Better safe than sorry? Police officers’ identification of and responses to vulnerable crime victims

ABSTRACT

This study addresses police officers’ identification of vulnerable victims and responses to these victims. Using an Articulated Thoughts in Simulated Situations (ATSS) approach, police officers (n = 51) responded to two scenarios regarding a victim who reported either a property crime or a stalking incident. After hearing each scenario, participants judged whether the victim was vulnerable or not and discussed which they would propose any protection measures. Findings showed that a majority of the participants identified the victim as vulnerable. Participants mainly based their judgements on indicators related to the incident or its consequences. Regardless of their judgement, participants proposed a variety of protection measures. Both this finding and the majority of vulnerability judgements suggest that police officers may act out of the principle of ‘better safe than sorry’.

KEYWORDS:

• Vulnerability
• police officers
• victimisation
• assessments

Introduction

The EU Victims’ Directive (2012/29/EU) requires EU member states to perform an individual assessment to identify whether a victim is vulnerable and, therefore, in need of protection measures. Ultimately, the goal of the individual assessment is to identify the victims who are most in need of protection. In most member states, the police are tasked with performing these assessments. Despite the growing attention to victims’ vulnerability in the area of policing (Manikis 2019, Menichelli 2019), there is a lack of conceptual clarity and robust systems for the identification of vulnerability (Enang et al. 2019, Skidmore et al. 2020). Vulnerability is a complex and highly heterogeneous concept with different meanings (Brown et al. 2017). With regard to victimisation by crime, the most prominent meanings relate to the (future) risk of (re)victimisation and the impact or harm of the experience of victimisation (Walklate 2011, Keay and Kirby 2017, Skidmore et al. 2020). Whether a crime victim is vulnerable depends on a variety of factors that may be used to identify the vulnerability. These indicators could be related to the victims’ individual characteristics, the specific incident or the context in which the victimisation takes place (Innes and Innes 2013). In practice, police officers often use these indicators to assess whether a crime has a high impact on a victim or, as is the case with the individual assessment, elevates the risk of re-victimisation.

The complexity of vulnerability poses a significant challenge to police practitioners in correctly distinguishing highly vulnerable victims from less vulnerable victims and subsequently adequately assigning protective measures. The lack of conceptual clarity and robust identification instruments trigger risks of over- or under-protection. On the one hand, too few victims are correctly identified, resulting in increased harm due to high impact or re-victimisation as these victims do not receive the protection measures they need (Aihio et al. 2017, Enang et al. 2019, Franklin et al. 2020), or too many victims receive protective measures, resulting in a waste of resources and possible negative side-effects, such as stigmatisation and patronisation (Walklate 2011, Peroni and Timmer 2013).

In light of these concerns, this study aims to analyse Dutch police officers’ understanding of victim vulnerability as well as their proposed responses in terms of protection measures. We use an Articulated Thoughts in Simulated Situations approach (ATSS; Davison et al. 1983, Zanov and Davison 2010) to study police officers’ reactions towards two scenarios in which a victim reports either a property crime or stalking incident. Each scenario includes four indicators that are often associated with victim vulnerability in academic literature: risky lifestyles (Pratt and Turanovic 2016), the type of crime (Innes and Innes 2013), victims’ responses to the crime in terms of fear and resilience (Walklate 2011, Maglione 2017) and prior police contacts as victim/witness or suspect (e.g. Grubb and Bouffard 2015, Clay-Warner et al. 2016, DeCamp et al. 2018). In this context, fear may indicate vulnerability due to higher perceived negative impact and lower self-efficacy, while resilience may point to low vulnerability and a higher susceptibility to self-regulated actions in overcoming the victimisation incident (Walklate 2011, Maglione 2017). By asking police officers to explain whether or not the victim in the scenarios was vulnerable, this study aimed to determine which indicators police officers use in the identification of vulnerability. Furthermore, this study strived to provide insight into which protection measures police officers would propose after their identification of vulnerability.

Identifying victim vulnerability

Several theories have been developed as to what constitutes a vulnerable person (Verza 2022). While some suggest that every individual is vulnerable (Fineman 2008, Verza 2022), others argue that vulnerability may refer to specific individuals or groups, based on certain characteristics (Peroni and Timmer 2013, Van der Aa 2016, Enang et al. 2019, Skidmore et al. 2020, Verza 2022). The latter approach applies to vulnerability assessments in which governmental institutions, such as the police, are tasked with identifying vulnerable individuals in need of support or protection (Menichelli 2019). Eventually, these assessments aim to prevent repeat victimisation or secondary victimisation and also contribute to a better allocation of limited resources.

Vulnerability assessments, however, are challenged by a lack of robust systems to identify vulnerable victims and confusion amongst police officers about what vulnerability means (Keay and Kirby 2017, Enang et al. 2019, Skidmore et al. 2020). Limited understanding of vulnerability may result in incomplete judgements and a risk of over- or under-protection (Gilson 2016, Van der Aa 2016). It is, therefore, important to improve our understanding of how police officers identify vulnerability. Previous studies analysed survey data, criminal records or vignettes of victimisation incidents. These studies demonstrated that police officers mainly based their assessments on indicators closely related to the victimisation incident, such as the threat of a weapon, injuries or the presence of children on the crime scene (Robinson et al. 2016, 2018, Turner et al. 2019, Petersson and Thunberg 2021). Yet, studies also showed that police officers based vulnerability assessments on characteristics of the victim, such as risky behaviour, age and mental health problems (Bach et al. 2021, Enang et al. 2021). While these studies provide some insights into how police officers understand victim vulnerability, the reasoning behind these identifications remains largely unclear.

Providing protection measures

Victims in need of support often refer to safety, protection from the suspect and support from the criminal justice system (Ten Boom and Kuijpers 2012, Henning 2016, Villacampa and Salat 2019). Vulnerability assessments serve as an instrument for police officers to identify such needs, formulate protection measures and thereby help to better protect them against future harm (Enang et al. 2021). The police may initiate protection measures, such as temporary restraining orders or extra alerts on high-risk victims in the police systems. Vulnerable victims may also be referred to victim support services, health professionals or multi-agency safeguarding hubs, which can also initiate support measures (Robbins et al. 2014, Shorrock et al. 2019, Enang et al. 2021). Measures initiated by police agencies or by multi-agency safeguarding hubs are often part of risk management strategies (Storey et al. 2014). In practice, these strategies are indeed offered more often to victims at risk of repeat victimisation (Belfrage et al. 2012, Storey et al. 2014). However, police officers may still be unaware of specific indicators of vulnerability when they are developing risk management stratgies (Storey and Strand 2017).

To better understand how police officers understand vulnerability and protect vulnerable victims, the current study addresses both steps of vulnerability assessments: the identification of vulnerability and the protection measures suggested for vulnerable victims. While previous studies address both steps, they rarely combine them and address police officers’s reasoning behind decisions in both steps. The current study, therefore, expands prior literature by using an ATSS approach to analyse Dutch police officers’ understanding of victim vulnerability as well as their proposed responses in terms of protection measures. Although the insights of this study apply to the Dutch context, they may also hold implications for EU member states where the police perform the individual assessment.

Methods

Data was collected in 2021 at four police stations geographically spread across the Netherlands. Police officers were eligible for participation if they had been trained to record victims’ crime reports. These police officers perform the Individual Assessment on a daily basis. They judge whether victims are vulnerable and in need of protection measures in order to identify those who are most in need of protection. To perform the Individual Assessment, Dutch police officers follow a policy guideline containing several indicators for vulnerability. These indicators refer to characteristics of the victim such as age; a list of high-risk types of crimes containing, for instance stalking, sexual assault and domestic violence; the circumstances in which the type of crime is committed; considerable harm resulting from the severity of the crime; crimes with a bias motive and crimes in which the victim has a dependency relation with the suspect. If at least one of the indicators in the guideline refers to the victim, the victim can be considered as vulnerable.

In total, 51 police officers participated in the study (M age = 41.67, SD = 12.07, range 22–65, 60.8% male, 39.2% female). Police officers had different functions: front-desk officers (n = 17), community officers (n = 8) and regular police officers (n = 26) with high variability in their years of experience in the police force (M = 14.60, SD = 10.05, range = 1–42). For this study, a data management plan and GDPR form were approved by the Ethics Review Board of the Tilburg Law School. Before the procedures were explained, participants received an informed consent form stating the goal and duration of the study guarantees that the data was stored and used anonymously, how long the data would be stored, who had access to the data and contact details in case they had questions after the study. Participants signed the form when they agreed to participate in the study. After this, the procedures were explained.

Procedure and design

The study followed the procedure of the ATSS design (Davison et al. 1983, Zanov and Davison 2010). In this procedure, participants receive an audio vignette that closely resembles a daily practical situation and are instructed to freely articulate their thoughts out loud. The vignette pauses several times, during which participants articulate their thoughts. After a predetermined break, the vignette promptly continues. This process is repeated until the vignette ends.

The freedom to articulate every thought they have allows participants unrestricted answering possibilities in contrast to fixed questionnaires (Davison et al. 1995). This is especially beneficial to study reactions to complex and sensitive topics (Rayburn and Davison 2002) and responses to victimisation (Bosma 2019, Mulder and Olsohn 2020). Additionally, the scenarios are designed and pre-tested to closely resemble practical and daily situations (Davison et al. 1983), which makes the procedure especially appropriate to study both reactions and professional judgements of participants, such as criminal justice personnel (Bosma 2019, 2023). ATSS is thus particularly appropriate to study police officers’ understanding of a heterogeneous and complex concept as vulnerability and their responses towards vulnerable victims.

Participants were carefully instructed about the procedures and then individually seated in a room without the experimenter present. Participants listened to two audio-fragments in which a victim reported either a property crime or a stalking incident. To enhance ecological validity, scenarios were pre-tested to identify possible discrepancies to everyday police practice (Bosma 2023). The procedure and the audio-fragments were, therefore, pretested and adjusted based on a small sample of police officers (n = 3) who did not participate in the final sample. Each audio-fragment automatically paused for 30 seconds and then resumed 3 times, dividing the vignette into 4 different blocks. During each pause, participants were requested to articulate their thoughts out loud. In the instruction, we emphasised that participants were free to articulate each and every thought they had. After each scenario, participants received several open questions and also answered out loud within a timeframe of one minute. Due to Covid-restrictions, these questions were also pre-recorded. The open questions after each scenario were the same for each vignette and addressed (1) participants’ understanding of the scenario, (2) the identification of vulnerability and (3) proposed protection measures.

The total procedure lasted approximately 18 min. One laptop played the audio fragments, while a second laptop made audio recordings of the procedure. These recordings were transcribed verbatim.

Materials

Vignettes

Each participant randomly received two audio-vignettes containing a fictional male victim reporting a property crime (vignette 1: PC) and then a female victim reporting a stalking incident (vignette 2: ST). The vignette was divided into four different blocks. Each block represented a possible indicator of vulnerability: risky lifestyle, type of crime, consequences and prior police contacts (see Tables 1 and 2 for an example of the scenario and see Supplementary Materials for the full texts). There were two versions per block: high vulnerability or low vulnerability. The high vulnerability version of each block was risky lifestyle, stalking, fear of crime and prior police contacts either as suspect or as victim/witness. The low vulnerability versions of each block included low-risk lifestyle, property crime, resilience and no prior police contacts. The age of the victim differs with the risky lifestyle. The PC victim is 25 years old (risky lifestyle) or 38 (less risky lifestyle), while the ST victim is 23 years (risky lifestyle) or 34 years old (less risky lifestyle). Vulnerability, in terms of risk of repeat victimisation, is negatively associated with age (Averdijk 2014).

Table 1. Example of a property crime scenario.

Vignette block	Text
Block 1: Less risky lifestyle	‘Yesterday, I cycled home from work. My nightshift in the retirement home ended around 10 p.m.. I was very tired, so I really wanted to go home. The largest part of my way home crosses a residential area near the retirement home. That’s is the quickest route, because you take a shortcut from the main roads. Near the edge of the residential neighbourhood, there is quiet parc. It happened near the playground in the parc’.
Block 2: Type of crime, property crime	‘Out of the blue, a scooter came driving straight towards me. Luckily, I could stop right in time. The driver got off the scooter and approached me. He told me to give him my wallet and phone. I could not see clearly what he looked like, but he sounded young. He stood right in front of me and said that I could better give him the stuff, otherwise I would have a big problem. I thought that he had a knife, at least I saw something shiny. I felt threatened so I did what he wanted. Who knows what he was capable of? After I gave him my wallet and phone, he ran towards his scooter and drove away. It all happened so fast, I did not see the licence plate’.
Block 3: consequences (fear)	‘After he left, I quickly cycled back home. I didn’t sleep all night. This morning, I struggled to keep my food down. I don’t know if I still dare to go outside alone at night. Maybe, I can better drive together with someone from now on. In any case, I will avoid that parc. I hope that you can catch the perpetrator, so I dare to walk the streets again’.
Block 4: Prior police contacts as victim witness	‘In the police registration system, the victim has the following police contacts’: “Incident: Assault; Role: Witness; Date: 04-12-2019” “Incident: Insult; Role: Victim; Date: 16-09-2019”.

Table 2. Example of a stalking scenario.

Vignette block	Text
Block 1: Less risky lifestyle	‘Yesterday, just after ten, I walked back to my car. I just went to the supermarket. Since I have a busy job, I often have to do the grocery shopping in the evening. At that time, it is fairly quiet in the supermarket. There were only a few cars left in the parking lot. As for the rest, there was no one in sight. I was standing close to my car looking for my car keys when I suddenly heard someone behind me’.
Block 2: Type of crime, stalking	‘I immediately recognized the voice: it was my sister’s friend. He sounded angry and asked why I did not respond to his texts. We had been on a few dates. That was fun, but I quickly closed it off. He, however, wanted more. For a couple of weeks, he has been sending me texts and voicemails. He is often looking for me at [the supermarket]. I have always managed to avoid him, until [yesterday]. I said that he should really stop this and that I would go to the police otherwise. This only made him more angry. He stood right in front of me, I thought that he was going to hit me. All of a sudden, two people walked past and stopped near us. He said that we would meet again some other time and walked away. After this, I quickly drove home’.
Block 3: consequences (resilience)	‘Once I was home, I was very upset, but I’m not going to hide in my home. This is the last time that he bothers me like this. This morning I enrolled in a self-defence course, so I can protect myself better in the future. From now on, I can solve it myself. Next time, I’m not going to let me feel threatened like that anymore. I hope you can arrest him, so he will leave me alone from now on and I can move on’.
Block 4: Prior police contacts as victim witness	‘In the police registration system, the victim has the following police contacts’: “Incident: Insult; Role: Victim; Date: 11-08-2019” “Incident: Fight/Brawl; Role: Witness; Date: 09-04-2019”.

In the first block (risky lifestyle), the victim explained how they arrived at the place of the incident – a place they often find themselves in. The victim either returns from work late in the evening (low-risk lifestyle) or in a drunken state returns from a party they often attend (risky lifestyle). In the second block (type of crime), the victim describes the crime, i.e. the property crime or the stalking. These two types of crimes were chosen as they may hold serious consequences for the victim. In this study, we chose property crime and stalking because property crimes are not included in the list of crimes that are indicators of vulnerability according the policy guideline for the Individual Assessment in the Netherlands. Yet, stalking is included in this guideline. The inclusion of both types of crime, therefore, enabled a comparison between judgments of one type of crime which is included in the guideline and another type of crime which was not included in the guideline but may also refer to vulnerability. In the third block (consequences), the victim elaborates the short-term consequences of the incident, either in terms of displaying fear and avoidance behaviour or in terms of displaying resilience: taking steps to cope with the aftermath. The last block (prior police contacts) presented (1) no prior history (equally distributed), (2) two prior contacts as victim/witness (only if block 2 is low-risk lifestyle) or (3) two prior contacts as suspect regarding driving under the influence or disturbing public order (only if block 3 is high-risk lifestyle). Block 4 thus adds to the manipulation of high- vs. low-risk lifestyle.

Eight versions of vignettes 1 and 2 were constructed. These versions varied in the lifestyles, consequences and the prior police contacts. Table 3 shows the number of participants that received each variation of a block and which variations were either low or high vulnerability. Note that while some numbers are equal, these do not necessarily consist of the same participants.

Table 3. Population per factor and vignette.

	Vignette 1	Vignette 2
Lifestyle
Less risky (low)	25	25
Risky (high)	26	26
Type of crime
Property crime (low)	51
Stalking (high)		51
Consequences
Fear (high)	26	25
Resilience (low)	25	26
History of police contacts
No history (low)	25	25
History of victim/witness contacts (high)	14	14
History of suspect contacts (high)	12	12
n	51	51

To ensure that participants received two distinct scenarios, several details in the narrative were presented differently between the two types of crime. For example, the location and the specific activity differed between the PC- and ST-victim (block 1) and the display of resilience (block 3) differs: the PC-victim continued to travel across the residential area where the incident took place and would like to confront the offender, while The ST-victim signed up for a self-defence course in order to personally handle the situation, if it were to occur again (near the parking lot of the supermarket).

Analytic strategy

Coding of the transcripts took place in two rounds and on the basis of a structured thematic coding scheme presented in Table 4. In the first round, the corresponding author coded all data and the second author coded a subsample, after which they compared codes and solved any disagreements. Several new codes that emerged from the discussion were added to the coding scheme. In the second round, the corresponding author closely checked and reviewed the assigned codes.

Table 4. Coding scheme.

Subject	Description
Vulnerability
Meaning
Impact	The direct harm caused by the incident.
Risk of (repeat) victimisation	An elevated risk, either observed or actual, of becoming a victim or being revictimised.
Indicators
Risky lifestyle	Frequent behaviour/activity displayed before the crime took place, i.e. returning home late from work or from a party.
Type of crime	The crime itself or details regarding the crime
Consequences	The state of the victim after the incident, either emotional or in terms of behaviour
History of police contacts	Either the presence or absence of prior contacts registered in police databases in the role of either victim, witness or suspect
Individual characteristics of the victim	e.g. the gender or age of the victim
Missing indicators	Participants who openly argued that information was missing.
Role police officers
Judicial	Remarks or measures referring to finding the suspect or proving the crime.
Victim support	Remarks or measures referring to support the victim emotionally or psychologically.
Protection measures
Physical protection measures	Measure where high-urgency cases are highlighted in police systems
Internal referral	Referrals within the police organisations, e.g. to a community officer
External referral	Referrals to other organisation than the police, e.g. to Victim Support Netherlands
Other protection measures*	Protection measures surfaced during the first coding round, referring to, for instance, the stop conversation or the SASH-form.
Victim measures*	Measures that the victim should take themselves

* Newley added codes after the first round of coding.

Based on the answers on the second question after each scenario, three categories were distinguished regarding participants’ identification: ‘vulnerable’ (V), ‘not vulnerable’ (NV) and ‘unsure’ (U). Reflecting the dichotomous nature of the European Individual Assessment as implemented in the Netherlands, the second question was a yes or no question. V- and NV-judgements could be distinguished when participants stated: ‘I think the victim is [or is not] vulnerable, because … ’. When participants considered both options without settling for either or were unsure about the answer, they were categorised in ‘unsure’. Vulnerability assessments, in terms of meanings and indicator, and protection measures proposed by participants were compared between the three vulnerability-categories and the two types of crime.

All quotes presented in this article have been labelled according to participant number, the scenario (PC: property crime; ST: stalking) and the vulnerability judgement (V: vulnerable; NV: not vulnerable; U: unsure). The percentages of the codes presented in the results section were based on the total number of participants per judgement in a scenario (as can be seen in total-row and columns of Table 5). If, for example, code A is mentioned by 10% for PC-V, this means that 10% of the participants who identified the property crime victim as vulnerable mentioned code A in their reasoning. An overview of the percentages and frequencies mentioned in the results section can be found in Table 3s and Table 4s in the Supplementary Materials.

Table 5. Number of high vulnerability indicators per judgement.

	1		2		3		4
	N	%	N	%	N	%	N	%	Total N
Property crime
Vulnerable	1	20.0%	8	40.0%	14	70.0%	4	66.7%	27
Not vulnerable	2	40.0%	8	40.0%	4	20/0%	1	16.7%	15
Unsure	2	40.0%	4	20.0%	2	10.0%	1	16.7%	9
Total	5	100%	20	100%	20	100%	6	100%	51
Stalking
Vulnerable	4	66.7%	15	78.9%	13	65.0%	5	83.3%	37
Not vulnerable	0	0.0%	2	10.5%	3	15.0%	0	0.0%	5
Unsure	2	33.3%	2	10.5%	4	20.0%	1	16.7%	9
Total	6	100%	19	100%	20	100%	6	100%	51

Results

Vulnerability judgements

Overall judgements

Table 5 shows that, in both scenarios, a majority of the participants judged the victim as vulnerable. For the PC-scenario, participants identified the victim mostly as vulnerable when there were more high vulnerability indicators present. The number of high vulnerability indicators did not influence judgements for the ST-scenario, as a large number of participants identified the victim as vulnerable across all scenarios. The majority of vulnerability judgements may indicate that police officers recognise the indicators for vulnerability, especially with respect to the PC-scenario. The ST-scenario, however, shows that the number of indicators did not influence the distribution of judgements. It may be possible that stalking in its own right is a signal for vulnerability, but it may also suggest that they assess vulnerability to be better safe than sorry.

For both scenarios, participants referred to impact (PC-V: 55.6%; ST-V: 35.1%) and the risk of (repeat) victimisation (PC-V: 37.0%; ST-V: 40.5%) as meanings of vulnerability. This corresponds with other studies where police officers are asked to identify victim vulnerability (Robinson et al. 2018, Bach et al. 2021). Especially for the fear-scenarios, the impact is often related to the emotional and possible traumatic effects of the incident. In the ST-scenario, participants argued that the incident had interrupted the victim’s private life: ‘she cannot lead her normal life, so I regard her as [a] vulnerable person’ [P1-ST-V]. Some participants also specifically noted the incident had a high impact on the victim. Participants mostly associated the risk of (repeat) victimisation to the victim’s perceived risks of future crimes. Especially in the scenarios with a fearful victim, participants argued that the victims view the risk of revictimization as high and thus feel a fear of crime: ‘he is also afraid to become victimised again, so yes I see him as vulnerable’ [P46-PC-V].

Indicators of vulnerability

Consequences

To identify vulnerability, participants most often used the consequences for the victim that manifests themselves shortly after the incident as an indicator for vulnerability (PC-V: 70.4%; ST-V: 59.5%). In the fear scenarios of both crimes, participants argued that anxiety and the subsequent behavioural change, i.e. avoidance of going outside, was important reasons to view the victim as vulnerable: ‘I think this guy would classify as vulnerable, in any case he said that he became more afraid, that he also didn’t dare to go outside alone anymore’ [P43-PC-V]; ‘I notice that it affects her a lot, that she locks herself up in her own house, closing the shutters, troubles with sleeping, thinking that he might be at the door [..] so I surely think she’s vulnerable’ [P44-ST-V]. The association between victimisation, fear and avoidance behaviour has been demonstrated in previous studies (Archer 2019, Randa et al. 2019). Yet, it has rarely been referred to in the context of identifying vulnerability, although fear of crime has been mentioned in prior studies as a possible indicator for vulnerability (Jackson 2009, Maglione 2017). In their identification, participants in this study discuss how avoidance behaviour overlaps with impact and the risk of repeat victimisation, i.e. the two meanings of vulnerability. The choice to avoid going outside can be associated with the way in which the incident affected the victim emotionally (impact). Furthermore, avoidance behaviour can be the result of the fear the victim feels and the subsequent perceived high risk of new victimisation (risk of repeat victimisation). Fear is an overlapping factor and the display of avoidance behaviour is the concrete manifestation of the emotional effects of the incident and the elevated perceived risk of victimisation.

The frequent mentioning of avoidance behaviour also suggests that police officers view behavioural change rather than fear itself, as a signal of vulnerability. Responses to the resilience-scenarios seemed to enforce this proposition as participants viewed both depictions of resilience in different lights. On the one hand, they argued that the PC victim was not vulnerable, as the victim continued with their behaviour: ‘I would not view him as vulnerable, simply because he states that this incident won’t get to him. Next time around, he will just ride through the industrial area again.’ [P8-PC-NV]. On the other hand, several participants deemed the resilient stalking victim as vulnerable, as they argued that the self-defence course was an indicator that the victim currently was vulnerable: ‘Of course she herself already states that she wants to become verbally stronger through the self-defence course, this means that she is vulnerable’ [P36-ST-V]. These views suggest that behavioural changes after an incident, regardless of whether negative (avoidance) or positive (moving towards resilience), function as a signal for vulnerability, as resilience is not yet reached.

Type of crime

A smaller proportion of the participants in both scenarios note the type of crime as indicators of vulnerability (PC-V: 37.0%; ST-V: 35.1%). Corresponding with previous research (e.g. Robinson et al. 2018, Turner et al. 2019), participants mentioned specific characteristics closely related to the incident or the type of crime. The specific information relating to the type of crime, however, differs between the two scenarios. For the PC-scenario, participants mentioned that the threat of violence, and potential presence of a weapon, seriously affected the victim: ‘I think he’s vulnerable, firstly, because he felt threatened during the incident’ [P19-PC-V]. Stalking in its own right seemed to be a signal for vulnerability in the ST-scenario, but several participants also mentioned the suspect’s behaviour and an elevated risk of subsequent or continuing incidents: ‘Yes, she’s vulnerable, because the man’s behaviour has not yet stopped’ [P6-ST-V]; ‘For me she’s vulnerable, because, she told him to stop and to leave her alone, but he doesn’t do this’ [P47-ST-V].

In their articulated thoughts, participants also tended to assess the seriousness of the incident in addition to the identification of vulnerability. Some participants argued that the scenario concerned an urgent stalking case: ‘definitely a form of stalking, this is a very serious crime’ [P3-ST-V], while others regarded it as early-onset stalking or viewed it as the lesser crime of unlawfully threatening: ‘apparently she implies that she is being threatened by him’ [P4-ST-V]. Additionally, participants had concerns about the legal threshold of evidence: ‘How can we prove a stalking case? there is no threat present, according to criminal law, I think. So, how many times she has been bothered and if she can prove this.’ [P2-ST-V]; ‘At this moment, there is little room for criminal liability’ [P10-ST-V].

Indicators left out of the judgement

Participants used risky lifestyles (PC-V: 18.5%; ST-V: 5.4%), victim characteristics (PC-V: 3.7%; ST-V: 24.3%) and the victim’s police contact history (PC-V: 14.8%; ST-V: 5.4%) to a lesser extent than the indicators mentioned above. This is surprising since prior studies argued that these indicators may be important for identifying vulnerability (e.g. Averdijk 2014, Clay-Warner et al. 2016, Pratt and Turanovic 2016, Savard et al. 2017, DeCamp et al. 2018, Bach et al. 2021). Although these indicators were mentioned to a lesser extent than expected based on the literature, there are still considerable differences in the percentages between both scenarios. Participants in the PC-scenario mentioned risky lifestyles with regard to frequent drinking behaviour and being in abandoned places: ‘I think that he’s vulnerable [..] you’re not how you’re supposed to be due to the alcohol consumption [..], he is cycling across that industrial area, with not a soul in sight’ [P27-PC-V]. Participants only mentioned the victim’s gender in relation to the female victim in the ST-scenario: ‘Yes, I think she’s surely vulnerable [..] as a woman you’re slightly more intimidated by a man than vice versa’ [P2-ST-V]. This may not be surprising as stalking is strongly associated with gender-based violence, whereas property crimes are not. Interestingly, participants rarely mentioned the other victim characteristic in the scenario: age. Across the three judgements, some participants mentioned indicators that were not included in the assessments but could have some relevance for the identification of vulnerability. These indicators could refer to, for instance, a victim’s mental health, disabilities or non-verbal cues such as a victim’s body language: ‘In these situations, I always prefer a face-to-face setting so you can see in his attitude and especially the eyes how it affects a person. You can tell a lot by non-verbal communication.’ [P28-PC-NV]. Although these indicators may be absent in the scenarios, studies argue that a victim’s mental health and nonverbal cues may be of importance when identifying vulnerability (e.g. Daigle and Teasdale 2018, Dinkins and Jones 2018).

Participants rarely associated a victim’s prior police contacts with vulnerability. In their articulated thoughts, some participants argued that either the first contact with the police or prior victim/witness contact with the police may influence how the incident affects a victim. A small number of participants also wondered whether the prior victim/witness contacts involved the same offender as the current incident and discussed that the victim did not report any earlier stalking incidents. Other participants, however, argued that they did not see the relevance of these prior contacts: ‘It tells me very little. Actually, I don’t find this relevant. She was a victim once before, but that doesn’t mean very much’ [P46-ST-V]. Additionally, some participants implied that prior police contacts mainly matter for offenders: ‘it doesn’t tell you a lot when he doesn’t appear [in the police systems], he is a victim and not an offender.’ [P50-PC-V]; ‘Looks like a decent hard-working man without a criminal record’ [P19-PC-V]. In responses to prior suspect contacts, participants argued that focussing on these contacts did not alter their view of the victim: ‘To me this would not mean that I would not take this man seriously, because this man too can become a victim of a crime’ [P29-PC-V]; ‘it is strange to judge someone based on this’ [P24-ST-V].

Protection measures

General measures

Similar to the assessment of the stalking incident, participants took on two roles both in their articulate thoughts and in their suggested response to the victim: a judicial role and a victim support role. For the judicial role, participants referred to gathering relevant information for criminal investigations, by asking for clarifications on details of the incident or documenting the crime (PC-V: 48.1%; ST-V: 70.3%). In their articulated thoughts of the PC-scenarios, participants also wanted to ask for a detailed description of the suspect or whether there was camera surveillance on the location of the crime. In their articulated thoughts for the ST-scenarios, the participants asked for details regarding the case, such as the content of the texts or whether there were witnesses.

The victim support role referred to supporting the victim’s well-being and safety. In both scenarios, a majority of the participants wanted to refer the victim to an external organisation (PC-V: 74.1%; ST-V: 59.5%), most notably to Victim Support Netherlands. Participants more often proposed external referrals in the PC-scenario than the ST-scenario. This may be due to internal police protocols for stalking cases. While external referrals are not protection measures in their own right, they play an important role in directing the victim to the adequate organisation for protection or support. Participants stated in both scenarios that the victim would benefit from speaking to victim support professionals, since they are specially trained and better equipped than the police to provide emotional and psychological support: ‘we as the police can only provide limited support of this kind, while victim support services may offer him more adequate support’ [P50-PC-V]. This resembles other European countries where victims are also referred to external services, such as multi-agency safeguarding hubs, for specific support measures, for specific support (Robbins et al. 2014, Shorrock et al. 2019).

Vulnerability seemed to be no condition for these referrals, since NV- and U-judgements also often mentioned external referrals in both scenarios (PC-NV: 73.3%; ST-NV: 80.0%; PC-U: 66.7%, ST-U: 66.7%). The arguments behind the referrals were also similar: ‘After documenting the crime report, I would refer him to victim support services (..) these people are professionals that can have a conversation with him and they deal with that better than we potentially can’ [P37-PC-U].

Similar to the external referrals, participants mentioned internal referrals for both V- and NV-judgements (PC-V: 22.2%; ST-V: 43.2%; PC-NV: 13.3%; ST-NV: 60.0%; PC-U: 33.3%; ST-U: 22.2%). Participants proposed internal referrals most in the ST scenario. Internal referrals serve to direct the victim to the adequate department or person within the police organisation who can provide specific protection measures. It is likely that internal referrals are more often proposed in the stalking scenario because police policy provides more options for specific protection measures provided by the police in case of stalking than in case of property crimes.

Yet, it is striking that police officers mention both external and internal referrals regardless of the judgment on vulnerability. This suggests that vulnerability is not conditional for these measures to be proposed: every victim can be referred to an external organisation. As one participant posed: ‘In any case, I would point out to her that she can always have a conversation with the people of victim support services’ [P29-ST-U]. While this may suggest the ambition to protect all victims from severe harm, it also poses several challenges for police workload as every victim is eligible for protection measures. Moreover, it poses questions on the function of the label of vulnerability as actions are not dependent on the identification of vulnerability.

Specific protection measures

In addition to the referrals, police officers also mention specific protection measures, especially in regard to the ST-scenario and regardless of the judgement (PC-V: 25.9%; ST-V: 70.3%; PC-NV: 33.3%, ST-NV: 40.0%; PC-U: 33.3%; ST-U: 55.6%). Specific measures in the ST-scenario mostly referred to the stop-conversation with the suspect, which is part of the police’s stalking protocol. This is a conversation between police and suspect, which discusses the severity of the suspect’s behaviour and aims to deter them from continuing their behaviour. Participants, however, differed in the urgency to advise a stop-conversation. Some participants stated this conversation ought to urgently take place: ‘that this man should at least get a stop-conversation to make clear that he should end these actions’ [P15-ST-V]. Others mentioned that it was a mere possibility: ‘In any case, I want to discuss with her if she would like us to have a stop-conversation with him, so it is really clear for him’ [P49-ST-V]. NV- and U-judgements also mentioned the stop-conversation, but also mentioned it as a mere possibility: ‘we are definitely going to have a conversation with that man and see how he experienced the situation and subsequently already have a sort of stop-conversation’ [P31-ST-U].

Some measures were only mentioned in case of a V-judgement in the ST-scenario. An example of this was physical protection measures (ST-V: 21.6%). This refers to measures where cases are highlighted in the police registration system to alert police officers that these cases are highly urgent and require extra attention. Similarly, V-judgements wanted to use the Screening Assessment for Stalking and Harassment form (SASH), a risk assessment instrument for assessing the risk of future incidents, and to determine the next steps, an officer would take: ‘a SASH-form can, in this case, also provide an indication on which follow-up actions should be taken’ [P1-ST-V]. Notably, several participants who identified the ST-victim as vulnerable advised that the victim should also take certain actions (ST-V: 37.8%). This advice mainly referred to keeping a journal of the stalking incidents, which may benefit later legal proceedings: ‘she has to collect and save all information: mails, emails, WhatsApp-messages, telephone-calls, voicemail-messages. This all has to be collected to eventually get a restraining order, for instance. That process takes long’ [P10-ST-V]. Participants took on their judicial role since the journal is a key part in the evidence building: ‘I think the crime report may be challenging, because it is not really [a case] according to criminal law (..) I would advise her to start keeping a journal, since this is necessary if we want to develop this in a report for stalking’ [P6-ST-V].

Conclusion and discussion

The identification and protection of vulnerable victims poses several complex challenges for daily police practice. Using an ATSS design, this study aimed to analyse police officers’ understanding of vulnerability as well as their response towards victims in terms of suggested protection measures. This study addressed both steps of vulnerability assessments and the police officers’ reasoning behind these steps. The findings provide more clarity on how police officers identify vulnerability in daily practice and why they propose protection measures based on this assessment. Hence, these findings may inform vulnerability-focused programmes in policing and help to enhance the protection of vulnerable victims. These insights apply to the Dutch context but may also hold implications for EU member states where the police perform the Individual Assessment.

In this study, a majority of the participants judged the victim as vulnerable. Furthermore, participants proposed protection measures regardless of the vulnerability judgement. On the one hand, this finding may suggest that police officers recognise indicators of vulnerability and furthermore strive to protect as many victims as possible. Alternatively, it may also be possible that participants identified vulnerability regardless of specific indicators. Police officers may fear that something goes horribly wrong and may act on this fear via the principle of ‘better safe than sorry’ (Richards et al. 2021). Participants may have identified victims as vulnerable or proposed protection measures to victims, out of fear for making incorrect decisions rather than judgement. If this reasoning is present amongst police officers, it may hold consequences for daily police workload since a large number of victims may be deemed vulnerable. Moreover, an even larger number of victims may receive protection measures, as these measures were proposed regardless of the judgement. By proposing protection measures regardless of the judgement, it appears that the measures are not necessarily related to the outcome of the assessment. The consequences then put more pressure to already limited police resources. More research is, however, needed to study the prevalence of the ‘better safe than sorry’ principle amongst police officers and to understand the psychological factors underlying this principle.

Interestingly, worries that victims may fall outside narrow definitions of vulnerability and thereby be excluded from receiving protection measures (Gilson 2016, Robinson et al. 2016, Van der Aa 2016, Enang et al. 2019) seem to be contradicted in the context of this study on the current Dutch police practice. However, if police officers propose measures regardless of the judgement, the vulnerability assessment loses its function of identifying those most in need, so that the distribution of limited resources is further pressured. To more thoroughly determine whether the assessment loses its function, more research is needed to study these effects on regional or national level.

The ‘better safe than sorry’ principle does not only hold consequences for the workload of police officers or the allocation of measures, but also impacts victims’ feelings of empowerment and resilience. The ambition to protect victims from harm or reduce their risk of repeat victimisation may culminate in overprotectiveness, which may damage victims rather than support them (Verza 2022). Being labelled as vulnerable may neglect a victim’s resilience and result in feeling stigmatised or patronised (Walklate 2011, Peroni and Timmer 2013). Victims’ wishes with regard to protection measures may be a secondary consideration for police officers, whereas their own judgement is primary (Meyer and Reeves 2021).

The findings discussed above present important challenges for police organisations in seeking the balance between the over and under protection of vulnerable victims. For police organisations, this is a cost–benefit analyses where the costs represent the available police resources and the benefits represent the number of victims who are protected from adversities. The fundamental idea is to protect victims who are most in need for protection. However, it is important that this group of victims is identified correctly. Therefore, it is necessary to aid police officers in the identification of vulnerability through trainings (Franklin et al. 2020) or risk assessment instruments based on indicators of vulnerability such as prior police contacts (ProVict: Geurts et al. 2021, Raaijmakers et al. 2022) or fear of crime (B-SAFER and DASH: Kebbell 2019). Moreover, these trainings should address the dual role police officers have in their interactions with victims, as demonstrated in the articulated thoughts. Traditionally, police work revolved around investigation processes and legal concerns, such as the availability of evidence and likelihood of prosecution (Johansen et al. 2022). Vulnerability assessments, however, introduce another aspect to police work: victim support and welfare (Rumney and McPhee 2023). Taking both these aspects into account demonstrates the complex nature of modern-day policing as it challenges police officers to position themselves (Saxton et al. 2020). To shape these trainings, future research should focus on getting a better grip of both police officers’ understanding of vulnerability and their views on the place vulnerability takes in their daily work.

Participants in this study mainly used the consequences of victimisation, in terms of fear of crime and avoidance behaviour, and indicators directly related to the incident as indicators for vulnerability. Regarding the consequences of the incident, the findings suggest that participants considered behavioural changes as an indicator of vulnerability. Regarding the indicators directly related to the incident, the findings resonate with earlier quantitative studies arguing that police officers often use this type of information in their risk assessment (e.g. Robinson et al. 2018, Turner et al. 2019). The articulated thoughts underlined these earlier findings, by demonstrating that participants took a judicial role in their approach towards the victim, thereby foccusing on the incident and its characteristics.

Yet, participants left out several indicators which, according to prior studies (e.g. Clay-Warner et al. 2016, Pratt and Turanovic 2016, DeCamp et al. 2018), are strongly associated with vulnerability, such as prior crime experiences or risky lifestyles. Discarding prior crime experiences as indicator of vulnerability, however, may influence how police officers use risk assessment instruments that use such experiences as indicators (see for instance Geurts et al. 2021, Raaijmakers et al. 2022), as police officers may not see the importance of prior crime experiences in the context of vulnerability assessments. While police officers may associate risky lifestyles with vulnerability (Bach et al. 2021), the display of such behaviour may influence other perceptions or biases of police officers, such as victim credibility (Sleath and Bull 2017). Differing perceptions may result in discrepancies between self-perceived vulnerability and vulnerability perceived by police officers (Aihio et al. 2017). From the victim’s perspective, factors such as socioeconomic status, gender and employment, and victimisation of certain types of crimes, such as hate crimes, may influence self-perceived vulnerability (Perry and Alvi 2012, Vieno et al. 2013, Aihio et al. 2017). Yet, police officers may not recognise indicators of vulnerability due to lack of understanding, perceptions or biases (Venema 2019, Franklin et al. 2020, St George et al. 2021). In conclusion, several indicators of vulnerability may currently fly under the radar while they are relevant for identifying and supporting vulnerable victims. In general, specific training on vulnerability indicators may aid police officers in understanding victims’ behaviour and protection needs (Franklin et al. 2020). Training police officers on the importance of prior crime experiences, individual indicators and risky lifestyles as indicators of vulnerability may aid them in obtaining a broader understanding of vulnerability. Moreover, these trainings may function as a platform for police officers to share their views on specific indicators, which can further improve how vulnerability assessments are designed and how they are implemented in police practice.

Limitations

The current study provided relevant insights in how police officers identify and protect vulnerable victims. Yet, there are several limitations to this study. First, we were unable to include practical aspects of police work that may affect police officers’ decision-making, such as heavy workload (Turner et al. 2019, Richards et al. 2021), personal opinions and biases (Sleath and Bull 2017, Venema 2019, Franklin et al. 2020, St George et al. 2021). The ATSS design, however, allowed participants to mention different factors that would affect their decision-making, but were not mentioned in the scenarios. The openness of the procedure provided participants with unrestricted answering possibilities (Davison et al. 1995), while the design of the scenarios would closely resemble the daily police practice, and thus daily decision-making (Bosma 2023). Although stress or time constraints were not mentioned here, future research could provide more in-depth insights on the possible influence of the hectic and complex nature of police practice on identifying vulnerability.

Second, this study was limited to specific vulnerability indicators. The ATSS design allowed participants to note other indicators, such as non-verbal cues, but scenarios with other vulnerability indicators, such as other types of crimes, may generate different responses from police officers. The association between gender-based violence and vulnerability and the association between race/ethnicity and vulnerability may be particularly important. Including a broader variation in types of crimes in future research may provide a further nuanced look on the identification and protection of vulnerable victims and possibly the ‘better safe than sorry’ approach. Third, the current study did not include scenarios of specific groups of victims, such as migrants, drug users or sex workers. These groups may display high vulnerability and also be less likely to be identified as vulnerable due to possible biases or marginalisation (Duggan 2018). Future research should, therefore, study these less visible and potentially marginalised groups of vulnerable victims.

Lastly, the current study only focussed on the police while the protection and support of vulnerable victims requires not only the efforts of the police, but also victim support organisations. Each organisation is specialised in providing either protection or support and both organisations need to collaborate in order to fully help the victim overcome the incident. Yet, this poses several challenges. First, definitions of vulnerability may differ between agencies (Bartkowiak-Théron and Asquith 2017). Second, the organisations involved in protecting and supporting a victim need to have a clear understanding of their role in this collaboration (Enang et al. 2021). Since this study only focussed on the police, future research could benefit from addressing the differences between agencies and their understanding and realisation of their role within the approach. Therefore, further research should include both the police and victim support organisations.

In conclusion, the current study provides insights into the identification of vulnerable victims and the responses to these victims. The findings apply to the Dutch context, but may also be interesting for other EU member states where police officers perform the Individual Assessment similarly. These findings provide implications for the implementation of vulnerability assessments and can guide trainings for these assessments. Hence, the current study may support police officers in identifying and eventually protecting this specific group of victims.

Author note

This research is part of a larger collaboration between Tilburg University, Praktikon/Radboud University and the Dutch National Police.

Acknowledgments

The authors wish to express their gratitude to the four police stations and their police officers for participating in this study. The authors would also like to thank Jacqueline Wientjes of the Dutch National Police and Prof. Dr. Toine Spapens of Tilburg University for their feedback on this manuscript. Special thanks to Manouk Thewissen and Joost Leenen for recording the vignettes.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the Internal Security Fund – Police and the European Union under grant number: S100111.