Specialized knowledge: Understanding crime analyst’s roles and responsibilities and the impact of their work

Abstract

Using qualitative interview data, this article examines the role of crime analysts in producing knowledge, as well as the challenges they face. Through the collection and organization of data outlining pertinent information about specific districts, analysts aid in the implementation of policing practices. As such, analysts regard themselves as possessing a specialized form of knowledge, which they incorporate and draw on in the outputs they produce. We conclude that analysts do not always employ rigorous, scientific methodologies, while producing their intelligence outputs, suggesting rather that they rely on their familiarity and specialized knowledge of offenders and crimes in their district. Our findings are important to evaluate and understand how ‘data-driven’ policing is occurring and identifying ways to improve and utilize crime analysis approaches within policing.

Crime analysis is a critical feature of modern policing strategies that centres on proactive problem-solving and intelligence-led decision-making (Hinkle et al., 2020; Innes et al., 2005). Crime analysts (hereon, in analyst) make sense of data using calculations, databases and computer systems to identify criminal trends and target high-risk areas and offenders to better allocate police resources and reduce overall criminality (Fyfe et al., 2017; Taylor et al., 2007). The reliance on data-driven knowledge, however, does not necessarily result in standard and rigorous practices (Innes et al., 2005). To present data in meaningful ways requires analysts interpret police data sources, which, in and of themselves, are not free of bias or limitations (Browning and Arrigo, 2020). Rather, the subjective nature of collection, sharing, recording and analysis practices may result in inadequate or incomplete intelligence (see Ballucci, under review; Cope, 2004; Hulnick, 2006; Innes et al., 2005). Thus, it is increasingly important to acknowledge the environment in which data is collected and translated, as it garners a human element that pervades within police practices and alters the reliability of these sources (Degeling and Berendt, 2018; O’Donnell, 2019). Despite the integration of crime analyst products into policing, there is little known concerning their overall role in the production of knowledge or the factors that influence their outputs. Studies to date acknowledge the importance of both police and analyst training, the limitations of data sources and the subjective nature of these products (Cope, 2004; Hulnick, 2006; Innes et al., 2005; Ratcliffe, 2004), yet fewer have specifically detailed crime analysts’ role in creating analysis that shapes policing decisions.

In our study, we specifically focus on understanding how crime analysts do their jobs. Using qualitative data, we detail analyst’s roles, responsibilities and practices to provide a greater understanding of the outputs they produce. We show that analysts rely on various data sources, such as analysts in other districts, reports from the courts and social services, as well as their personal knowledge of offenders and criminality in their district. Their job responsibilities, which involve a range of analysis to, for example, identify high-risk offenders and hotspots, results in a familiarity with the offenders and crime trends in their district. Analysts perceive the knowledge accumulated through their on-the-job experiences as unique and, thereby, privilege their perspective. We refer to this knowledge as specialized because it is developed through experience with policing systems, techniques, and exposure to offender and crime trend information. We find that crime analysis involves the application of a range of techniques, and that there are variations in practices despite the goal of standardization. In demonstrating the multiple factors that operate in this process, we show that, despite the aim to produce objective, scientific outcomes, discretion, experience and personal knowledge shape the products that analysts produce. Overall, our detailed analysis of practices contributes to the important goal of understanding the epistemological and ontological basis of intelligence work (see Innes et al., 2005), as is necessary to provide realistic knowledge about the ability of crime analysis and data, in general, to produce rational and objective truths.

Crime analysis and policing

Crime analysts’ roles and responsibilities include assisting in the allocation of police resources towards high crime areas and offenders most at risk of recidivation, with the aim of minimizing incidence (Browning and Arrigo, 2020; Ratcliffe and Guidetti, 2008). This approach is characterized as problem-oriented policing, as it aims to employ proactive strategies to prevent and reduce crime (Hinkle et al., 2020). To achieve this, crime analysts methodically gather, organize and analyse data to develop crime patterns and trends, which can then be used to produce crime maps and network charts to provide intelligence to predict and prevent future occurrences (Agarwal et al., 2013; Carter and Phillips, 2013; Sanders et al., 2015). Crime maps, for example, aid in identifying hotspots that indicate an area where a disproportionate number of crimes occur (Boba Santos and Taylor, 2014; Taylor et al., 2007). This technique depends heavily on Geographic Information Systems (GIS), which are technologies that require information and data to produce an accurate depiction of crime clusters (Kumar and Chandrasekar, 2011; Subhashini and Milani, 2015). Using data mining to sort large quantities of data, crime analysts provide information concerning crime trends to better allocate police resources (Perry, 2013), and aid in preventive-policing goals. As such, crime analysis emerged as a resource to assist in the transition away from standard police models that were described as being a ‘means to an end’, focusing on arrest numbers and response times, to a more supported problem-oriented policing model. Over 30 studies in the United States, Canada and the United Kingdom found this method to be effective in reducing crime and disorder, while also addressing its root causes (Hinkle et al., 2020).

Moreover, crime analysts are perceived as an addition to policing, in that they are able to objectively analyse data, and uniquely and systematically make inferences (Evans and Kebbell, 2012). Advocates of crime analysis argue that the algorithms, data and equations used to develop analytical tools offer an objective view of crime patterns and trends. For instance, research by Mohler et al. (2015) discovered that predictive policing algorithms (forecasting when and where crime was going to occur) greatly assisted police officers in their efforts to temporarily reduce crime. In this way, crime analysis is viewed as a successful instrument for addressing an array of crime problems, on both a micro and macro level. This is due to the ability of crime intelligence to range from tactical, strategic and evaluation-oriented levels (Boba Santos and Taylor, 2014); tackling smaller scale crimes involving everyday prolific offenders to larger scale organized crimes (Ratcliffe and Guidetti, 2008).

Of the police departments that applied preventive measures based on analytical products and data, several in major cities in the United States were able to significantly reduce crime in specific areas (Perry, 2013). Analogously, another study in the United States found random foot patrols, a favoured element of community policing, were less effective in reducing crime (particularly violent crime) when compared to targeted foot patrols (Ratcliffe et al., 2011). Thus, crime analysis provides information that directs officers to high crime areas that can assist in prevention and decreasing crime overall (Boba Santos and Taylor, 2014; Chan, 2001; Cope, 2004; Ratcliffe et al., 2011). This is because problem-oriented policing expects officers to develop an encompassing understanding of both victims and offenders, while recognizing the larger picture and uncovering crime patterns and trends (Hinkle et al., 2020). Nonetheless, although crime analysis is increasingly popular and integrated in many countries ¹ to improve policing practices, scepticism remains.

Despite the potential for crime analysis to offer rational, objective knowledge, studies clearly show the limitations of these processes. In particular, scholars question whether crime analysis can be impartial when the collection and transcription of intelligence largely occurs at the hands of officers (Browning and Arrigo, 2020). Furthermore, most algorithms input data surrounding specific crimes, offenders or situations and, thus, patterns are predetermined. Some also argue that data mining, or the process of discovering data patterns and trends, has inherent biases, especially in its role in intelligence-led policing (Degeling and Berendt, 2018). Crime analysis was designed to remove the uncertainty and ambiguities that are a result of human experiences and knowledge (Murray, 2005) to better predict and control crime by inserting a scientific and systematic formula that relies purely on available data and limits the dependence on individual police officers’ experiences (Chan, 2001; Innes et al., 2005; Sanders et al., 2015). However, research findings have suggested the opposite, stating that crime analysis, specifically crime maps, are ambiguous and require interpretation on the officers’ end to understand and utilize the product effectively in the field. This, in turn, further illustrates the point that crime analysis does not result in the full removal of the subjective and human elements of policing practices (Manning, 2001).

Part of the appeal of crime analysis is generated by the fact that it is rooted in data-driven decision-making. To make data useful, however, requires large amounts of translating and processing before analysation by prediction software. Thus, the available raw data can be perceived and labelled in a variety of ways depending on the interpretation of the information (Kaufmann et al., 2019). The resulting patterns must be ‘recognized’ to be actionable intelligence, suggesting officers need to make sense of the trends to properly direct their patrols. Analytical predictions, however, do not guarantee concrete policing decisions and may not be an accurate representation of on-the-beat experiences (Williams et al., 2020). With contextual information limited, police officers are responsible for implementing the findings correctly to reduce crime (Kaufmann et al., 2019). In this way, crime data are an artefact of police practices and becomes an illusion of objectivity (Innes et al., 2005: 52–53), rather than a truly objective, scientific account of crime problems and trends. Innes et al. (2005) describes this phenomenon as an ‘analytic bricolage’ (Innes et al., 2005: 50–52), wherein analysts’ products are deeply rooted in the inherently biased and human elements of policing data. Furthering this is the role that analysts play in inflating police data by connecting any breaks in information using their deductive skills and reasoning (Innes et al., 2005).

Critics of crime analysis and intelligence-led policing (ILP) have also identified predictive policing as harmful for those residing in areas targeted by hotspots. Along these lines, Kaufmann et al. (2019) used the term ‘patterned nudging’ to explain the direction of police patrols to neighbours characterized by higher levels of crime. This increased police presence leads to higher numbers of reported incidents, which results in the emergence of a cyclical pattern, wherein it becomes impossible to remove the bias and assumptions from police data that will later be used to compose analysts’ crime intelligence (Kaufmann et al., 2019; Williams et al., 2020). The concern regarding biased information is further emphasized when considering crime data is typically viewed from a naïve positivist approach, regarded as wholly objective and value free. Contrary to this assumption, crime data are often a product of interference from those collecting, transcribing and interpreting the information. However, once displayed quantitatively, it is often perceived as an objective measure of crime. This concept further supports the claim that crime analysis is not as objective and scientific as originally believed.

The limitations of police data also play a pivotal role in the effectiveness of crime analysis. More specifically, analysts are regularly held to a high standard of work that is dependent on the information of others (Boba Santos and Taylor, 2014). Police data, however, have been found to be of poor quality; lacking the pertinent and objective information required for analyst to produce intelligence of high calibre (Burrell and Bull, 2011; Evans and Kebbell, 2012). While police officers tend to submit every detail of their patrols, this requires analysts to sift through copious amounts of information to detect useful intelligence (Innes et al., 2005). Equally problematic is that on-the-beat officers’ data are inherently biased, reflecting an unreliable account of crime in their district (Cope, 2004; Innes et al., 2005). In addition, crime analysis is more often used to support policing decisions, rather than influence them (Hulnick, 2006), while overall patterns are generalized accounts of crime, meaning crime trends are only detectable with large amounts of data. Thus, due to the infrequent occurrence of particular types of crime, intelligence data are unable to develop a pattern regarding non-recurring incidents, such as violent and rural area offences. These challenges compromise the ability to develop certain crime patterns and trends that can be actionable for police forces (Kaufmann et al., 2019).

Finally, crime patterns and trends often only identify the symptoms of criminal acts and, as a result, are unable to provide police with the root causes of crime. Therefore, although crime analysis is meant to be a preventive approach to crime (Hinkle et al., 2020), analysts cannot provide insight into effective measures for the long-term prevention of recidivism among prolific offenders (Kauffman et al., 2019). Crime analysis also lacks a temporal component, meaning it is unable to show police services whether crime is rising or falling within a clustered geographic area (Ratcliffe and McCullagh, 1998). Analytical tools such as crime mapping and hot spot analysis reflect historic crime trends, which results in the allocations of scarce police resources based on past, rather than current, data (Kaufmann et al., 2019; Perry, 2013; Williams et al., 2020). Ultimately, predictive policing reacts to short and medium-term patterns of incidents rather than to individual incidents as they occur or are reported and, thus, raises questions as to whether intelligence-led policing accomplishes what it originally intended. Our research begins to answer these questions, addressing what the role of analysts is within intelligence-led policing and how their knowledge of crime data and trends is inherently subjective and biased.

In the following section, we describe our methodology, followed by our analysis, where we discuss the responsibilities of crime analysts, providing the necessary background to examine how analysts understand and apply their specialized knowledge. We conclude with a discussion on how our work builds on and advances the existing literature on crime analysis practices.

Methodology

Our study uses data from all available crime analysts in a single Canadian jurisdiction. These 12, semi-structured interviews focus on understanding the roles and responsibilities of crime analysts within their respective police organizations. The reasoning for pursuing qualitative research methods came down to its ability to allow us to delve into individual understandings of crime analysis and its role within police practices through in-depth analysis. Not only does this prove to be more effective than quantitative surveying (Boba Santos and Taylor, 2014; Piza and Feng, 2017), but it also brought forth novel information not available in previous research. This is especially important given that, to date, a majority of the data on crime analysis is quantitative in nature, centres around the point of view of police officers, or compares interviews with both officers and analysts (Cope, 2004; Evans and Kebbell, 2012). Thus, an understanding of crime analysts’ roles from the perspective of analysts themselves will prove beneficial in establishing how these individuals develop their own form of specialized knowledge that is comparable to that of police officers.

The interviews identified which types of products crime analysts’ construct, including offender profiles, prolific offender lists and hot spot analysis. The participants consisted of 10 crime analysts, plus two male police officers who engage in crime analysis strategies, in one Canadian province (to remain anonymous). Among the analysts, six worked for the provincial police forces, while the others were employed federally or municipally. Six were female, the other four were male. Of these 10 respondents, eight reported having completed an undergraduate degree, with more than half also having received a master’s degree in a variety of disciplines. Furthermore, more than half of the analysts interviewed had some sort of prior work experience in analysis or the criminal justice system. In addition, consistent with the current literature (Piza and Feng, 2017; Sanders and Condon, 2017), most of the analysts received no formal training prior to their employment, but rather developed relevant skills and specialized knowledge while on-the-job. Prior to and during the interviews, the participants were made aware of the aim of the interviews, familiarized with the interviewer, and were ensured confidentiality, as well as their right to withdraw their participation or statements at any time.

Upon completing the interviews, one author transcribed each, using QSR NVivo software to help compile and organize the content. Subsequently, this author thematically coded the data. After discussions and reviewing the literature, the other author used NVivo to then thematically identify codes relevant to the analysis of this article. Overall, two major themes emerged from the data: ‘What Crime Analysts Do: Specialized Knowledge: Research, Quality, and Legitimacy’ and ‘How Crime Analysts’ Work is Taken up and/or Perceived’. After establishing the core themes, a number of smaller codes prevailed.

Although qualitative research typically uses an exclusively inductive approach, this research study utilized both inductive (bottom up) and deductive (top down) methods, with greater reliance on the former (Soiferman, 2010). An inductive approach requires the themes to emerge from the interview data itself, rather than with guidance from a review of the existing literature (Soiferman, 2010). Thus, the inductive–deductive methodology best suites this analysis, as it reduces biases while coding, with this literature acting only as a support for the emerging themes. The first author coded the data, beginning with a broad research query on the literature surrounding crime analysis; followed by an in-depth examination of the interview data to develop the main themes. From there, the research questions were established and used to navigate the rest of the relevant literature and support the development of further ideas. In addition, the second author also coded and reviewed the defined themes to ensure they were supported by the interview data, warranting the internal validity of this research study.

Analysis

What crime analysts do: Specialized knowledge: Research, quality and legitimacy

Crime analysts are responsible for translating crime data into actionable products, with the specific aim of assisting in guiding patrols and the allocation of limited police resources. Furthermore, analysts produce timelines for major crime in the district, compute statistics on crime trends for community meetings, and act as a liaison between the police department and community partners, such as probation services. In general, their role focuses on practices that will identify high-crime areas to manage the potential of recidivism by collecting data from multiple sources, including officers, other analysts and social services personnel. As a result, analysts develop a variety of products, including, for example, hotspot analysis, which directs police patrols to areas with a high number of incidents:

[If] we’ve had a real hotspot of break and enters, I’ll print them out a map [. . .] and provide that to [the officers] so that, if they’re wondering whether to turn left or turn right when they leave the detachment, hopefully that b and e hotspot will encourage them to turn right. (Keith)

As reflected earlier, the interviewees consistently discuss practices such as crime forecasting, identifying crime trends, hotspot analysis and high-risk offender designations as key components of analyst roles and responsibilities.

To complete these tasks, analysts must develop a familiarity with criminal activity in their district. This includes being aware of what type of crimes are being committed, as well as where and which high-risk or prolific offenders have been released or recently re-offended. To do this, analysts engage in several practices, including the reviewing of all data sources (which, to re-iterate, include other analysts, courts and social services reports, and their personal knowledge of offenders and crime trends), reading daily crime reports and gathering intelligence from police agents. ² Clare, an analyst, states, ‘I review files in my region every morning and I put those of interest aside’. As a result of regularly reviewing crime data, analysts become very familiar with the neighbourhoods and offenders typically involved in crime, thereby providing them with a sense of specialized knowledge (understanding the crime, offenders and victims in their district).

Moreover, analysts heavily rely on the imputation of data into formulas, methods and software to produce data-driven decisions. The goal of these scientific procedures is to reduce bias in the development of crime data because of its ability to direct policing practices, such as surveillance decisions. Despite the use of instruments to produce rational and scientific products, analysts incorporate their specialized knowledge into these procedures and, as we show, perceive their knowledge as specialized and unique.

That is, analysts consistently discuss their familiarity with the offenders and crime trends in their area, and how this knowledge impacts the ways they construct analytical products. In particular, interviewees commonly feel that they ‘know’ the repeat offenders and have an intuitive sense concerning the connections between offenders and crimes recently committed, even before reviewing the files and reports. One analyst speaks to this idea further, stating their familiarity with offenders in their district:

If there [are] certain types of crimes, I can say which one of our offenders usually does that, or which one lives near if we have a theft or something from a vehicle [. . .] I’m getting to know the offenders. (Yusuf)

Through this interviewee’s comment, we can gather that, with extensive knowledge of offenders’ profiles, crime analysts gain a detailed understanding of the offenders within their district that results in a specialized knowledge. This, in turn, lends confidence regarding one’s familiarity with not only the crime patterns and offenders in their district, but also with their ability to identify and predict crime and criminality. This confidence often results in a sidestepping of typical analytical procedures:

At this point, to be honest with you, because I’ve been here four years, I know my offenders so well because usually they’re up and coming. I haven’t had to research someone in probably over a year, because I know them and I’m up to date on them. (Warren)

While these levels of certainty aid in the speed of the process, they also arguably result in outputs that do not necessarily follow the intended standard protocol when incorporating crime analysis procedures. Nonetheless, analysts perceive their specific knowledge as a benefit that increases and betters their ability to produce analysis. In the context of identifying high-risk offenders, another analyst emphasizes this point:

I feel like I know the list pretty well because, to be honest, [. . .] I’ve worked here for 13 years, so I know who the people are. And that’s really the main reason why I know the list well. (Felix)

This analyst, among others, perceives their work experience and, particularly, the length of time in their role, as resulting in a specialized form of knowledge. This knowledge is considered not only valuable but also increases analysts’ abilities to solve and identify potential criminality. As one crime analyst expresses, ‘we know who our offenders are; it’s not very hard to catch them’ (Eileen). This statement exemplifies the extent to which an analyst feels as though they are aware of their subject matter, and that this awareness enables them to identify and match offenders to crimes within their districts with ease. What this then shows is that, with time, analysts rely on and prioritize their specialized knowledge, even if it contradicts the data available to them.

The extent to which analysts feel their knowledge is specialized is further illustrated by their assertion that their intel can provide police officers with answers. In particular, one analyst recalls an incident where members of their team overlooked the predictions they provided based on offender knowledge, which was later proven to be beneficial:

I had looked at the pattern, [. . .] we had an operational briefing five weeks ago [where] we talked about this series that was going on and I said listen, what’s going on right now, based on this, [. . .] I can tell you it’s going to happen either on a Wednesday or on a Friday and it’s going to happen during the daytime and I’m telling you it’s probably going to be this vehicle that's going to be in the area, which was this guy’s vehicle. [. . .] So, long weekend came, I left the Friday afternoon, came back the Tuesday, started reviewing the files [. . .] [and, when] I got to the fifth file of about 103, [. . .] [there was] a break and enter in [a] community where there was a series of break and enters. It was during the daytime, it was on a Friday, and it was the darn guy’s car in the driveway, so I was three for three. [S]o yes, I did say I told you so then. [sic] (Nina)

Through this, we can see that analysts see their knowledge and abilities as valuable and specialized, yet face the added difficulty of their outputs not being utilized to their full potential.

How crime analysts’ work is taken up and/or perceived

Analysts expressed disappointment with how their information is not fully taken up and utilized in policing practices. To counter this issue and increase the legitimacy and value of their work, analysts include external sources of data (outside of the police or correctional field), such as research, to promote their products. One analyst explains,

Hotspot policing is another big one because that’s something that I want to see the officers do a little bit more of. And, in order to justify that, I have to have research to back it up that shows, ‘Look, all these studies show that this works to help reduce crime, so let’s do it’. (Audrey)

Echoing this sentiment, the same analyst expands on the use of research within their work:

Right now, [I’m] trying to work with some of the Loss Prevention officers to come up with a shoplifting prevention program. So, I go into the literature and see the research that has been done on what works and what doesn’t. Things that have been empirically validated. It just helps me to know what to propose when I’m trying to develop something here. (Audrey)

Thus, crime analysts, to support policing prevention programmes, combine what they learn (from literature and empirical research) with their on-the-job skills and specialized knowledge. In policing practices, this research component becomes a way for analysts to gain legitimacy and support for their products.

Nonetheless, although crime analysts perceiving their work and abilities as specialized, they are aware that it is not always well received. This perception impacts how they construct products and justify their work. For instance, as a result of officer scepticism, analysts often feel as though they are forced to legitimatize their work through process breakdowns, with one analyst explaining that they are ‘always having to justify what [they’re] doing’ (Warren). Analysts will, however, engage in these processes in order to have their work considered not only scientific but also as effective means to determine targets to reduce crime.

Moreover, the implementation of systematic processes further impacts how analysts are able to respond to officers who challenge their findings. In the context of identifying high-risk offenders, analysts explain that they have a process they follow:

If Joe can’t make the list based on the methodology, [. . .] there is no changing the list. [. . .] If they don’t make the average or don’t make the cut, they’re not going to just superficially make it because Constable so and so said that they should be on there. (Nina)

In this way, analysts are not receptive to changing their outputs at the request of police officers. This position illustrates a level of confidence and commitment on the part of the analysts, wherein they stand by their methods and analytical approaches and, therefore, defend the outputs they produce. These outputs, which guide policing, are further asserted as superior through the contrasting of analytical methods with that of police officers, while also suggesting that, when officers draw on analyst knowledge or ask for their opinions, it further exemplifies the value of their work. In describing how police officers approach decision-making, one analyst explains,

Honestly, they have no method or anything. They usually ask me who I think, and they don’t generally listen to that. But they ask me what I think and then they’ll decide [what to do with that]. (Eileen)

Thus, although officers consult analysts on their opinion, analysts continue to perceive their knowledge and intelligence as remaining inferior in the eyes of the police. Without support through the implementation and use of their knowledge by police officers, analysts, therefore, will be ineffectual in assisting predictive policing practices and risk management. This lack of regard for and utilization of analyst outputs within policing may be the result of a number of factors. These include police management neglecting to provide enough resources and manpower to act upon information provided by analysts, the feeling that officers perceive analyst products as inaccurate and inferior sources of data within police practices, and the lack of their products being shared with frontline workers.

The analysts interviewed also revealed that the officers may feel overwhelmed in their responsibilities, which can result in the perception that they do not have the time to inform analysts of the specific incidents and details of their shift. Furthering this point, analysts accounted limited police resources as a reason why officers are not always able to engage in surveillance, such as curfew checks, on the offenders that analysts have deemed high risk. This creates a cycle of frustration, as the inability or failure of officers to act on recommendations to monitor these offenders leads to disregarded work done by crime analysts. One analyst epitomized this recurring dilemma:

. . . there is something missing in the strategy – [. . . and that] is equipping the front-line members or [. . .] the team with the tools in order to enforce what the science is giving us. I can contribute a list, ³ or I can produce a list and provide it and identify, but if we do nothing about those people, then the list is obsolete. (Nina)

This lack of police resources, combined with the perception that analysts’ intelligence is substandard, can leave analysts and analytical products feeling underutilized or dismissed.

Moreover, the resistance experienced by analysts extends beyond officers to include other levels of law enforcement. This includes areas such as management, commonly referred to as ‘the brass’. The main role of this division is to outline clear job expectations for officers while on-the-beat; however, officers’ detailed patrol duties usually do not generate the type of information analysts required to put together comprehensive data-driven products. For example, one analyst describes the typical response from officers when they provide them with an intelligence product outlining curfew checks:

After a while, the offenders were repeated so much that they know who is prolific. Other than the curfew, there is not much proactiveness that they can do to get them unless they do surveillance or something, but if there is no information coming in suggesting that they are committing a crime, there is no surveillance that needs to be done. Now, [my impression] is that they open the list and see the faces and say ‘oh yeah, okay’ and move on. [sic] (Eileen)

Here, it becomes clear that analysts view officers as willing, perhaps even eager, to use their products if it provides information on new or unfamiliar offenders; otherwise, analysts see their products as underutilized and overlooked as secondary information to police knowledge. Thus, the analysts reported resistance against their specialized knowledge when it contradicts an officer’s gut instincts (i.e. enforcing curfews checks and general offender management). This disconnect comes from what analysts perceive as officers preferring to rely on their experiential knowledge (developed on-the-beat) to respond to calls or direct patrols, rather than utilizing the information provided to them by analysts. As a majority of crime analysts are civilians, many do not have the hands-on experience of policing, which impacts officers’ respect towards, and use for, analysts’ products. As one analyst explains,

. . . if you identify to higher-ups that there’s a certain problem, sometimes they’re less likely to take that seriously if it’s coming from a civilian because they’re the police officers. [. . .] There is a bit of that divide between the uniform and the civilian within the police agency. (Audrey)

Nonetheless, although police officers value their own experiential knowledge above the products of analysts, as mentioned earlier, analysts believe they have also acquired a specialized knowledge. In other words, analysts consider their knowledge and awareness of offenders, as well as crime hotspots, as a unique form of expertise to identify target offenders and predict crime in their district. Furthermore, although crime analysts were initially implemented to execute an objective and systematic understanding of crime, our analysis shows that their practices overlap with those of police officers. Despite analysts’ adaptations of specialized knowledge, however, the tension between officers and analysts impacts final intelligence outputs.

Also noteworthy is that while crime analysis plays a role in supporting predictive policing, analysts themselves cannot force officers to monitor and surveil particular offenders. Warren, a crime analyst, explains, ‘I don’t assign the curfew checks, we have a team leader that does that because I can’t task the members. [. . . sometimes] it’s just a matter of reminding the officers to do the curfew checks’. As a result of this resistance to utilizing analyst products, data collection is ultimately impacted, as without curfew checks and surveillance, crime analysts are not provided with all the necessary forms of intelligence.

Overall, as shown earlier, analysts interviewed expressed concern regarding police departments deciding which analytical tools or products to proceed with, if at all any. The hesitation of police agents to unreservedly integrate crime analysis into ILP begins to raise questions as to the true role of analysts within police practices and whether they are intended to portray a façade of objective, scientific, data-driven decisions in the name of crime prevention and prediction.

The impact of officer knowledge

Nonetheless, analysts recognize the value of police work, acknowledging officers as often well informed and knowledgeable. As one analyst explains, ‘the officers have a really good feel [for] which people are being active’ (Sara). Further expounding on this point, one of the police officers interviewed provides their perspective on the specialized knowledge exemplified by the police in their district, asserting that ‘every cop in the city knows exactly [who the offender is], because they see it three times a day in briefing’ (Riley). It is then understood, by both sworn and unsworn personnel, that officers’ familiarity with offenders is one reliable source of information for guiding police decisions, despite its implications on biased and random patrols. Fundamentally, police officers’ information completes the picture.

Analysts are cognizant of the value of data and information in their methodologies, and, particularly, the type of information police officers can provide them, as police officers have access to a level of knowledge that can impact the quality and output of intelligence products. As one analyst explains,

[. . .] the little pieces of information that they [police officers] get on every single call, if there could be any way to share all of that with not only me, but with the other officers [it would be helpful] [. . .] I just find that there’s a lot of knowledge and there’s a lot of experience, but it gets so trapped in these little, not really silos, it’s more just within individuals on a certain platoon. [. . .] I don’t know how much of that [information] actually trickles down to the front-line officers that are the ones who need to know about it. And, vice versa. I don’t know how much of what all those front-line guys know actually gets back up to us. So, there’s kind of a disconnect between management and frontline workers, or frontline officers, and there really needs to be a better flow of information in both directions. (Audrey)

In this way, analysts are aware that the lack of intelligence and information sharing between differing levels of policing impacts the process of data-driven analysis, which, in turn, has detrimental impacts on overall outputs. Further illustrating this point, Audrey continues to explain:

I need to know what is going on [in] the street, you know? Where the problems are. It’s one thing for me to pull up my little crime mapping software and plot some hotspots and say, ‘Ok, there’s some break and enters going on here’, or ‘This area seems hot for assault’, or ‘This bar district is getting a little wild’ or something, but that’s not the same as having officers share information with me. (Audrey)

Thus, there are distinct differences in information gathered from data sources and that of police officers and, as such, there is value in combining these to produce the best results. In other words, police data complete the picture for analysis and are an integral part of the analyst data collection process, as it complements the information from their other sources. The ability to effectively make data-driven decisions is reliant upon the quality and quantity of components such as street check information provided by team members following their patrols. For instance, field interrogations and street checks conducted by police officers support the analysts in their attempts to connect crime series and understand their local offenders. In turn, the information they provide acts as a supplement and can enhance their analysis process. From this develops an internal conflict among analysts; despite being aware of the benefits of using police data and information, analysts remain frustrated with the lack of acceptance of their specialized knowledge and information on the side of the officers. Crime analysts, in practice, are not being utilized the way they were theoretically intended, resulting in consequences for the scientific, objective nature of ILP.

Discussion and conclusion

In the last two decades, an organizational shift has transformed the strategy of frontline policing practices (Deukmedjian, 2006; Hinkle et al., 2020). Previously, the focus was on a community-based approach, comprised reactive patrols. Recently, however, the concentration has been on proactive, problem-solving police practices that emphasize the objectives of ILP in an attempt to predict and reduce crime (Hinkle et al., 2020; Innes et al., 2005). As highlighted throughout this study, crime analysts are at the centre of these crime reduction initiatives. Using multiple data sources, they construct various intelligence products such as crime maps, hot spot analysis, network charts and prolific offender lists to direct police patrols while optimizing scarce resources (Agarwal et al., 2013; Boba Santos and Taylor, 2014; Carter and Phillips, 2013; Hinkle et al., 2020; Sanders et al., 2015; Taylor et al., 2007). Initially, analysts were introduced into ILP to remove the biased and subjective elements of police work by implementing objective, systematic formulas and methodologies to produce data-driven intelligence.

Our work gives nuances to the study by Innes et al (2005), building on the idea that officers shape the data analysts receive and, thus, there is an ‘illusion of objectivity’ surrounding crime data. In particular, our research questions whether analysts are fulfilling their intended roles and responsibilities, suggesting analysts do not always remain impartial and unbiased. Rather, our study points to the finding that analysts too insert their own form of specialized knowledge, similar to on-the-beat officers. The interview data revealed that increased time in the crime analyst position ensues an understanding and familiarity of offenders and crime in an analysts’ respective district. Many analysts interviewed, therefore, valued and preferred their specialized knowledge, referencing times it permitted them to not use the information or data available to them in their files. In this way, analysts considered their specialized knowledge as useful for predicting and preventing crime and, thus, believed it to be a strong and actionable resource for police practices.

The specialized knowledge of analysts, however, contradicts their task to produce an objective and scientific report of crime patterns and trends. When analyst engage and rely on their specialized knowledge, they start to reflect a role similar to police officers. This begs the question of what the roles and responsibilities of crime analysis are if they are producing intelligence based on their job-related knowledge.

Furthermore, our findings are consistent with other existing literature, finding that crime analysts and their products often face resistance by police officers (Chan, 2001; Cope, 2004). Despite analysts feeling that their knowledge is valuable, they described having to constantly explain and justify their actions to police personnel. As a result, analysts regularly mentioned the use of research to substantiate their intelligence, rooting their products in successful, empirical findings. Our research suggests that, while analysts privilege their specialized knowledge, they found using research adds a convincing component to their products for improved officer uptake.

Even with the use of research to uphold their claims, however, our study found analysts continued to face alleged resistance from officers as a result of inadequate police resources, and more importantly, due to officers’ reliance on their own experiential knowledge (Cope, 2004; Evans and Kebbell, 2012; Piza and Feng, 2017). In other words, analysts perceive police officers as preferring grounded work that is rich in context and collected by patrolling the streets. Surmising, officers favour their knowledge over that of civilian analysts, leading to a cyclical pattern of incomplete crime data, which further affects analytical products.

Future research should expand on our findings, addressing what compels analysts to develop their specialized knowledge that, in turn, causes them to stray from data-driven decisions. Our data suggest that a reason for this shift stems from a lack of comprehensive police data, which limits the scope of practice of the analysts, forcing them to fill in the blanks and affecting their ability to remain objective. Nonetheless, an in-depth look at our data is required to analyse how the working relationship between analysts and officers shapes the collection and sharing of crime data. Analysts anticipate officers to effectively communicate with them in order for them to properly perform within their position. Thus, future studies should try to better understand the relationship between analysts and officers, and its influence on analysts’ ability to remain objective. Moreover, to improve the outcomes of problem-oriented policing and crime analysis, local empirical site studies are needed, as their effectiveness is dependent on the implementation and operations of the police agency (Hinkle et al., 2020). Studies that examine police perspective on crime analysis would also increase the likelihood of effective outputs. In addition, research that addresses understandings of crime analysis on behalf of police officers can help generate a comprehensive understanding and perhaps promote change on the officers’ end.

To improve the utility and use of analytical outputs, one suggestion our work points to is the formation of aligned goals across all subsets of police departments (both patrol and brass at the agent end). Moreover, our findings have implications for future policy decisions, suggesting the need to address data biases within intelligence-led policing. As crime analysts were introduced into contemporary policing strategies to be an objective and scientific perspective on crime data and trends, our research is important to understanding how crime analysts actually operate in practice. Our study’s novel research acknowledges what analysts’ roles and duties are and how that contradicts their performance of their responsibilities. By recognizing that analysts deploy a specialized knowledge (a perceived understanding of crime and offenders in their area which guides their analytical products and tools, and which develops from years of experience on-the-job), matched with a lack of police data to compensate for their products, we can understand how and why analysts perform within their role. Accordingly, future policy can aim to address the fundamental cause for why analysts develop a specialized knowledge, thereby returning crime analysis to its objective and scientific roots. Furthermore, it is important to understand where the knowledge and data comes from that comprises analyst products. Gaining insight from analysts themselves allows us to understand what goes into producing their outputs and, ultimately, how crime analysis directs police patrols. Our findings are important to evaluate and understand how ‘data-driven’ policing is occurring and identify ways to improve and utilize crime analysis approaches within intelligence-led policing.