Skip to main content
SearchLoginLogin or Signup

'Top 10' Policing as an Alternative Place-based Strategy: Responding to the Overcomplication and Underestimation of the Law of Crime Concentration

Connealy, N. & Hart, T. (2023). ‘Top 10’ policing as an alternative place-based strategy: Responding to the overcomplication and underestimation of the law of crime concentration. Police Practice and Research, DOI: https://doi.org/10.1080/15614263.2023.2183202

Published onMar 08, 2023
'Top 10' Policing as an Alternative Place-based Strategy: Responding to the Overcomplication and Underestimation of the Law of Crime Concentration
·

Abstract

Recent trends in crime and place research emphasize a micro-level focus on the concentration, stability, and patrollability of crime hot spots. Empirical findings consistently suggest that hot spots are disproportionate in crime concentration, are time stable, and have high crime reduction potential for place-based interventions. Due to a growing literature on hot spot identification techniques, research may be overcomplicating and underestimating the degree to which these concepts manifest, creating unnecessary challenges to crime prevention strategies. The current study analyzes robbery in three cities to determine whether crime hot spot concentration, stability, and patrollability observed at different hot spot aggregations (1%, 0.1%, and the “Top 10”) have characteristics that make them amenable to more efficient crime-reduction initiatives at smaller aggregations. Collectively, current findings suggest that “Top 10” Policing, which focuses on just the 10 most criminogenic hot spots within an agency’s jurisdiction, could have a meaningful impact on overall crime rates.

Crime hot spots are defined and identified as the few places where crime statistically and disproportionately concentrates (Block & Block, 1995; Sherman & Weisburd, 1995). The concept of focusing on and allocating police resources to crime hot spots has become widely adopted by law enforcement (Weisburd & Majimundar, 2018) due to the well-demonstrated crime reduction benefits reported (Braga et al., 2019). The heralded utility of crime hot spots has also led to rapid growth in the evidence base concerning the relationship between crime and place in recent years. Important conclusions pertaining to the degree to which crime concentrates within a select few identifiable hot spot locations (Weisburd, 2015), the stability of crime hot spot locations over time (Hart, 2021; Schnell & McManus, 2020), and the importance of micro-level units of analysis in understanding hot spot composition and operationalizing solutions (Groff et al., 2010; Steenbeek & Weisburd, 2015) have been largely established in the field. Standard practice and general operating procedure in law enforcement now include identifying and deploying specifically tailored crime reduction strategies to known hot spot locations.

However, the rapid increase in crime hot spot research has made it difficult for law enforcement to effectively translate existing scholarship into meaningful practice. The process of selecting crime hot spots often requires a relatively arbitrary set of decision points based on highly technical criteria for identifying hot spots across differing calculation techniques, spatial extents and units of analysis, time periods, jurisdictional settings, and crime types (Chainey et al., 2008). The specificity of such nuanced processes and research findings has disconnected our increased academic understanding of crime hot spots from the ability to readily identify, analyze, and implement interventions and solutions among practitioners.

There currently exists a missed opportunity to apply the widely established empirical findings pertaining to micro-level crime hot spot concentration, stability, and patrollability to a relatively small number of hot spots that may be primarily responsible for driving the disproportionate level of crime observed. For example, studies have consistently found that about 50% of crime in some jurisdictions can be attributed to about 5% of street segments (Curman et al., 2015; Weisburd & Majimundar, 2017), with research also determining that the degree of crime concentration is even further intensified for specific crime types (Sherman et al., 1989). These findings adequately demonstrate the degree of crime concentration across large-scale areas of interest, but 5% of street segments can sometimes reflect hundreds to thousands of disconnected micro-unit locations in a jurisdiction. Furthermore, research findings pertaining to concentration and stability have examined an appropriate number of “research units” to the neglect of a meaningful number of “intervention units”. Existing research using micro-level units of analysis with high unit counts (e.g., all street segments in a city) can lead to conclusions that lack practical applicability, rendering them less actionable and generalizable regarding patrollability.

In response, through a multi-city case study, the current research proposes a new approach to hot spot policing called “Top 10” Policing1, which focuses on the concentration, stability, and patrollability of only the 10 most criminogenic robbery micro-places within a jurisdiction. For comparative purposes, our research is conducted using three robbery hot spot aggregation scales (i.e., 1% of all street segments, 0.1% of all street segments, and the “Top 10” most criminogenic street segments) in three US cities (i.e., San Francisco, CA; Seattle, WA; and Tucson, AZ). We aim to examine whether there are differences in the operational utility of hot spots across each of the three aggregational thresholds tested. This approach highlights the importance of translating research across various settings with differing city sizes, unit counts, crime levels, and available law enforcement resources. We believe this approach holds strategic and empirical advantages for both researchers and practitioners alike.

Literature Review

The Law of Crime Concentration

Spatial approaches to understanding crime patterns have determined that crime is not random or evenly distributed (Eck & Weisburd, 1995). Instead, crime concentrates in a select few places (Weisburd et al., 2009), and the resultant observable concentrations can be specifically attributed to highly localized spatial scales (Braga et al., 2017). The Law of Crime Concentration explored this phenomenon and uncovered some general bandwidths at which crime tends to cluster (Weisburd, 2015). Consistent conclusions have emerged in this literature, suggesting that about 50% of all crime occurs in about 5% of places (Curman et al., 2014; Weisburd et al., 2004) and about 25% of all crime occurs in about 1.5% of places (Weisburd, 2015). Findings across both urban (Weisburd et al., 2017) and non-urban (Gill et al., 2016) jurisdictions suggest that crime is significantly concentrated to a select few micro-level places. Research also indicates that disaggregating crime types results in even fewer criminogenic locations and potentially more intense concentration patterns (Haberman, 2017; Sherman et al., 1989). These results have influenced the approaches routinely used to identify the places that statistically experience the most crime.

Crime Hot Spots

Because crime is exceedingly spatially concentrated, both crime and place-related research and practice have prioritized identifying the most criminogenic and disproportionately risky locations. In this effort, there has been a growing emphasis on determining which factors are conducive or potentially causal to crime concentration (Caplan et al., 2017). Research has considered the relationship of proximate businesses, types of establishments, and other environmental features in the immediate area to crime locations (Brantingham & Brantingham, 1993). Several environmental features have been found to be associated with crime, including those that increase activity and population concentrations (Malleson & Andresen, 2016) and those that provide and attract known criminogenic and illicit opportunities (Brantingham & Brantingham, 1993; Kinney et al., 2008). Other environmentally derived approaches have considered the visual, physical and structural conditions of places, finding that both environmental disorder and decay may impact the likelihood of crime and hot spot formation (Braga & Bond, 2008; Connealy, 2021). A growing body of research also exists that tries to explain the factors, settings, and opportunity structures for crime to occur in hot spots (Connealy, 2022) among other social and health outcomes that co-occur (Weisburd & White, 2019).

An additional avenue of crime hot spot research has focused on the process of identifying and predicting future hot spots (Drawve, 2016). Several key criteria have emerged as important in the hot spot identification process, including (a) spatial unit size, (b) temporal parameters, (c) calculation technique, and (d) specific crime (sub)types. In terms of spatial unit size, crime and place research has confirmed the importance of micro-level units of analysis, suggesting that the geospatial relationship between the environment and crime plays out in a highly localized way (Bernasco, 2010). Highly localized, micro-level units of analysis like street segments (Groff & Lockwood, 2014), intersections (Braga et al., 2010), configural behavior settings (Hart & Miethe, 2015), and point-level addresses (Adepeju et al., 2016) have been found to be comprised of unique micro-communities that resultantly vary in characteristics and crime concentration from unit-to-unit (Weisburd, et al., 2012). The micro-scale of such small and refined hot spot units of analysis has enabled researchers to better minimize error and uncover potentially causal factors, while facilitating the ability of practitioners to conduct more targeted interventions at real, existing, physical locations (Lum et al., 2011; Lum et al., 2018).

The temporal parameters operationalized have also proved important for identifying hot spots (Hu et al., 2018). In the research sphere, crime hot spots are largely determined based on longitudinal calculations that are more suitable for evaluation and causality (Schnell & McManus, 2020). Contrarily, crime hot spots in the practitioner arena are often assessed across varying short-term intervals that correspond to critical law enforcement-related delineations and timeframes, such as 1-day, 7-day, 1-month, and/or 1-year windows. Despite the differing reasoning for each approach in considering time, crime hot spots tend to be largely stable in both short-term intervals (Hart, 2021) and across longer time periods (Braga et al., 2010). Most crime-prone places (i.e., hot spots) are resultantly assumed to be time stable and persistent over time.

The calculation techniques used to identify hot spot units have also been a central area of research. Hot spots are defined as the units that experience the phenomenon of interest in a non-random pattern and in a disproportionately elevated capacity based on analyses of past crime data and/or prospective forecasts. However, the ways in which they are identified as “non-random” or “disproportionately elevated” are relatively subjective. Hot spots can be identified as non-random if they are spatially adjacent to other high crime places or reflect a high level of clustering or density. Or they can be considered disproportionately elevated if they have a high rate of crime occurrence relative to the unit size or population count, if they continuously demonstrate high crime levels over specified periods of time, or if they have unique contextual settings that warrant prioritization. Similarly, hot spots can be calculated using raw counts, standardized rates, or count thresholds based on equal intervals or class breaks. The reasons and criteria applied to identify hot spots also tend to vary based on the crime(s) of interest in hot spots. High-frequency crimes like theft and burglary tend to utilize different approaches than relatively lower frequency crime events like robbery or homicide. Further, crime-specific subtypes have been shown to have unique hot spot locations (Haberman, 2017) and environmental features (Connealy & Piza, 2019). The reality is that there are myriad ways to calculate and identify hot spots, each one conditional on the specific needs of the involved entities, the jurisdiction, and the study or effort at hand.

Hot Spots Policing

Despite a lack of uniformity in defining and identifying crime hot spots, there has been a widespread shift in law enforcement approaches and practices towards hot spot policing. Emphasizing the importance of focusing on places at the most elevated risk for crime has led many agencies to adopt tactics that reallocate and deploy resources and personnel directly to areas with high crime concentrations. A previous National Police Research Platform survey suggested that more than 75% of US police agencies use some form of hot spot policing or related crime hot spot approaches (Mastrofski & Friedell, 2011), with that number purported to be much higher now. Meta-analyses have also found that hot spots policing is effective in reducing crime relative to control locations (Braga et al., 2019), that the positive impact of hot spots policing on crime reduction may be even more salient than prior conservative estimates (Braga & Weisburd, 2020), and that the net benefits and crime control have been found to positively diffuse to other nearby areas (Bowers et al., 2011; Weisburd et al., 2006).

Law enforcement agencies have used various hot spot policing tactics to reduce crime in targeted places, including problem-oriented policing (Skogan & Frydl, 2004), foot patrol (Haberman & Stiver, 2019; Ratcliffe et al., 2011), officer saturation (Piza et al., 2018), and other strategies to increase police presence and visibility (Koper, 1995). A common thread across these approaches is that the specific places (i.e., hot spots) to focus personnel and resources must first be identified. However, with often already strained resources and capacities, the number of “hot spot” places that agencies can meaningfully identify and impactfully intervene upon is remarkably few (Rinehart-Kochel & Gau, 2021). In short, there exist some critical challenges to continuing to apply hot spots research in the practitioner sphere.

Translating Research into Practice

Crime is an exceptionally rare event, though, the locations at which crime occurs are regularly patterned. The notion that crime events are disproportionately attributable to a select few micro-level places is well established (Weisburd, 2015), but the full extent to which crime is concentrated into these micro-level hot spots can be interpreted and communicated in simpler and more meaningful terms. Across a range of studies and cities, crime has been found to concentrate to a small percentage of street segments, often with most crime occurring at between 1% and 5% of all units (Weisburd, 2015). The micro-level units of analysis used in research have also demonstrated utility in capturing unique variation across places and potentially causal factors (Steenbeek & Weisburd, 2015). However, despite the shock factor of such statistics and the obvious degree of clustering to micro-places at face value, the practicality of these types of conclusions is relatively more understated. Further, without a consensus or generalized approach to hot spot selection, the identification of the appropriate number of hot spots and the prioritization of the most criminogenic places is often mired by complex and arbitrary decision points related to spatial unit size, temporal parameters, calculation method, and crime type.

Crime Concentration Considerations

The notion that crime is disproportionately attributable to such few units is a byproduct of the fact that crime does not occur in most micro-places, and that jurisdictions with high unit counts (e.g., street segments) and low crime frequencies will have even more exacerbated levels of concentration (Bernasco & Steenbeek, 2017; Chalfin et al., 2021). Identifying and intervening at such a high volume of units—relative to incidents—is also not feasible for hot spot policing approaches. In other words, interpreting crime concentration levels becomes less meaningful when the 1-5% of “all units” refers to a potentially high volume of disconnected micro-places that cannot be readily considered for crime prevention solutions. Moreover, with higher identified hot spot counts meeting the criteria for an intervention, there is an increased need for police resources, a greater variance in crime levels across “hot spot” units, and a resultant issue in the ability to discern a true “treatment effect” in any subsequent intervention evaluations (Weisburd et al., 1993). The concentration of “crime” as a general construct is also less conducive to prevention efforts, which often are best served through an explicit focus on specific crime types (Andresen & Linning, 2017). For these reasons, we may be understating the degree to which crime, especially specific crime types, concentrate into small aggregation thresholds (e.g., just 10 micro-places or 0.1% of micro-places within a jurisdiction, as opposed to 1-5% of all micro-places).

Temporal Stability Considerations

Hot spot policing research has also attempted to determine the degree to which identified hot spot units remain hot over time. Often, the questions researchers ask center on identifying how long hot spots remain “hot”—or how long a unit is disproportionately elevated in their respective crime level. Regardless of the conception of time, results have consistently indicated that the micro-level hot spots identified have a high degree of staying power and tend to remain “hot” over time. Again, though, as a secondary consequence of including such a high volume of units as hot spots, the temporal stability of high crime places is often examined in the aggregate. Instead of identifying and testing only the select few most criminogenic places and determining whether they persist as hot spots, the ability to make temporal conclusions about specific places is largely washed out by aggregated results. These identified units may also remain “hot” simply because a small count of crime occurrence is the threshold for inclusion. There may be important takeaways and implications if a small number of places with high crime volumes, say just 10 units for example, are high crime hot spots over time. This may indicate that the most crime-prone places are not prioritized and could remain high in crime over time because they are overlooked in current approaches. Considering only the select few most criminogenic hot spots could also allow for identified hot spot units to be better contextualized and operationalized on an individual basis for comprehensive problem-solving and situational crime prevention efforts.

Patrollability Considerations

A key point of crime and place research is to create actionable solutions for practitioners. This requires research to include a rigorous but replicable hot spot identification methodology that simultaneously considers crime levels, crime types, temporal parameters, units of analysis, an appropriate number of units, and the ability to evaluate and translate findings meaningfully and specifically. As data capacities and research have advanced, the disconnect between research and practice has also resultantly widened. Many hot spot identification techniques require multiple complex steps, and may incorporate elements like hot spot adjacency, density, or standardization into the selection criteria, all of which may inadvertently hinder or mask the simple identification of the most criminogenic places. Similarly, common micro-level units of analysis like fishnet grids, hexagons, or cells are often highly technical, high frequency, and are not readily conformable relative to law enforcement or patrol-specific boundaries. Using units of analysis and aggregation thresholds that are actionable in quantity and discernible in the real world is an important patrollability consideration in crime reduction efforts.

The Current Study

To better align ongoing crime hot spot research to practice, the current study provides an in-depth examination of the concentration, stability, and patrollability of micro-level hot spots across three different aggregation thresholds. First, we propose a simple and replicable way for hot spots to be identified that prioritizes the most at-risk places based on smaller spatial units, singular crime types, and total crime count. We then extend the law of crime concentration by measuring the percentage of crime attributable to different aggregation thresholds like 1% of street segment units, 0.1% of street segment units, and just the “Top 10” street segment units in three cities based on total robbery count. Potential differences in concentration intensity across the three hot spot aggregation thresholds within each city were measured using Gini coefficients and visualized with Lorenz Curves (Bernasco & Steenbeek, 2017; Hart, 2020). The level of crime concentration at smaller aggregation thresholds (i.e., the “Top 10” street segments) than commonly explored may remain high and provide law enforcement with the potential to enact a meaningful crime reduction strategy at a more actionable number of places.

Second, the current study considers the temporal stability of hot spots at each aggregation scale to determine whether crime-prone places remain hot spots over time in each city. Hot spots were ordinally classified based on whether they reoccur, emerge, or disappear from one time interval to the next. This classification scheme was used to create a Dynamic Variability Index (DVI), which is an established indicator of hot spot stability over time (Adepeju et al., 2016). Intra-city differences in DVI scores across the three aggregation scales were examined using one-way repeated measures analysis of variance (ANOVA) for ranked-ordered data, based on the Friedman’s test, to determine whether temporal stability varies for hot spots identified at each threshold (1% of all street segments, 0.1% of all street segments, and the “Top 10” street segments). Crime hot spot stability has been demonstrated in the literature before, but this examination provides specific insight into the temporal stability of the highest crime units across three different aggregation scales not previously tested.

Third, the study examines the patrollability of hot spots and the potential crime reduction capacity of each aggregation threshold. Hot spot unit locations in each city and at each threshold were assigned a small spatial effects buffer zone and situated as either standalone, discontinuous “islands” (Caplan, 2011; Wheeler, 2021), or as spatially contiguous, connected, “clumpy” units (Adepeju et al., 2016; Hart, 2021) to determine operationality and patrollability as a hot spot patrol area. Then, area-to-perimeter (AP) ratios (Bowers et al., 2004) were calculated for the hot spot patrol areas to determine geographic compactness as an indicator of effective patrollability. Differences in AP ratios were compared across the three different aggregational thresholds in a city to examine operational utility and patrollability on an intra-city basis.

By expanding upon empirically demonstrated concepts related to concentration, stability, and patrollability across three aggregational thresholds in multiple cities, the present research is positioned to meaningfully contribute to an actionable understanding of the appropriate aggregation for the most crime prone, time stable, and operationally useful micro-level hot spots. The study is guided by the research questions listed below. We hypothesize that there will be a demonstrable utility for law enforcement in focusing efforts on the “Top 10” aggregation of criminogenic locations.

  1. Does a meaningful proportion of crime (i.e., robbery) concentrate among a relatively small aggregation of hot spots, defined as 1% of street segments, 0.1% of street segments, and the “Top 10” street segments experiencing the most robbery, and if so, is the distribution of crimes evenly dispersed across units within each aggregation?

  2. Does the temporal stability of crime hot spots vary significantly across the three hot spot aggregation thresholds examined?

  3. Does the patrollability of crime hot spots vary significantly across the three hot spot aggregation thresholds examined?

Methodology

Study Setting

The present study identifies and examines robbery hot spots in three US cities: San Francisco, Seattle, and Tucson. Robbery data for each city were obtained from online open data portals.2 Data collected included crime data from 2015 through 2021 with associated XY coordinates for each incident and relevant city-maintained street segment files amenable to a geographic information system. Robbery data for each year were then assigned to the nearest street segment unit using ArcGIS Pro’s spatial join tool.3 Each city is unique to the study due to variation in city size and total unit count, population, robbery volume and rates, and available police resources. San Francisco is the smallest in size with ~47 square miles of land area, but the largest in terms of population (874,784) and police resources (2,140 total officers and 1,316 deployable patrol officers in 2021). Seattle is ~84 square miles in total and has 1,325 total sworn police officers and approximately 878 deployable patrol officers for the 741,251 people living within the city. Tucson is notably larger, with ~227 square miles of total land area. Though, the population of 545,340 and the recorded total number of officers (807) and deployable patrol officers (406) is considerably lower.4

Unit of Analysis and Robbery Hot Spots

The unit of analysis for the current study is individual street segments in each of the three cities examined. Street segments are commonly used in crime and place research because they provide fixed unit boundaries, minimize error, and adequately capture the unique variation across each unit (Weisburd et al., 2012). Summated counts of recorded robbery incidents were generated for each street segment within each city from 2015 through 2021. This approach best prioritizes the most criminogenic street segments in each city, so that robbery hot spots can be conceptualized and compared at three distinct aggregation thresholds not previously tested in the field: 1% of all street segments with the highest concentration of robberies, 0.1% of street segments with highest concentration of robberies, and the “Top 10” street segments with the highest concentration of robberies.

The hot spot identification strategy proposed in the present study simplifies several critical decision points researchers and practitioners face in selecting hot spots including spatial unit size, crime level, and crime type (or problem of focus). The “Top 10” Policing approach allows for a consistent, replicable technique for selecting hot spots. First, we advocate for the spatial unit to be individual street segments, as street segments have been found to demonstrate variation in characteristics and crime levels while providing fixed, real, discernible boundaries for identification and intervention. Second, we suggest that hot spots are selected based on total crime volume, which prioritizes the most active and at-risk locations that disproportionately drive crime levels upward and simplifies the complexities associated with considering hot spots based on density, standardization and weights, or temporal sequencing. Third, we further limit the focus to robbery due to its environmentally derived opportunity structure (Zhou et al., 2021), its potential discernible spatial patterning, its severity and public fear factor (Camacho-Doyle et al., 2022), and its emphasis on many police interventions (Hatten & Piza, 2022). Testing a specific crime type also appropriately extends the concepts of concentration, stability, and patrollability of hot spots routinely posed in crime and place research.

Table 1 presents descriptive statistics for the street segments analyzed in the current study at the three hot spot aggregation scales considered and the robbery incidents associated with them. Details presented in the table include the average number of street segments and robbery incidents contained within each hot spot aggregation scale and the average minimum number of robberies recorded from 2015 through 2021 that defined the threshold for being included in each hot spot aggregation.5 It can be readily observed in the table that even for a presumably small aggregation of hot spots like the 1% of street segments experiencing the most robbery, the number of incidents defining the threshold for inclusion is relatively low. For example, on average, the threshold to be included in the 1% of street segments experiencing the highest concentration of robberies was just four in San Francisco, two in Seattle, and one in Tucson. Furthermore, statistics presented in Table 1 shows that a meaningful percentage of all robberies concentrated among even the smallest proportion of hot spot street segments in each location. On average, nearly 9% of all robberies in San Francisco and Tucson, and nearly 11% of all robberies in Seattle, concentrated within the “Top 10” most criminogenic units from 2015 through 2021.

Table 1

Descriptive Statistics for Street Segments and Robbery Incidents, by Study Area and Hot Spot Aggregation Scale, 2015-2021

Location/level of

Street segments

Robbery Incident (%)

hot spot aggregation

Min

Max

Mean

SD

 

Min

Max

Mean

SD

 

Threshold

San Francisco, CA

N = 16,995

 

N = 22,580

1% of street segments

152

198

174.00

17.08

35.10

52.17

44.33

7.04

4

0.1% of street segments

17

20

18.43

0.98

9.12

14.63

12.55

2.07

13

“Top 10” street segments

10

12

10.71

0.76

6.65

10.63

8.66

1.34

17

Seattle, WA

N = 23,086

N = 10,554

1% of street segments

209

240

227.29

10.70

53.80

62.84

57.60

2.62

2

0.1% of street segments

17

20

18.43

0.90

16.35

22.52

18.26

2.26

6

“Top 10” street segments

9

11

9.71

0.76

9.02

13.07

10.58

1.42

11

Tucson, AZ

N = 60,126

N = 8,152

1% of street segments

545

740

635.43

64.88

100.00

100.00

100.00

0.00

1

0.1% of street segments

51

71

63.14

7.54

27.10

38.14

32.44

3.25

4

“Top 10” street segments

7

12

10.00

1.73

 

6.09

12.44

8.87

1.96

 

7

Note. Descriptive statistics for each location and level of hot spot aggregation are based on counts of street segments from 2015 through 2021. Threshold is the annual average minimum number of robbery incidents associated with a street segment that was included in a hot spot. As ties were not broken at the threshold cutoff value, the annually derived minimum and maximum values can vary slightly above or below the aggregation (for example, in one year there was only 7 units that met the “Top 10” criteria in Tucson.

Temporal Stability

Knowing whether crime patterns are more stable over time can have important implications for place-based policing strategies (Adepeju et al., 2016; Hart, 2021). Concentrations of robbery that remain more stable, for example, could be locations where targeted interventions could be more effective compared to robbery hot spots that are more fluid over time. Because the focus of this study is on understanding patterns of crime concentration at small hot spot aggregations, we examined whether hot spot temporal stability varied across the 1% of street segments, 0.1% of street segments, and the “Top 10” street segments that experienced the highest levels of robbery in each city. In our study, we define crime hot spot stability as the degree to which most robberies consistently concentrate among the same street segments where crime clustered the most from one year to the next. Stability is measured using Adepeju and colleagues’ Dynamic Variability Index (DVI; Adepeju, et al., 2016).

To assess hot spot stability with DVI scores, we started by comparing all street segments where robberies concentrated in 2016 to the segments where they concentrated in 2015, at each of the three hot spot aggregation thresholds considered. We then classified the 2016 segments as either reoccurring, newly emerging, or disappearing. The 2017 segments were then compared to the 2016 segments, classifying the 2017 segments in the same manner. We repeated this process at each of the three hot spot aggregations in a city for the remaining years (i.e., through 2021). Next, we used the street segment classification information to compute a DVI score as the percentage of all segments within a hot spot aggregation group that changed (i.e., either disappeared or emerged) from one year to the next. Scores were calculated for each year except 2015, which was the first interval or base year used in the classification process. The DVI values indicate the proportion of all hot spots that were emerging or disappearing across each iteration of time (i.e., from one year to the next). Smaller DVI values represent hot spots that are more time stable, whereas larger DVI values indicate hot spots that are relatively less time stable. Friedman’s test was used to compare differences in average DVI scores across the three micro-granular levels of aggregation within each city.

Patrollability

The third aspect of the study considers the proximity and size of the identified crime hot spots to make conclusions about patrollability. Even with a small number of places to prioritize, hot spots policing efforts may be less operational if many of the identified hot spot units are isolated. Although each hot spot must be assessed individually to better understand its contextual drivers of crime occurrence, it may be more patrollable if the locations of interest are small in quantity and can be functionally tied into other hot spots or existing patrol efforts. This study examines the potential overlap and spatial proximity of hot spots to form patrol areas, and then uses area-to-perimeter (AP) ratios of the formulated patrol areas to give estimates for compactness and patrollability across the three aggregation thresholds tested.

The first step in this process required identifying the yearly hot spots from each hot spot aggregation threshold. Second, the identified hot spots were assigned a spatial effect zone that considers the criminogenic influence of the hot spot unit on the surrounding units which can be likened to a deterrability area based on officer presence. Hot spot street segments were extended to a 1200ft x 1200ft buffer (i.e., equivalent to about three street blocks [see Piza et al., 2017]) and overlapping hot spot buffers were dissolved into a singular patrol area to identify possible “clumps” while isolated hot spots were identified as “islands” (Wheeler, 2021). Third, the compactness of the hot spot patrol areas was assessed by calculating AP ratios across the three aggregation scales to indicate the patrollability of the hot spot patrol areas. A larger AP ratio corresponds to an area that could be patrolled more efficiently in that the hot spots are more compacted. A non-parametric Friedman’s test was conducted to compare differences in average AP ratios for the three hot spot aggregations within each city.

Results

Crime Concentration

The degree to which crime concentrates at each hot spot aggregation threshold was measured using standard Gini coefficients6 and visualized with Lorenz Curves, revealing a consistent pattern within all three study cities. Specifically, as the spatial aggregation threshold becomes smaller, crime concentration becomes more uniformly distributed across hot spots. This means that at lower aggregation thresholds, the places identified as hot spots have similar crime volumes, whereas the units identified as hot spots at the 1% aggregation threshold have a great degree of variance; some units are especially high crime areas, while other units have low robbery counts and likely should not constitute a hot spot. For example, Panel A in Figure 1 shows a very uneven distribution of all robbery incidents across all street segments in San Francisco (G = .892), a pattern consistent with the law of crime concentration (Weisburd, 2015). However, Panel D shows that the distribution of incidents in San Francisco is most evenly distributed when the geographic scales are the smallest, like the “Top 10” most criminogenic places (G = .135; 95% CI [.093, .175]). These patterns suggest that hot spot policing tactics could focus on the “Top 10” street segments and could be very successful, given that fewer resources would be required to patrol these areas, a large proportion of crime concentrates in these places (see Table 1), and the distribution of crimes is most uniformly distributed across these hot spots (Panel D of Figure 1).

Figure 1: Lorenz Curves and Gini Coefficients by Study Area and Hot Spot Aggregation

Panel

San Francisco

Seattle

Tucson

A

Chart, line chart Description automatically generated

Chart, line chart Description automatically generated

Chart, line chart Description automatically generated

(G = .892)

(G’ = .873)

G’ = .898

B

Chart, line chart Description automatically generated

Chart, line chart Description automatically generated

Chart, line chart Description automatically generated

(G = .304; 95% CI [.266, .356])

(G = .409; 95% CI [.372, .462])

(G = .460; 95% CI [.442, .479])

C

Chart, line chart Description automatically generated

Chart, line chart, scatter chart Description automatically generated

Chart, line chart, scatter chart Description automatically generated

(G = .203; 95% CI [.156, .253])

(G = .179; 95% CI [.134, .262])

(G = .125; 95% CI [.102, .165])

D

Chart, line chart, scatter chart Description automatically generated

Chart, line chart Description automatically generated

Chart, line chart, scatter chart Description automatically generated

(G = .135; 95% CI [.093, .175])

(G = .147; 95% CI [.086, .215])

(G = .081; 95% CI [.057, .107])

Note: Lorenz Curves presented in Panel A are for all street segments; curves in Panel B are for 1% of street segments; curves in Panel C are for 0.1% of street segments; and curves in Panel D are for the Top 10 street segments. Bernasco and Steenbeek’s (2017) generalized unbiased Gini coefficient (G') was used when the number of street segments was greater than the number of robberies. General patterns show that crime concentration is more evenly dispersed across units at smaller scales.

Temporal Stability

Temporal stability was analyzed by calculating DVI scores at the three hot spot aggregation thresholds for each city. A Friedman’s test was conducted to determine whether there were significant differences in dynamic variability at the three levels of hot spot aggregation considered. Results are presented in Table 2 and indicate that the observed DVI scores do not vary significantly across any of the levels of hot spot aggregation considered within any of the three case-study cities. Like the patterns of crime concentration, the patterns of stability observed in the data were also consistent within each city. For example, the stability of crime hot spots associated with the 1% of street segments where crime is concentrated the most in Tucson, based on the median DVI score, was 33.91, compared to 36.27 for the “Top 10” street segments. However, the apparent difference in stability between these two levels of hot spot aggregation was not statistically significant, χ2 (2, N = 6) = 4.00, p = .135. This suggests that the effectiveness of hot spot policing strategies that focus on the very small hot spot aggregations (e.g., the “Top 10” criminogenic street segments) may not be in jeopardy due to hot spot stability, because hot spots are just as stable at very small aggregations as they are in relatively larger aggregations of hot spots where crimes cluster.

Table 2

One-way Repeated Measures Analysis of Variance (Friedman’s Test) for Ranked DVI Scores, by Study Area, 2016-2021

Location/level of

DVI scores

Friedman’s test

hot spot aggregation

Min

Max

Mdn

 

X2

df

p

W

San Francisco, CA

2.33

2

.311

n.s.

1% of street segments

24.45

45.62

26.28

0.1% of street segments

22.22

46.03

30.95

“Top 10” street segments

18.42

36.84

28.95

Seattle, WA

1.33

2

.531

n.s.

1% of street segments

29.45

35.06

31.42

0.1% of street segments

20.00

43.64

29.09

“Top 10” street segments

16.00

56.00

28.00

Tucson, AZ

4.00

2

.135

n.s.

1% of street segments

29.52

38.57

33.91

0.1% of street segments

36.81

40.66

39.01

“Top 10” street segments

23.53

41.18

36.27

 

Note. Descriptive statistics for each location and level of hot spot aggregation are based on the six DVI scores, one score for each year from 2016 to 2021. A DVI score for 2015 was not calculated because it was the first interval or base year used in the hot spot classification process used to compute scores.

Patrollability

Patrollability was analyzed by calculating AP ratios at the three hot spot aggregation thresholds for each city. A Friedman’s test was conducted to determine whether there were significant differences in the compactness of hot spots, or patrollability, across the hot spot aggregations considered within a city. Results are presented in Table 3 and indicate that—unlike DVI scores—AP ratios vary significantly in some cities, but only at certain levels of hot spot aggregation. For example, significant differences in operationality were observed in Seattle (χ2 [2, N = 7] = 13.56, p = .001) and Tucson (χ2 [2, N = 7] = 12.29, p = .002), but not in San Francisco (χ2 [2, N = 7] = 0.86, p = .651). As shown in Table 4, however, in Seattle the differences in the rank-ordered AP ratios were observed only between the “Top 10” street segments and the 1% of street segments aggregation levels (W+ = 1.93, z = 3.61, p = .001). A similar pattern was observed in the data for Tucson between the “Top 10” and 1% hot spot aggregations (W+ = 1.86, z = 3.47, p = .002). This suggests that patrolling the “Top 10” street segments may be more operationally burdensome because hot spots are less compact, but only when compared to a relatively larger level of micro-granular aggregation (i.e., 1% of all street segments). Paradoxically, relatively larger levels of hot spot aggregation could be more operationally burdensome than smaller levels of hot spot aggregation (i.e., “Top 10” street segments) due to the overall number of hot spots that would warrant targeted patrol (see Table 1). Therefore, decisions about which hot spots to target based on aggregation levels would likely be driven in large part by an agency’s available resources.

Table 3

One-way Repeated Measures Analysis of Variance (Friedman’s Test) for Ranked Area-to-Perimeter Ratios, by Study Area, 2016-2021

Location/level of

Area-to-perimeter ratio

Friedman’s test

hot spot aggregation

Min

Max

Mdn

 

X2

df

p

W

San Francisco, CA

0.86

2

.651

n.s.

1% of street segments

719.40

812.30

756.60

0.1% of street segments

688.10

875.20

732.60

“Top 10” street segments

711.70

811.00

734.20

Seattle, WA

13.56

2

.001

.968

1% of street segments

785.60

840.70

802.90

0.1% of street segments

704.00

745.60

726.40

“Top 10” street segments

676.20

745.60

706.50

Tucson, AZ

12.29

2

.002

.878

1% of street segments

799.50

860.80

834.40

0.1% of street segments

694.80

734.50

714.50

“Top 10” street segments

658.40

739.30

674.60

 

 

Note. Descriptive statistics for each location and level of hot spot aggregation are based on the seven AP ratios, one ratio for each year from 2015 to 2021.

Table 4

Post Hoc Comparisons of Area-to-Perimeter Ratios Between Levels of Hot Spot Aggregations in Seattle, WA and Tucson, AZ

Location/

Post hoc comparisons

comparison groups

W+

z

p

r

Seattle, WA

“Top 10” - 0.1% street segments

0.85

1.60

.326

n.s.

“Top 10” - 1% street segments

1.93

3.61

.001

1.36

0.1% - 1% street segments

1.07

2.00

.135

n.s.

Tucson, AZ

“Top 10” - 0.1% street segments

0.71

1.34

.544

n.s.

“Top 10” - 1% street segments

1.86

3.47

.002

1.31

0.1% - 1% street segments

1.14

2.14

.098

n.s.

Note: Mean ranked differences were assessed using Wilcoxon Signed Rank test and significance values were adjusted by the Bonferroni correction for multiple tests.

Discussion and Conclusion

The present study used a multi-city and multi-threshold approach to examine several field concepts related to crime hot spots: concentration, stability, and patrollability. The first section of the study provided a simplified approach to hot spot identification through “Top 10” Policing and analyzed crime concentration within each of the three cities into identifiable crime hot spots at different aggregation thresholds than previously examined in the field (i.e., 1%, 0.1% and the “Top 10” street segments). The findings confirm prior research related to the Law of Crime Concentration (Weisburd, 2015) and offer further support and evidence suggesting that the merit of the “law” could be meaningfully reduced to even more refined hot spot aggregation thresholds. The present application of the Law of Crime Concentration, particularly in the academic and evaluation spheres, currently underestimates the degree of crime concentration across jurisdictions by using higher-level aggregations like 1-5% of places, generalized “crime” categories, and non-uniform units of analysis in a capacity that does not translate amenably to place-based law enforcement practices. Thus, evidence from the present study suggests that “Top 10” Policing could be an effective alternative hot spot policing strategy for agencies seeking to identify and reduce crime in hot spots in a more effective and efficient manner.

For example, the first research question examined the level of crime concentrations at each of the three aggregation thresholds within a city and the dispersion of crime within them. Within each of the three cities analyzed, crime concentration patterns remained highly localized down to aggregation thresholds like 0.1% of street segments or just the “Top 10” most criminogenic places. This indicates that the “Top 10” hot spots were responsible for a disproportionate share of robbery incidents in each respective year. In Seattle, just the “Top 10” places experienced at least 9.1% of all robberies each year, on average; in San Francisco and Tucson, those percentages were 6.7% and 6.1%, respectively. In some study years, the “Top 10” places accounted for as much as 13.1% of all robbery incidents (i.e., Seattle in 2021). Lorenz Curves and Gini coefficients also indicate that the concentration of crime is significantly more evenly distributed among the “Top 10” hot spots than the other aggregation thresholds. With a significant and disproportionate share of crime in the “Top 10” street segments, crime reduction benefits may be facilitated by prioritizing the “Top 10” units experiencing the highest level of crime and initiating comprehensive problem-solving efforts at this more actionable threshold. Hot spots identified at the “Top 10” aggregation share a similarly high—and disproportionately high relative to other aggregations—levels of crime. The degree of explicit prioritization, comprehensive problem-solving, and situationally specific interventions needed to achieve meaningful crime reductions may be more readily achievable for departments pursing hot spots policing at an aggregation of just 10 units (i.e., the most criminogenic units in the entire jurisdiction).

The current trajectory of crime hot spot research may be placing too much emphasis on nuanced calculation methods and theoretically oriented units of analysis, with the highest crime micro-level places often going unnoticed or unprioritized amidst the noise. Future crime and place research can still operationalize technical calculation methods and localized units of analysis. However, considering evidence produced from the current study, such research may be better positioned to create a more efficient impact on crime by also articulating any hot spots related findings within the scope of an actionable framework. As noted previously, we refer to this framework as “Top 10” Policing. This approach could better empower law enforcement to apply, replicate, and act on current crime and place/hot spot research through an explicit focus on the most criminogenic places for problem-solving. The concentration statistics produced in the analysis also serve to depict the even more intensified levels of concentration for specific types of crime. We hypothesize that future analyses that explore other types of crime will likely reflect similar degrees of concentration, especially for low-frequency crime types.

The second research question was related to the temporal stability of hot spots. Regardless of the time period or technique, research has largely concluded that hot spots are time stable. However, building on these examinations, it is important to determine whether the select few places where crime concentrates the most are persistent hot spots over time. Results of a Friedman’s test for intra-city differences across the three hot spot aggregation thresholds indicated that there was no difference in stability between the aggregations. Hot spots that are among the “Top 10” most criminogenic places are just as time stable—despite having exceedingly higher identification thresholds—as those being identified at 1% or even 0.1% thresholds. For example, the threshold for being classified as a 1% hot spot in some cities (e.g., Tucson) only requires experiencing one crime a year at the micro-level units of analysis tested. With temporal stability being similar across all hot spot aggregation thresholds tested, crime prevention efforts may be best served by sending resources to the most criminogenic “Top 10” places—the places that are ascribing a high level of both crime and temporal stability.

The third research question considered the patrollability of the hot spots examined. Focusing on the “Top 10” places also proved to have operational- and patrol-specific utility. The AP ratios calculated for each hot spot aggregation threshold within each city indicated that there were some significant differences in perceived patrollability. The “Top 10” units were less compact and more spatially distal than the than 1% of the units in Seattle and Tucson. This suggests that the “Top 10” units are less patrollable, in that they are not always contiguous or compacted in physical shape and size. However, this presents some interesting implications. First, the fact that the “Top 10” most criminogenic units in some cities are not immediate spatial neighbors may contradict many widely held assumptions that the most criminogenic units are just the most commercialized and densely populated units often nested in a downtown or entertainment district. In Seattle and Tucson, in particular, the identified “Top 10” hot spot units were located disparately across the city. Second, the “Top 10” hot spot units may be less compact as there are far fewer units to possess any degree of contiguity. Even when not compact, the “Top 10” units include a maximum of 10 hot spot patrol areas (in this case just 10 individual street segments). The other aggregations may demonstrate more compactness and seem more patrollable, but likely average many more patrol areas that are much larger (and thereby require more resources for comprehensive problem-solving and subsequent interventions). Thus, despite being potentially more distal, the “Top 10” individual street segment hot spots are inherently easier to patrol than 150-plus units at the 1% aggregation threshold (152 in San Francisco, 209 in Seattle, and 545 in Tucson [see Table 1]).

The current study demonstrates that place-based policing can be simplified and translated into practice more easily and efficiently through the “Top 10” Policing framework. The continued focus on complex hot spot identification techniques and highly nuanced statistical or spatial approaches has mired the practicality that crime hot spots were designed to afford. There is important information we can learn and equip law enforcement agencies with by simply including a focus on a smaller quantity of places with real, physical boundaries that experience the highest levels of crime. A fractionally small number of street segment units (i.e., the “Top 10”) can account for significant shares of crime and crime concentration, can be stable over time, and can be more operationally practical and patrollable relative to available police resources.

In short, focusing efforts on the “Top 10” hot spots may lead to a more effective and measurable intervention strategy with the potential for more sustained positive effects over time. The simplicity of the “Top 10” approach to identifying hot spots, its efficiency in physical space, manpower, and resources, and the potential crime reduction benefits it possesses position it to be an immediately implementable hot spots policing solution. The “Top 10” approach can readily complement existing hot spots policing efforts that are more general in scope, which could also serve to help identify and prioritize “Top 10” locations of interest. Other strategies, including the approach being replicated across a multitude of crime types and looking for overlap, as well as higher-level hot spot strategies that focus on larger patterns and aggregations, would still appropriately co-exist with the “Top 10” approach. “Top 10” Policing is not beholden to singular crime types, limited to just 10 locations, or incompatible with other strategies. The framework is meant to illuminate that at just the “Top 10” most criminogenic micro-level locations, meaningful crime reductions can be achieved through more intensive interventions.

There are several important research implications for “Top 10” Policing. Focusing on a more actionable number of units can facilitate research that is better at contextualizing high crime environments through techniques like systematic social observation, which has recently been used in remote capacities to better understand high crime or novel environments of interest (Schnell et al., 2019; Sytsma et al., 2021). Future research can also examine the extent to which other crime types concentrate to select aggregations like the “Top 10” and can incorporate jurisdictions of varying size with different units of analysis. Replicating the “Top 10” approach across different crime types and examining the potential for overlap in hot spot locations across different crime types could serve to maximize crime reduction returns at the fewest number of micro-level places. In terms of temporal stability, research can examine the extent to which the most criminogenic places change over time, and if changes are observed, whether they can be attributable to law enforcement, legislation, community-specific, or other social initiatives, and so on. Lastly, research needs to ensure that the results produced are operationally relevant, in terms of actionability and patrollability. As the general goal of identifying and intervening upon hot spots is crime prevention, future research should also consider the level of resourcing needed to achieve comprehensive problem-solving and meaningful crime reductions at the identified hot spots. Results that incorporate real, physical boundaries that can be conformed to patrol areas and research that uses an appropriate number of intervenable units may yield a greater impact on crime reduction. “Top 10” Policing provides a framework for researchers to link methodological and data-centric crime hot spot concepts to real-world practitioner purposes.

Limitations

Despite the current study’s contribution to the extant place-based scholarship, certain limitations must be acknowledged. First, the cities selected were largely predicated on longitudinal data accessibility through public-facing open data portals. Current results may not hold constant for other cities or jurisdictions that have different units-of-analysis-to-crime-occurrence ratios. Second, robbery was selected as the crime of interest to further refine the Law of Crime Concentration and other hot-spot-related concepts (e.g., stability, operationality, patrollability) in the ways proposed (testing of different aggregation thresholds), though, other crime types may have yielded differing results based on their varying frequencies or environmentally situated patterning. Third, the methods used to identify hot spots were intentionally simplistic, which may have led to prioritizing total robbery volume over other intuitive reasons a unit may be classified as a hot spot like population counts, clustering or density with other criminogenic units, the presence of problematic businesses and establishments, or law enforcement informed narratives.

The use of 10 units in the “Top 10” Policing approach is also in itself an arbitrary aggregation threshold. However, the concentration, stability, and patrollability of crime were demonstrated using just the “Top 10” units, which reflects a quantity of places that most departments could reasonably allocate resources to and conduct meaningful problem-solving at. Despite being an arbitrary threshold, the “Top 10” aggregation does have applied, practical, and empirical merit. In an applied sense, the “Top 10” aggregation provides a catchy, interpretable moniker to galvanize law enforcement “hot spots” strategy around. In a practical sense, the “Top 10” micro-level places enable deep, problem-solving, and strategically tailored interventions for a specific crime type, with specifically identified causes, at specific places—which research has shown reduces crime. Lastly, in an empirical sense, the “Top 10” shows the extent and degree of micro-level crime concentration with a relatively high percentage of incidents occurring at an exceptionally small threshold of places (much less than the 1-5% metrics commonly cited across the field). In cities with lower total units of analysis, the approach could be extended to the 0.1% of places (e.g., San Francisco and Seattle averaged just over 18 places at this threshold level) where resources are available. The impetus of this research is to move away from research informed thresholds like 1-5% of places that may lack practical merit relative to aggregation thresholds like the “Top 10” most criminogenic places.

Conclusion

The current study demonstrates that common crime and place findings regarding the concentration, stability, and patrollability of crime hot spots are preserved and improved at smaller aggregation thresholds than previously examined. The study reveals that the 1% aggregation threshold counts too many units as hot spots (that are potentially not high in crime at such micro-levels) and often does not involve an actionable number of places for practitioners to prioritize. Based on the significant and disproportionate levels of crime concentration, stability and patrollability observed at just the “Top 10” hot spot street segments, we contend that at smaller aggregation thresholds: (a) hot spots can be more simplistically and meaningfully identified, (b) the concentration, stability, and patrollability of hot spots are preserved and/or improved, and (c) the degree of resource allocation and comprehensive problem solving necessary to initiate crime reduction is more achievable. With this approach, departments of all sizes and resource constraints could enact hot spot policing interventions and have a meaningful impact on crime reduction by extending the law of crime concentration to even smaller scales.

The micro-unit of analysis trend in place-based research has appropriately prioritized micro-level units of analysis due to their ability to minimize error and capture unique micro-community compositions. However, the quantity of micro-units of analysis often leads to high n hot spot identification, analyses, and conclusions that prohibit the contextualization of specific locations, problems, and solutions. “Top 10” Policing provides a framework for researchers to advance the concepts of concentration, stability, and patrollability observed in the literature in a way that satisfies practitioner needs to home in on specific places. Law enforcement agencies adopting “Top 10” Policing may be able to yield substantial crime reduction benefits through comprehensive problem-solving efforts at an actionable number of the most problematic places. With a simple hot spot identification strategy, a reasonable number of hot spot units, and the potential to generate a sizeable crime reduction, “Top 10” Policing is immediately actionable for law enforcement entities interested in adopting, augmenting, or revamping hot spot policing.

References

Adepeju, M., Rosser, G., & Cheng, T. (2016). Novel evaluation metrics for sparse spatio-temporal point process hotspot predictions-a crime case study. International Journal of Geographical Information Science, 30(11), 2133-2154.

Andresen, M. & Linning, S. (2017). The (in)appropriateness of aggregating across crime types. Applied Geography, 35(2012), 275-282.

Bernasco, W. (2010). Modeling micro-level crime location choice: Application of the discrete choice framework to crime at places. Journal of Quantitative Criminology, 26(1), 113–138.

Bernasco, W. & Steenbeek, W. (2017). More places than crimes; Implications for evaluating the law of crime concentration at place. Journal of Quantitative Criminology, 33(3), 451-467.

Block, R. & Block, C. (1995). Space, place, and crime: Hot spot areas and hot places of liquor related crime. In Eck, J. & Weisburd, D. (eds.) Crime Places in Crime Theory, 145-183.

Bowers, K., Johnson, S., Guerette, R., Summers, L. & Ponyton, S. (2011). Spatial displacement and diffusion of benefits among geographically focused policing initiatives. Campbell Systematic Reviews, 3.

Bowers, K., Johnson, S. & Pease, K. (2004). Prospective hot-spotting: The future of crime mapping? British Journal of Criminology, 44(5), 641-658.

Braga, A., Andresen, M. & Lawton, B. (2017). The law of crime concentration at places. Journal of Quantitative Criminology, 33(3).

Braga, A. & Bond, B. (2008). Policing crime and disorder hot spots: A randomized controlled trial. Criminology, 46(3), 577-607.

Braga, A., Papachristos, A.V., & Hureau, D. (2010). The concentration and stability of gun violence at micro places in Boston, 1980-2008. Journal of Quantitative Criminology, 26, 33-53.

Braga, A., Turchan, B., Papachristos, A. & Hureau, D. (2019). Hot spots policing and crime reduction: An update of an ongoing systematic review and meta-analysis. Journal of Experimental Criminology, 15(3), 289-311.

Braga, A. & Weisburd, D. (2020). Does hot spots policing have meaningful impacts on crime? Findings from an alternative approach to estimating effect sizes from place-based program evaluations. Journal of Quantitative Criminology, 38(1), 1-22.

Brantingham, P.J. & Brantingham, P.L. (1995). Criminality of place: Crime generators and crime attractors. European Journal of Criminal Policy, 3(3), 5-26.

Camaco Doyle, M., Gerell, M. & Andershed, H. (2022). Perceived unsafety and fear of crime: The role of violent and property crime, neighborhood characteristics, and prior perceived unsafety and fear of crime. Deviant Behavior, https://doi.org/10.1080/01639625.2021.1982657.

Caplan, J. (2011). Mapping the spatial influence of crime correlates: A comparison of operationalization schemes and implications for crime analysis and criminal justice practice. Cityscape, 13(3), 57-83.

Caplan, J., Kennedy, L., Barnum, J., Piza, E. & Rennsion, C. (2017). Crime in context: Utilizing risk terrain modeling and conjunctive analysis of case configurations to explore the dynamics of criminogenic behavior settings. Journal of Contemporary Criminal Justice, 33, (2), 133-151.

Caplan, J., Neudecker, C. H., Kennedy, L. W., Barnum, J. D., & Drawve, G. (2021). Tracking risk for crime throughout the day: An examination of Jersey City robberies. Criminal Justice Review, 46(2), 259–273.

Chainey, S., Tompson, L., & Uhlig, S. (2008). The utility of hotspot mapping for predicting spatial patterns of crime. Security Journal, 21, 4-28.

Chalfin, A., Kaplan, J., & Cuellar, M. (2021). Measuring marginal crime concentration: A new solution to an old problem. The Journal of Research in Crime and Delinquency, 58(4), 467-504.

Connealy, N. (2021). Understanding the predictors of street robbery hot spots: A matched pairs analysis and systematic social observation. Crime and Delinquency, 67(9), 1319-1352.

Connealy, N. (2022). The influence, saliency, and consistency of environmental crime predictors: A probability score matching approach to test what makes a hot spot hot. Justice Quarterly, 1-24.

Connealy, N. & Piza, E. (2019). Risk factor and high-risk place variations across robbery targets in Denver, Colorado. Journal of Criminal Justice, 60, 47-56.

Curman, A., Andresen, M. A., & Brantingham, P. J. (2015). Crime and place: A longitudinal examination of street segment patterns in Vancouver, BC. Journal of Quantitative Criminology, 31(1), 127–147.

Drawve, G. (2016). A metric comparison of predictive hot spot techniques and RTM. Justice Quarterly, 33(3), 369-397.

Eck, J. & Weisburd, D. (1995). Crime places in crime theory. In John E. Eck and David Weisburd (Eds.), Crime and Place: Crime Prevention Studies, 4, 1-33. Monsey, NY: Willow Tree Press.

Gill, C., Wooditch, A., & Weisburd, D. (2016). Testing the “Law of Crime Concentration at Place” in a suburban setting: Implications for research and practice. Journal of Quantitative Criminology, 33(3), 519–545.

Groff, E. & Lockwood, B. (2014). Criminogenic facilities and crime across street segments in Philadelphia: Uncovering evidence about the spatial extent of facility influence. The Journal of Research in Crime and Delinquency, 51(3), 277-314.

Groff, E., Weisburd, D., & Yang, S.-M. (2010). Is it important to examine crime trends at a local “micro” level? A longitudinal analysis of street to street variability in crime trajectories. Journal of Quantitative Criminology, 26(1), 7–32.

Haberman, C. (2017). Overlapping hot spots? Examination of the spatial heterogeneity of hot spots of different crime types. Criminology & Public Policy, 16(2), 633-660.

Haberman, C. & Stiver, W. (2019). The Dayton foot patrol program: An evaluation of hot spots foot patrols in a central business district. Police Quarterly, 22(3), 247-277.

Hart, T. (2020). Identifying situational clustering and quantifying its magnitude in dominant case configurations: New methods for conjunctive analysis. Crime & Delinquency, 66(1), 143-159.

Hart, T. (2021). Investigating crime pattern stability at micro-temporal intervals: Implications for crime analysis and hotspot policing strategies. Criminal Justice Review, 46(2), 173-189.

Hart, T. C., & Miethe, T. D. (2015). Configural behavior settings of crime event locations: Toward an alternative conceptualization of criminogenic microenvironments. Journal of Research in Crime and Delinquency, 1-30.

Hatten, D. & Piza, E. (2022). When crime moves where does it go? Analyzing the spatial correlates of robbery incidents displaced by a place-based policing intervention. Journal of Research in Crime and Delinquency, 59(1), 128-162.

Hu, Y., Wang, F., Guin, C. & Zhu, H. (2018). A spatio-temporal kernel density estimation framework for predictive crime hotspot mapping and evaluation. Applied Geography, 99, 89-97.

Kinney, B., Brantingham, P. L., Wuschke, K., Kirk, M. G., & Brantingham, P. J. (2008). Crime attractors, generators, and detractors: Land use and urban crime opportunities. Built Environment, 34(1), 62–74.

Koper, C. (1995). Just enough police presence: Reducing crime and disorderly behavior by optimizing patrol time in crime hot spots. Justice Quarterly, 12(4), 649-672.

Lum, C., Koper, C. S., & Telep, C. W. (2011). The evidence-based policing matrix. Journal of Experimental Criminology, 7(1), 3–26.

Lum, C., Koper, C. S., Wu, X., Johnson, W., & Stoltz, M. (2018). The proactive policing lab. Final report to the Laura and John Arnold Foundation. Fairfax: George Mason University

Malleson, N. & Andresen, M. (2016). Exploring the impact of ambient population measures on London crime hotspots. Journal of Criminal Justice, 46, 52-63.

Mastrofski, S. & Fridell, L. (2011). Police Departments Adoption of Innovative Practice. National Police Research Platform.

Piza, E., Chauhan, P. & Travis, J. (2018). The effect of various police enforcement actions on violent crime: Evidence from a saturation foot-patrol intervention. 29(6-7), 611-629.

Piza, E., Feng, S., Kennedy, L. & Caplan, J. (2017). Place-based correlates of motor vehicle theft and recovery: Measuring spatial influence across neighborhood context. Urban Studies, 54(13), 2998-3021.

Ratcliffe, J., Taniguchi, T., Groff, E. R., & Wood, J. D. (2011). The Philadelphia foot patrol experiment: A randomized controlled trial of police patrol effectiveness in violent crime hotspots. Criminology, 49(3), 795–831.

Rinehart-Kochel, T. & Gau, J. (2021). Examining police presence: Tactics, and engagement as facilitators of informal social control in high-crime areas. Justice Quarterly, 38(2), 301-321.

Schnell, C., Grossman, L. & Braga, A. (2019). The routine activities of violent crime places: A retrospective case-control study of crime opportunities on street segments. Journal of Criminal Justice, 60, 140-153.

Schnell, C. & McManus, H. (2020). The influence of temporal specification on the identification of crime hot spots for program evaluations: A test of longitudinal stability in crime patterns. Journal of Quantitative Criminology, 38(1), 51-74.

Sherman, L., & Weisburd, D. (1995). General deterrent effects of police patrol in crime "hot spots": A randomized, controlled trial. Justice Quarterly, 12(4), 625-648.

Skogan, W. & Frydl, K. (2004). Fairness and effectiveness in policing: The evidence. Committee to Review Research on Police Policy and Practices. The National Academies Press.

Steenbeek, W. & Weisburd, D. (2015). Where the action is in crime? An examination of variability of crime across different spatial units in The Hague, 2001–2009. Journal of Quantitative Criminology, 32(3), 449–469.

Sytsma, V., Connealy, N. & Piza, E. (2021). Environmental predictors of a drug offender crime script: A systematic social observation of Google Street View Images and CCTV footage. Crime and Delinquency, 67(1), 27-57.

Weisburd, D. (2015). The law of crime concentration and the criminology of place. Criminology, 53(2), 133–157.

Weisburd, D. (2018). Hot spots of crime and place-based prevention. Criminology & Public Policy, 17(1), 5-25.

Weisburd, D., Bernasco, W. & Bruinsma, G. (2009). Putting Crime in its Place: Units of Analysis in Geographic Criminology. Springer-Verlag.

Weisburd, D., Braga, A. A., Groff, E. R., & Wooditch, A. (2017). Can hot spots policing reduce crime in urban areas? An agent-based simulation: hot spots policing and urban area crime reduction. Criminology, 55(1), 137–173.

Weisburd, D., Groff, E. & Yang, S. (2012). The Criminology of Place: Street Segments and our Understanding of the Crime Problem. Oxford University Press.

Weisburd, D. L., & Majimundar, M. K. (2017). Proactive policing: Effects on crime and communities. Washington DC: National Academy of Sciences.

Weisburd, D., Petrosino, A. & Mason, G. (1993). Design sensitivity in criminal justice experiments. Crime and Justice, 17, 337-379.

Weisburd, D. & White, C. (2019). Hot spots of crime are not just hot spots of crime: Examining health outcomes at street segments. Journal of Contemporary Criminal Justice, 35(2), 142-160.

Weisburd, D., Wyckoff, L., Ready, J., Eck, J., Hinkle, J. & Gajewski, F. (2006). Does crime just move around the corner? A controlled study of spatial displacement and diffusion of crime control benefits. Criminology, 44(3), 549-592.

Wheeler, A. (2021). Clumpy Hotspots. https://andrewpwheeler.com/2021/03/09/clumpy-hotspots/.

Zhou, H., Liu, L., Lan, M., Zhu, W., Song, G., Jing, F., Zhong, Y., Su, Z., & Gu, X. (2021). Using Google Street View imagery to capture micro built environment characteristics in drug places, compared with street robbery. Computers, Environment and Urban Systems, 88, 101631.

Author Bios

Nathan T. Connealy is an Assistant Professor in the Department of Criminology and Criminal Justice at the University of Tampa. His research focuses on the spatial analysis of crime patterns, the intersection of policing, crime prevention, and technology, and quantitative research designs. His scholarship has recently been published in peer-reviewed outlets including Criminology, Justice Quarterly, Criminology & Public Policy, and Crime & Delinquency among others.

Timothy C. Hart is a faculty member of the Department of Criminology and Criminal Justice at the University of Tampa. His areas of interest include survey research, applied statistics, geographic information systems (GIS), and victimization. His scholarship appears in various academic journals, including the Journal of Quantitative Criminology, the Journal of Research in Crime and Delinquency, Criminal Justice and Behavior, and the British Journal of Criminology.

Comments
0
comment
No comments here
Why not start the discussion?