Tracking Risk for Crime Throughout the Day: An Examination of Jersey City Robberies

Review of Literature

Routine activity and opportunity theories provide theoretical frameworks for the convergence of different people, places, and crime outcomes at various times of a day (Cohen & Felson, 1979; Cohen et al., 1981). Routine activity theory posits that criminal opportunities are the result of the convergence of a motivated offender (often assumed), likely target, and a lack of capable guardianship. Whereas people routinely conduct their daily activities at particular times of the day, such as going to work or school, using public transportation, or running errands before dinner. Opportunities for crime at particular places connected to these activities, such as schools, bus stops, or grocery stores, become temporally bounded with them (Bernasco et al., 2016; Haberman & Ratcliffe, 2015; Hagerstrand, 1970; Miller, 2005; Ratcliffe, 2006). Understanding how criminal opportunities vary throughout the day based on environmental features and their interactions at particular places is the focus of this article.

A place’s social relevancy (Kinney, 2010) changes throughout the course of a day. Social relevancy refers to the way in which the value, attraction, or expected utility of a place becomes salient to particular people or groups under certain circumstances. It relates features of a landscape to situational contexts of crime at particular places and times. For example, the social relevance of a school may differ on weekdays during school hours compared to weekends or evening hours (Bernasco et al., 2016). Bars do not typically attract large crowds of patrons until the later evening hours (temporal vibrancy), and weekends may appear different than weekdays. So, the social relevancy of bars on crime at 10 p.m. on a Friday may differ compared to 10 a.m. on a Tuesday (Rossow & Norström, 2012; Walker et al., 2014). While some features of an environment could generally create conditions conducive to crime, their influence can ebb and flow throughout the course of the day. The tempo and rhythm of human activity at places varies dramatically over time according to the ways in which people use their environment.

It is generally known that spatial crime patterns are a function of the environmental backcloth of the area under study (Brantingham & Brantingham, 1995) due in part to particular attractors and generators of crime distributed throughout the landscape (Caplan, 2011; Caplan et al., 2011). Attractive or generative qualities of place features are likely to vary throughout the day, thus altering the crime risk landscape. Caplan and Kennedy’s (2016) work on spatial vulnerabilities concludes that high-risk places depend largely on the spatial and temporal dynamics of crime. This view is supported by Grubesic and Mack (2008) who have argued that the spatial and temporal dimensions of crime are not independent but rather interdependent entities that interact to create situational risks.

Said another way, the spatial influence of environmental features is dynamic and changes according to the way people utilize the physical landscape over time (Caplan & Kennedy, 2016). Along these lines, Lemieux and Felson (2012) report that the risk for violent crime victimization is greatest when traveling between activities. Haberman et al. (2017) show further support for this based on their study of robbery hot spots at different temporal scales in Philadelphia, PA. Time influences the role that various environmental features have on vulnerabilities to crime across a landscape (Irvin-Erickson, 2014).

Just as time influences routine activities, environmental features influence crime patterns and shape the spatial distribution of vulnerabilities to crime that vary over time. This could make places more likely to experience hot spots of one type of crime, but not another, as Haberman (2017) demonstrated. But it also supports the possibility that the same type of crime can present hot spots that ebb and flow at different places within the same jurisdiction throughout a day due to the periodic relevancy of environmental attractors or generators of illegal behavior located there (Rengert, 1997). Fixed activities such as school, work, recreation, and entertainment may limit the movements of people in space and time. Additionally, many businesses, for example, maintain regular hours of operation and can only be visited when open. Despite the assertions of the importance of these time constraints on opportunities for crime, research on this effect to date is limited.

A plethora of existing research identifies several environmental features that could potentially aggravate spatial vulnerabilities to robbery (Bernasco & Block, 2009, 2011; Duffala, 1976; Hart & Miethe, 2014, 2015; Roncek & Bell, 1981; Roncek & Faggiani, 1985; Roncek & Maier, 1991; Smith et al., 2000; St. Jean, 2007; Wright & Decker, 1997). For instance, Bernasco and Block (2011) found that bars, restaurants, barbers/beauty salons, laundromats, liquor stores, grocery stores, general merchandise stores, gas stations, pawn shops, check cashing services, and public transportation stops were positively related to robbery at nearby places in Chicago. Hours of operation probably played a role in the social relevancy of some of these facilities located at particular places in the jurisdiction. Even though facilities such as these are not erected or demolished on an hourly or daily basis, their spatial influence on crime at a microtemporal scale can be quite dynamic (Gerell, 2018).

This dynamic could be explained using the analogy of a kaleidoscope (Kennedy, 1983). Barnum et al. (2017) tested the crime risk kaleidoscope concept across three cities—Chicago, IL, Newark, NJ, and Kansas City, MO—and found differentiating factors among settings for robbery in each jurisdiction. More recently, Connealy and Piza (2019) found that risk factors could vary within subcategories of the same crime type within the same jurisdiction, suggesting that variation in spatial crime patterns could be attributed to situations and opportunities offered by particular environmental features. Absent from these studies though is the inclusion of time. A figurative turn of the kaleidoscope could reflect a change in time within the same study area, similar to Szkola et al.’s (2019) seasonality approach. The current study builds from the robust crime and place literature by simultaneously accounting for time and space in crime occurrences and related generators and attractors of the environmental backcloth. We hypothesize that spatial vulnerability to crime will vary according to the influences of distinct physical features of the landscape over the course of a day.

The Current Study

We focus on robberies in Jersey City, NJ, a densely populated urban environment located in northeastern New Jersey along the Hudson River, across from New York City (Figure 1). Its population of 264,000 residents is the second largest in New Jersey, while geographically being one of the smallest municipalities, with a total land area just under 15 square miles. Approximately 17,000 people live within each square mile; the statewide average is about 1,200 residents per square mile (U.S. Census Bureau, 2017). A lot of people in a relatively small space makes Jersey City ideal for testing the spatial and temporal dynamics of attractors and generators of crime because there are many potential environmental features and routine activities at play.

OPEN IN VIEWER

Figure 1.

Jersey City.

Data and Methods

The dependent variable includes all incidents of armed (firearm, knife, and other weapon) and unarmed street robbery that occurred in Jersey City during calendar year 2014 (N = 445) which are known to and recorded by the police. These crime data were provided at the address level by the Jersey City Police Department (JCPD) administrative records management system and then geocoded to a street centerline shapefile in ArcGIS. Incidents were classified as street robbery, as opposed to, for example, robbery inside a business, by JCPD personnel in accordance with the definition provided by the Federal Bureau of Investigation for the purposes of the Uniform Crime Report. We focused on street robberies for several theoretical and practical reasons. First, environmental criminology suggests focusing on a single crime type (Andresen, 2014). Different crime types may have different spatial and temporal signatures, and aggregating incidents into violent or property crime categories may misrepresent important nuances among individual crime types. Second, street robberies occur outdoors in public locations. As opposed to other types of crime that occur indoors in private spaces, street robberies are more susceptible to the influences of their surrounding milieu in ways consistent with the theoretical propositions laid out above. Third, street robberies occur at discrete moments in time that can be easily ascertained by victims. Time indicators for street robbery incidents are much more likely to be accurate in comparison to those for other types of crime, such as burglary (Lersch & Hart, 2011). Finally, street robberies are a substantial source of fear among members of a community (Wright & Decker, 1997). Henceforth, “street robbery” will simply be referred to as “robbery.”

Robbery crime data were separated into 12 two-hr time intervals: 00:01–2:00 (n = 67), 02:01–04:00 (n = 50), 04:01–06:00 (n = 13), 06:01–08:00 (n = 11), 08:01–10:00 (n = 19), 10:01–12:00 (n = 19), 12:01–14:00 (n = 42), 14:01–16:00 (n = 27), 16:01–18:00 (n = 46), 18:01–20:00 (n = 51), and 20:01–24:00 (n = 73). Our goal here was to use smaller temporal scales rather than longer proportions of a day. Prior risk terrain modeling (RTM) analyses included a temporal element focused on 8-hr policing shifts which fits for a policing-focused study of shifts but overlooks potentially smaller temporal shifts in risk factor significance (Drawve & Barnum, 2018). We sought to explore a more microtemporal scale to identify whether meaningful patterns emerge. Initially, we considered 1-hr time intervals; however, low crime counts precluded this option. ¹ Two-hour time frames were the next best option and were grounded in research by Koper (1995) who found that police can create residual deterrence by spending approximately 15 min or less at target areas every 2 hr.

Environmental Features

We tested the spatial influences of 17 environmental features: bus stops, banks, bars, check cashing service, convenience stores, gas stations, grocery stores, laundromats, liquor stores, manicuring establishments, parking lots, pawn brokers/second hand stores, pharmacies, restaurants, schools, vacant buildings, and variety stores (Table 1). Most environmental feature data sets were acquired from Infogroup. All others—bus stops, liquor stores, parking lots, and vacant buildings—were obtained from the JCPD. Robbery incident and risk factor data sets were collected at the same time and are considered temporally consistent with the then-present state of the city in 2014.

OPEN IN VIEWER

Table 1.

Potential Environmental Risk Factors for Street Robbery, Feature Counts, and Spatial Operationalization(s) Tested.

Potential Risk Factor	n	Spatial Operationalization
Banks	69	Proximity and density
Bars	33	Proximity
Bus stops	812	Proximity and density
Check cashing service	19	Proximity
Convenience stores	68	Proximity and density
Gas stations	46	Proximity and density
Grocery stores	175	Proximity and density
Laundromats	33	Proximity
Liquor stores	175	Proximity and density
Manicuring businesses	43	Proximity and density
Parking lots	86	Proximity and density
Pawnbrokers and secondhand dealers	44	Proximity and density
Pharmacies	65	Proximity and density
Restaurants	574	Proximity and density
Schools	107	Proximity and density
Vacant buildings	885	Proximity and density
Variety stores	39	Proximity and density

Note. All features tested to a maximum extent of three blocks at half-block increments.

Results

Table 2 demonstrates that the 12 2-hr time intervals each had a different set of environmental features—attractors or generators—for street robbery. Figure 2 represents the movement of that spatial vulnerability, supporting the hypothesis and suggesting that robbery does not remain static in space and time even within a day’s time span. Based on these results, we fail to reject the null hypothesis and conclude that spatial vulnerability to robbery varies according to distinct spatial influences of physical features of the landscape over the course of a day. Of the 17 environmental features tested, we identified nine that significantly increased the risk of robbery (banks, bus stops, convenience stores, grocery stores, manicuring establishments, restaurants, schools, vacant buildings, and variety stores). Bus stops were the most prevalent risk factor among all of the models, present in nine of the 12 time periods. Grocery stores and vacant buildings were the next most influential factors, each identified 5 times in the models. Schools appeared in two of the time periods. Risk factors found to be significant in only one time period were banks, convenience stores, manicuring establishments, restaurants, and variety stores. These, perhaps, are the most noteworthy as they help to create highly unique behavior settings for robbery at particular times and places in Jersey City.

OPEN IN VIEWER

Table 2.

Risk Terrain Model Results for 2014 Robbery in Jersey City, NJ.

Risk Factors
			Banks	Bus Stops	Convenience Stores	Grocery Stores	Manicuring	Restaurants	Schools	Vacant Buildings	Variety Stores
Time (24-hr clock)	00:01–02:00	O/SI		D/382					P/1146	D/1146
		RRV		3.16					6.09	2.52
	02:01–04:00	O/SI	D/382	P/1146		D/1146				D/1146
		RRV	6.25	12.13		2.78				2.74
	04:01–06:00	O/SI		P/382		P/1146
		RRV		3.97		14.07
	06:01–08:00	O/SI								P/382
		RRV								27.57
	08:01–10:00	O/SI		D/1146
		RRV		16.41
	10:01–12:00	O/SI		D/382
		RRV		9.12
	12:01–14:00	O/SI				P/764		P/382
		RRV				14.37		6.30
	14:01–16:00	O/SI		P/764						P/382
		RRV		19.51						4.31
	16:01–18:00	O/SI		P/764	D/382
		RRV		22.19	4.99
	18:01–20:00	O/SI				P/764					P/382
		RRV				6.23					5.13
	20:01–22:00	O/SI		D/764					P/1146
		RRV		3.01					12.72
	22:01–24:00	O/SI		D/382		P/764	D/382			D/1146
		RRV		2.94		3.79	3.06			3.77

Note. O = operationalization (P = proximity or D = density); SI = spatial influence (in feet); RRV = relative risk value.

OPEN IN VIEWER

Figure 2.

Risk terrain maps for three time intervals (2:01–4:00 a.m.; 2:01–4:00 p.m.; 10:01 p.m.–12:00 a.m.).

To demonstrate the variety of uniquely vulnerable places for robbery in Jersey City, Figure 2 highlights three time periods: 2:01 a.m.–4:00 a.m. (early morning), 2:01 p.m.–4:00 p.m. (afternoon), and 10:01 p.m.–12:00 a.m. (late evening). The maps categorized as “highest risk,” here meaning “highest vulnerability,” are symbolized with dark gray areas indicating places with relative risk scores (RRSs) two standard deviations or more above the mean, while the lighter gray shaded areas are places equal to or greater than the top 5% of RRSs. These maps indicate the combined risks, or vulnerabilities to crime, of the significant environmental factors in their respective models. We are primarily concerned with the darkest gray areas because they indicate places with the greatest vulnerability to robbery.

The time frame from 2:01 a.m. to 4:00 a.m. exemplified a model when there were four environmental factors present (banks, bus stops, grocery stores, and vacant buildings). The risk terrain map during this time frame includes both dark gray and light gray shading of risk, whereby risks of robbery appear concentrated in the core of the downtown area of Jersey City, with some other vulnerabilities located the northern part of the city too. Note the relatively few places that are symbolized as being at the highest risk (dark gray) during this time period. The risk terrain map for the afternoon time period (2:01 p.m.–4:00 p.m.) was the most dispersed of the three maps and there were just two environmental factors (bus stops and vacant buildings) found to be significant during this time. The late-night time period from 10:01 p.m. to 12:00 a.m. had four significant environmental factors (bus stops, grocery stores, manicuring establishments, and vacant buildings). The risk terrain map during this time frame has the most “highest risk” places, mainly in the downtown area and a few other areas located in the north and east ends of the city. Viewing all 12 maps chronologically depicts a turbulent landscape of vulnerability to robbery in Jersey City throughout a day.

Discussion and Conclusion

This study examined temporal spatial vulnerabilities for robbery in Jersey City, NJ. Grounded in environmental criminology theories, this research affirms that the relevancy and interactions of particular environmental features on crime varies by time of day, over the course of a day, within the same jurisdiction. Time-stable features of the physical environment have a dynamic influence over the timing and location of robbery incidents. Crimes are already well known to cluster at particular places (Weisburd, 2015) of the environmental backcloth (Brantingham & Brantingham, 1995) under study. This article builds on this knowledge to add that within short temporal windows, such as 2-hr periods, attractors and generators of crime fluctuate to influence daily micro spatial crime patterns. This supports Grubesic and Mack’s (2008) notion that the spatial and temporal dimensions of crime are interdependent entities and expands on routine activity theory as it relates to physical features of the environment. Routine activities of people over the course of a day may interact at particular places to shape the social relevancies and influences of particular features of the landscape on illegal behaviors and crime outcomes (Ratcliffe, 2006).

Returning to the three risk terrain models exemplified above, the significant environmental factors in each suggest potential “risk narratives” (Caplan & Kennedy, 2016) that could help explain these spatial vulnerabilities. In the “early morning” hours of 2:01 a.m.–4:00 a.m., four factors were found to significantly connect with robbery incident locations: banks, bus stops, grocery stores, and vacant buildings. Considering the likely routine activities and situational contexts of people in Jersey City at this time, banks may involve the use of ATMs where patrons become potential victims because motivated offenders know they have cash in their wallets. Perhaps cash withdrawals from ATMs near bus stops were intended to pay for transportation. Bus stops may also be viable sites of shelter for homeless people who could be the victims of robbery. Taken together, bus stops located near ATMs might create a potential victim pool for motivated offenders to take advantage of. Grocery stores are inclusive of convenience stores in Jersey City. Opportunities to rob people of cash or recently purchased goods might increase at this time period when there are fewer capable guardians out and about. Perhaps cash from ATMs is intended for use at nearby convenience stores on a person’s way to wait for the bus at a nearby stop. This time period is closing time for bars and some other facilities, so perhaps employees of these workplaces are especially at risk on their way home from work. Vacant buildings near these other risk factors add to the void of place managers or capable guardians and could aggravate nearby risks of robbery (Kennedy et al., 2018).

The significant environmental factors of bus stops and vacant buildings during the “afternoon” hours of 2:01 p.m. to 4:00 p.m. are relevant to school closing times and after-school activities of youth in Jersey City. Prior research by Kennedy et al. (2018) found that youth tend to hang out after school at or near vacant properties, which affords them privacy from adults or other capable guardians. During this period of the day, bus stops and vacant properties may create unique opportunities for turf conflict, offending, or victimization as students are dismissed from schools across the city.

Of the 17 potentially criminogenic features tested here, nine were significant at various times of the day. Considering all of the potential situational contexts for crime spurred by the identification of these features of the landscape can help researchers, practitioners, and policymakers connect spatial–temporal vulnerabilities to robbery in a way that allows for robust problem solving and resource allocations to the places that need them most. Metaphorically, this study brings the focus of crime analysis to the level of the locations of trees that comprise the forest and where they take root. But, stepping back to broadly view the full canopy, it’s evident that if spatial vulnerabilities to crime are unmitigated at local levels, the micro spatial and temporal patterns of crime will likely become more evident hot spots. Pragmatically, police responses to robbery incidents that emerge at particular places within short temporal spans of a day would likely be reactive and only possible if real-time assessments of reported crimes were ongoing throughout a shift. However, if time-stable features of an environment connect with the emergence, persistence, or desistance of hot spots at particular moments of a day, then these vulnerable places could be focused on preemptively to mitigate place-based risks or enhance deterrence at particular times of the day (Koper, 1995; Lum & Koper, 2013; Sherman, 1990).

Police patrols are more effective when they operate with a narrow focus on the most problematic places for illegal behavior (Skogan & Fyrdl, 2004). Many agencies have adopted this operational strategy, but they often focus on the places where crimes have happened in the past (Eck et al., 2005). This study suggests that agencies could identify the underlying environmental features that drive vulnerability at criminogenic behavior settings rather than wait for crimes to emerge or spike. Settings and their constituent features could become the focus of tailored, proactive police interventions designed to mitigate spatial risks for tactical crime prevention at key moments of the day (Caplan & Kennedy, 2016). Multiple community resources can be prioritized at vulnerable places to deliver evidence-based programs designed to mitigate place-based risks and alter the situational contexts for crime (Clarke, 1980; Kennedy et al., 2018). A 2015 study on place-based risk reduction in five cities across the United States found that police and their community partners can yield significant reductions in crime (Kennedy et al., 2015). The current study suggests that timing should also be considered when allocating resources for place-based crime prevention initiatives. The Koper Curve principle (Koper, 1995) states that police officers do not have to stay at target areas for long periods of time to create a deterrent effect after they leave; police can create residual deterrence by spending approximately 15 min or less at target areas every 2 hr. This study offers new insights to further optimize police resources for deterrence effects. If, for example, the riskiest place for robbery was grocery stores between 4 and 6 a.m., then police and other resources could be focused there for up to 15 min toward the beginning of the 2-hr high vulnerability window.

Akin to Rengert’s (1997) study wherein he found that crime clusters were moving at the macrolevel over time, we found that spatial vulnerabilities to crime also fluctuate at the micro levels of both time and space, providing support for our hypothesis. Environmental attractors or generators of robbery at midnight in Jersey City were not the same as spatial features associated with robbery in the morning or afternoon hours. While there were instances where features aggravated robbery risks across multiple time periods, their spatial influence fluctuated and never held the same RRV from one time period to the next. Variations among environmental factors and their spatial influences on surrounding places suggest a temporality of their social relevancy, perhaps due to routine activities and related expectations of motivated offenders and place-based opportunities for crime (Barnum et al., 2017; Bernasco & Block, 2011); Rengert, 1997). As the social relevance of settings change over time, spatial influences of environmental features located there stimulate or revive opportunities for crime at these places. Notably, examining vulnerable places does not remove the importance of the human factor. It simply shifts the focus away from personal characteristics to personal preferences and routine activities observed at places throughout the course of a day. This suggests a way of looking at crime outcomes more as function of a dynamic interaction among people that occurs at places. We now know more about how the attributes of criminogenic places are not constant nor necessarily set in place over time.

The influence of environmental features made some places higher risk, or more vulnerable, to robbery relative to other places. But vulnerability was greatest at places where the spatial influences of many environmental factors colocated. The most vulnerable places for robbery were visually different across risk terrain maps. It appears that each temporally unique analysis figuratively marked a new turn of the “crime risk kaleidoscope” (Barnum et al., 2017; Kennedy, 1983). Adding to that line of research, this study suggests there to be a spatial–temporal kaleidoscope of criminal activity across times of the day within the same jurisdiction (Wicker, 1979).

It is important to note a few important limitations. First, we utilized just 1 year of data, for a single crime type, in a single study setting. Future research could examine whether the current findings are generalizable for different crimes types or within other settings over longer durations of time. Further, we examined 17 potentially criminogenic features based on the extant literature, but additional potentially relevant factors may be tested in future work. Moreover, we looked at general categories of environmental features. Schools included elementary, middle, and high schools; however, the effects of schools on robbery may vary depending upon the grade level considered. It could be worthwhile to disaggregate broader categories of environmental features to more fully understand the underlying crime-generating mechanisms. Our results may be contingent upon the temporal units we employed. We examined risk factors and their spatial influences across 2-hr time intervals because it allowed for a microscale breakdown of the data over a 24-hr period, and 1-hr time intervals did not contain enough crime counts in some temporal periods. Additional research is necessary to determine the most appropriate temporal units with which to examine the relationship between environments and crime.

Despite these limitations, this study offers valuable insights for researchers and practitioners interested in spatial–temporal vulnerabilities to crime. With readily available analytical tools and open data portals in many cities and towns across the country, this study is easily replicable. Insights obtained from the spatiotemporal crime diagnostics performed here should encourage data-informed decisions for crime prevention programming that are transparent because police and their community partners can explain what’s being focused on at the places that need the most attention at particular times of the day. ² This study shows that a small number of places and unique sets of features of the landscape can attract or generate robberies at different times of the day. At the microlevel, people use places throughout the course of the day that creates distinctive spatial and situational contexts for crime. Expectations of illegal behavior or opportunities for crime at each place change as people come and go and utilize different features of those settings. This study also shows that even a physically unchanged built environment can have “fantastically dynamic places” (Jacobs, 1961/1992, p. 14), with different risks of crime at different times because of the social relevancy of particular features of the landscape at particular times and circumstances.