therefore, is what differentiates officers who believe it is appropriate to turn the camera on in both scenarios, only the scenario with safety concerns, or neither scenario. Further, although both of the following scenarios may represent a safety concern, there was very strong support for BWCs to be used when a prisoner threatens an officer, but less support when an officer threatens a prisoner. Again, we wanted to determine what differentiates officers who were singularly concerned with officer safety from those who additionally considered prisoner safety (or neither scenario) as a reason to activate their BWC.

Outcome variables
Policy adherence (privacy) was created by coding participant responses to the two items asking if an officer should turn on their BWC "in close vicinity of a doctor's or nurse's consultation with a prisoner" (a general privacy issue) and "a fight in a medical unit" (a general safety issue, but with potential privacy implications). The variable was coded as 0 "No medical unit use" (participant believes BWC should not be turned on in either situation relating to medical units; n = 34), 1 "Safety concern use" (participant believes BWC should be activated in the fight/safety scenario only; n = 437), or 2 "Any medical unit use" (participant believes BWC should be activated in both scenarios; n = 39). 4 This variable allowed us to explore what differentiated officers who were willing to violate the DCI, which states "a BWC is not to be used in consultation rooms of health centres, clinics, and hospitals, unless the officer is responding to an emergent situation" (QCS 2017). On that basis, the "Safety concern use" group would act in adherence to the policy.
Threatening situations was created by coding participant responses to items where participants indicated if an officer should activate their BWC in the scenarios of "A prisoner threatening a correctional officer" (Prisoner threat) and "A correctional officer threatening a prisoner" (Correctional officer threat). The variable was coded as 0 "No threats" (participant believes BWC should not be turned on in either situation; n = 20), 1 "Prisoner threat only" (participant believes BWC should be activated in the prisoner threat situation only; n = 131), or 2 "Both threats" (participant believes BWC should be turned on in both scenarios; n = 358). 5 Although the DCI does not specifically cover threatening behaviours, we suggest officers who activate their BWC in both scenarios would likely be acting in alignment with the policy's intention (by recognizing the importance of both officer and prisoner safety). Response categories for each statement range from 1 'strongly disagree' to 7 'strongly agree'

Analytic strategy
We adopted the analytical approach of conducting multinomial logistic regression (MLR) models, as both outcome variables were categorical in nature. This analysis technique can also be used where an outcome variable has three or more categories (Kaufman 2018). The goal of this type of analysis is to find the best fitting model to describe the relationship between an outcome variable and a set of predictor variables (Hosmer et al. 2013). Specifically, logistic regression predicts an outcome variable given one or more continuous or categorical predictor variables and emphasizes the probability of a particular outcome for each case (Tabachnick and Fidell 2007). This approach was best suited to answering our research question of what differentiates officers who believe it is appropriate to activate their BWCs in certain scenarios. All predictor variables were included in the final MLR models. Given the small sample size and their lack of significant relationships with the outcome variables, participant characteristics (e.g. gender, age) were not included in the final models. Previous research also shows that demographics do not predict BWC activations (Adams et al. 2021).

Research phase 2: interviews
We conducted follow-up interviews with a sample of correctional officers and other QCS staff to gain a more comprehensive understanding of the survey results. Interviewees were recruited through the survey or via word-of-mouth. This resulted in a sample of 34 interviewees, including 25 correctional officers from nine prisons across Queensland. The remaining interviewees worked as intelligence analysts, in violence prevention, and centre or QCS management roles. Interviewees ranged in corrections experience, with the largest proportion (26%) having between 2 and 5 years of experience. Both male (n = 24) and female (n = 10) staff were interviewed, and most interviewees had experience wearing a BWC (n = 23), while the remaining 11 interviewees were in positions where BWCs were not typically worn. Interviews 6 As shown in Table 2, reliability analyses showed all scales had a Cronbach's alpha value of more than .70, the traditionally accepted cut-off value for scale construction (Bonett and Wright 2015).
were conducted in person (n = 27) or by telephone (n = 7) and lasted on average 35 min. All interviews were audio-recorded (with interviewees' permission) and transcribed. A thematic analysis of the transcripts was carried out using NVivo to identify the key themes (Boyatzis 1998).

Research findings: phase 1 quantitative results
We turn now to the presentation of our findings, beginning with the quantitative results from the MLR analysis.

MLR model 1: predictors of policy adherence (privacy)
Table 3 shows the MLR model results predicting officers' willingness to activate their BWC in scenarios involving potential privacy concerns in the medical unit. The first contrast is between "Safety concern use" participants (those who adhere to the DCI policy and support recording during a fight in a medical unit, but not in the general vicinity of a medical consultation) and "No medical unit use" participants (who do not support BWCs being turned on in either scenario). The latter group acts as the reference group in this comparison. The MLR results show that three predictor variables were significant in this model. To begin, for every 1-point increase in officer support for the use of BWCs in prisons, the odds of a participant activating their BWC for safety concerns only, relative to those who would not activate at all, increase 1.98 times. Similarly, for every 1-point increase in perceptions that BWCs Table 3 Multinomial logistic regression model predicting officers' willingness to turn on BWCs in scenarios involving privacy concerns (in a medical unit) N = 456. 'Safety Concern use' n = 393; 'No medical unit use' n = 29; 'Any medical unit use' n = 34 *p < 0.05, **p < 0.01, ***p < 0.001 Contrast 1: 'Safety concern use' vs 'no medical unit use' protect officers professionally, the odds of activation for the safety situation increase 1.62 times. Finally, for every 1-point increase in participants' preferences for BWCs to be turned on in all prisoner interactions, the odds of activating the BWC for the safety concern, compared to no medical unit use, increase 1.49 times. The second contrast is between those officers who would activate their BWC for "Any medical unit use" (those who would activate for both a fight and a medical consultation) and the "No medical unit use" group (those who would activate in neither scenario), with the latter category acting as the reference group. In this comparison, with every 1-point increase in perceptions that BWCs protect officers' professional safety, the odds of BWC activation in both scenarios, relative to activation in neither scenario, increase 1.93 times. Further, with every 1-point increase in participants' preferences for BWCs to be turned on in all prisoner interactions, the odds of activating the BWC for both medical scenarios increase 2.33 times.
Finally, we compared officers in the "Safety concern use" group with those in the "Any medical unit use" group, with the latter acting as the reference category. Here, two predictor variables significantly predicted BWC activation. Namely, with every 1-point increase in participants' perceptions that BWCs affect prisoner interactions (through correctional officers behaving more fairly), the odds of BWC activation for safety concerns alone decrease by 0.64. Because of issues associated with interpreting odds ratio values of less than 1, this odds ratio can be inversely represented and interpreted as an odds ratio with a value greater than 1 (Osborne 2006). Thus, it is equivalent to saying that for every 1-point decrease in officers' support for the beneficial impacts of BWCs on officers' behaviour, the odds of supporting activation for only the safety scenario increase by 1.56 times (i.e. 1/0.64). Likewise, with every 1-point decrease in participants' preferences for BWCs to be turned on for all prisoner interactions, the odds of BWC activation for the safety scenario alone increase by 1.56 times (1/0.64). In terms of model fit, the MLR significantly predicts the outcome variable better than the intercept-only model, χ 2 (16, N = 456) = 79.207, p < 0.001. The pseudo-R 2 measure of McFadden's index has a value of 0.175, a moderate effect size (Tabachnick and Fidell 2007). Based on these measures, the MLR model appears to be a good fitting model. Table 4 shows the MLR model predicting officers' willingness to turn on BWCs during threatening situations. The first contrast is between the "Prisoner threats only" group (those who would only turn on when prisoners represent a threat to officers) and the "Both threats" group (participants who would activate their BWC where either a prisoner or correctional officer is displaying threatening behaviour). The latter group acts as the reference category in this comparison. In this scenario, officers who would activate their BWC for both threatening situations are likely to be acting in accordance with the DCI.

MLR Model 2: predictors of activation in threatening situations
In this model, only general views regarding the fair treatment of prisoners was a significant predictor. For every 1-point increase in officers' support for the fair treatment of prisoners, the odds of BWC activation for the prisoner threat scenario, relative to the recording of both threats, decreased by 0.44. Put another way, for every 1-point decrease in support for fair treatment of prisoners, participants were 2.27 (1/0.44) times more likely to support activation only in scenarios involving a prisoner threatening an officer (and not also when an officer was threatening a prisoner).
Regarding the second contrast, comparing participants who supported turning BWCs on for "No threats" and those who supported BWC activation for both prisoner and officer threats, two predictor variables significantly predicted camera activation. Namely, with every 1-point decrease in officers' perceptions that BWCs ensure their professional safety, participants were 2 (1/0.5) times more likely to activate in neither scenario, relative to activating in both scenarios. Further, with every 1-point decrease in officers' perceptions that BWCs have a beneficial effect on improving prisoner interactions by altering officer behaviour, participants were 2.70 (1/0.37) times more likely to have their camera turned off in both scenarios.
Finally, we compared officers in the "Prisoner threat only" group with those in the "No threat" group, with the latter acting as the reference category. Here, with every 1-point increase in participants' perceptions that BWCs affect prisoner interactions by improving correctional officers' behaviour, the odds of supporting BWC activation in situations involving prisoner threats towards officers (relative to not recording at all) increase 2.73 times.
Model fit statistics showed the MLR model significantly predicts the outcome variable better than the intercept-only model, χ 2 (16, N = 455) = 139.221, p < 0.001. The pseudo-R 2 measure of McFadden's index has a value of 0.214, a moderate effect size (Tabachnick and Fidell 2007). Based on these measures, the model appears to be a good fitting model.

Research findings: Phase 2 qualitative findings
We also had the benefit of interview data to more deeply explore the cues used by correctional officers to guide their BWC activation and, further, to understand how officers navigate privacy and security concerns when using this technology with prison populations.

Cues used by officers to guide BWC activation
We asked officers to describe the behavioural or other cues they may rely upon when deciding whether to activate their BWC. For many officers, their decision-making relied on specific physical and verbal cues from prisoners: I think you can usually tell by the prisoner's body language, nine out of ten [times]. The unit that I'm in, I've gotten to know the prisoners. I manage that unit. So, I know the live wires in there who can potentially tip. You can see it in their body language. A lot of the time, they go very rigid, the fists start to clench, or they start pacing up and down. You can see that indication. If you know that you don't have a rapport with that prisoner, then you're not going to be able to calm the situation down. That would be my indicator to start recording. (05) Some officers explained that they tend to activate their cameras before delivering upsetting news or when they are about to make a potentially unwelcome request of a prisoner. As one officer explained, they would activate their BWC when entering "a potentially risky situation, knowing that, you know, you're giving [the prisoner] bad news or you're, you know, that you're going to have to demand something of a prisoner that you're fairly sure that he won't want to do" (02). Some officers also mentioned using the cameras outside of incident responses (for example, during interviewing of prisoners). In these cases, officers drew on their knowledge of a prisoner's past behaviour (and tendency to lodge vexatious complaints) in determining the appropriateness of activating their camera in advance of that prisoner interaction.

Recording interactions not permitted by the DCI
Officers also discussed how they might approach the use of their BWC in circumstances where recording would be in direct contravention of the DCI. The DCI instructs officers not to use their BWCs in specific areas of the prison or during certain procedures. For example, it states that recordings should not take place in prison locations where there is a "reasonable expectation of privacy" (Queensland Government 2017, p. 3). Most notably, the DCI makes clear that "a BWC recording must not be made of a search of a prisoner requiring the removal of clothing" (Queensland Government 2017, p. 3).
Officers generally acknowledged that they did not have unfettered discretion to record using their BWCs in all circumstances. Most often, they referred to ROC searches and, less often, to locations including the staff lunchroom or medical rooms, or circumstances involving private conversations. Importantly, though, while many officers recognized there were circumstances outlined in the policy specifying that they should not activate their BWC, some felt that they would rely on their intuition and discretion to activate their camera to record in an unsafe situation-regardless of the policy: I know you're not meant to have it on while they're doing medical consults and whatnot, but I think I would still have it on if I was going into a potentially dangerous situation. I know that the Deputy Commissioner's put out that instruction .  (02) Further, while officers said they were aware that the DCI prohibited the filming of prisoners during ROC searches, some officers nonetheless said they would activate their BWC during these searches. In many officers' view, this policy contravention was justifiable because safety considerations trumped a prisoner's right to privacy. However, to counteract privacy concerns, officers said they faced their BWC away from the prisoner: Well, I know within our guidelines for the use of body-worn cameras, we're not supposed to take body-worn camera footage of prisoners when we're doing removal of clothing searches. Because we've got to think about the dignity of the prisoner when they're doing these planned events… I have left my body-worn camera rolling even though we're doing a removal of clothing search…The staff's safety and my safety and the recording of the incident, in my opinion, overrides the privacy concerns of the prisoner. However, what I do when I do this, because I don't turn my body-worn camera off... because it also shows continuity of the footage, as opposed to turning it on and off, which is no continuity. I just turn my body to the side so the camera's not directly pointing at the prisoner being ground-stabilized and with the ROC being conducted, but the audio is still picking up. (03) By adopting this approach, officers felt they were able to successfully navigate any prisoner privacy concerns while maintaining the safety and security of officers.

Discussion and conclusion
BWCs are increasingly being introduced as a tool to make prisons safer-for correctional officers and prisoners alike (Polley and Smith 2020). However, the utility of this technology depends to a large extent on both the amount of discretion provided to BWC users and the way in which that discretion is exercised (Newell and Greidanus 2017). Thus, this study set out to better understand correctional officers' decisions about using their BWCs while working with prisoner populations.
Perhaps not surprisingly, our findings showed there were several scenarios where officers were almost unanimous in their support for activating their cameras-for example, situations including riots or where prisoners were threatening or being violent towards correctional officers. There were also, however, scenarios where officers' views were more mixed. Notably, circumstances involving potential privacy concerns (such as recording in the vicinity of a prisoner's medical consultation) or where it was a correctional officer who was threatening a prisoner revealed some disparity in officers' views on the appropriateness of recording.
Concerning the scenarios involving privacy considerations (namely, officers' willingness to activate the BWC in circumstances involving a prisoner's medical consultation), our results suggest that correctional officers who support the use of BWCs in prisons and believe the cameras protect their professional safety were more likely to abide by the policies governing the use of BWCs (by not recording in the vicinity of a prisoner's medical consultation but activating their camera to capture a fight in a medical unit). Conversely, officers who believe more strongly that BWCs impact prisoner interactions (by improving officer behaviour) are more likely to activate their BWC in both medical scenarios, thereby violating the DCI requirements around privacy. Finally, officers who support the idea that BWCs should be activated for all prisoner interactions were most likely to activate their camera in one or both medical scenarios, as opposed to not recording at all.
Next, we considered scenarios involving threatening behaviour with threats directed at both prisoners and officers. We note that while the DCI does not explicitly cover threatening behaviours, correctional officers who activate their BWC in both situations (prisoner and officer threats) would likely conform with the policy's intention by prioritizing both officer and prisoner safety. Our results indicate that officers who are less likely to value the fair treatment of prisoners were more likely to activate their BWC only when a prisoner was threatening an officer (compared to that scenario and one where an officer is threatening a prisoner). Additionally, officers who were less likely to perceive BWCs as a tool for improving their professional safety or to believe BWCs improve officers' interactions with prisoners were less likely to activate their BWC in either threat scenario compared to both scenarios. Lastly, officers who were more likely to feel BWCs improve prisoner-officer interactions were more likely to activate their camera in the prisoner threat only scenario (compared to neither threat).
Our findings also provide evidence of officers' attempts to exercise their discretion in the form of resistance to the relevant policy guidelines, with some officers revealing that they knowingly turn their BWC on in circumstances not permitted under the relevant policy. Specifically, officers said they would activate their BWC during searches that required the removal of a prisoner's clothing-despite the policy expressly forbidding such a recording from being made. To justify that decision, officers explained they had weighed the competing considerations, but had chosen to prioritize officers' safety over prisoners' right to privacy. They also pointed to the perceived need for continuity of footage; a consideration likely linked to officers' perceptions that such footage would protect them from any subsequent allegations of misconduct.
Together, these findings show that while having a BWC policy may, to some extent, influence officers' activation decisions, it does not guarantee their compliance with the policy. This may suggest the need for increased consultation with officers, as one party greatly impacted by the introduction of BWCs, during the development of BWC policies. More importantly, though, our findings highlight the difficulties that may be faced by correctional officers when it comes to using BWCs in prisons; difficulties caused by officers' responsibility for having to balance privacy and other ethical concerns relating to prisoners against considerations of the safety and security of the prison.
There is an "amplifying debate" in the US about the potential for BWCs to erode people's right to privacy, particularly society's most vulnerable, including minorities and people in prison (Murphy and Estcourt 2020, p. 368, see also Ringrose 2019). The idea that people in prison may be entitled to any element of privacy is not without its controversy. The deprivation of a prisoner's privacy has long been considered to simply represent one of the 'pains of imprisonment' (Sykes 2007) or a "functional prerequisite" of correctional institutions (Schwartz 1972, p. 229). More recently, Moran et al. (2013) went so far as to question whether the idea of privacy or private space is even possible in prison.
Despite this, researchers have begun to question whether the use of more intensive surveillance tools like BWCs unreasonably infringes upon individuals' right to privacy in carceral spaces (Taylor and Lee 2019), suggesting it would be remiss of corrective services agencies not to consider this issue when introducing this technology within prisons . The introduction of BWCs may be particularly problematic in jurisdictions that experience Indigenous overrepresentation in prison, given the disproportionate impact on these individuals when more intensive state surveillance is introduced (Murphy and Estcourt 2020). Further, Murphy and Estcourt (2020, p. 373) argue that "in other fields where individual privacy is significantly curtailed by the government, we normally require the matter to be dealt with by legislation, not internal policy documents". Despite this, corrective service agencies, like police before them, have largely been allowed to formulate their own BWC policies, "with little legislative oversight or democratic input" (Murphy and Estcourt 2020, p. 373). The consequence as Adams and Mastracci (2017, p. 313) point out is that the significant privacy concerns relating to the use of this technology are "rarely anticipated". These are matters that must be addressed should the use of BWCs continue, particularly in places that house vulnerable populations.