Yelping about a good time: casino popularity and crime

Full Terms & Conditions of access and use can be found at
https://www.tandfonline.com/action/journalInformation?journalCode=gjup20
Criminal Justice Studies
A Critical Journal of Crime, Law and Society
ISSN: 1478-601X (Print) 1478-6028 (Online) Journal homepage: https://www.tandfonline.com/loi/gjup20
Yelping about a good time: casino popularity and
crime
Virginia Sosa, Gisela Bichler & Lianna Quintero
To cite this article: Virginia Sosa, Gisela Bichler & Lianna Quintero (2019) Yelping about
a good time: casino popularity and crime, Criminal Justice Studies, 32:2, 140-164, DOI:
10.1080/1478601X.2019.1600820
To link to this article: https://doi.org/10.1080/1478601X.2019.1600820
Published online: 01 Apr 2019.
Submit your article to this journal
Article views: 136
View related articles
View Crossmark data
Yelping about a good time: casino popularity and crime
Virginia Sosa, Gisela Bichler and Lianna Quintero
Criminal Justice, California State University San Bernardino, San Bernardino, USA
ABSTRACT
Electronic word-of-mouth (eWOM) is fast becoming a standard
medium of communication when people are looking for information about where to go for a good time. Exploring the content of
reviews posted online could expose criminogenic conditions associated with properties, thereby providing an opportunity to examine crime problems afflicting private facilities with public data. This
study investigates whether Yelp comments, together with property characteristics, can account for crime and deviance occurring
at or near casinos in Southern California. While a correlational
analysis revealed that two Yelp-based variables – property magnetism and star ratings – were associated with fewer reported
issues of crime and deviance, a multivariate analysis revealed the
complex nature of risky facilities. Holding all other factors constant, perceived staff friendliness and casino magnetism are significantly related to higher crime rates. Notably, the presence of
slot machines is associated with lower crime, irrespective of crime
measure used. Study limitations restrict generalizability; however,
these findings suggest that public data can be used to investigate
the criminogenic capacity of risky-facilities.
KEYWORDS
Casinos; crime pattern
theory; risky facilities; Yelp.
com; magnetism; crime
generator; crime attractors
Introduction
Modern casinos can be thought of as entertainment hubs, offering an assortment of
adult recreation and leisure amenities (Maheshwari, 2016). While successful casinos draw
large crowds and generate economic benefits that extend far beyond the facility, they
are also linked to crime problems that affect patrons and neighboring communities
(Williams, Rehm, & Stevens, 2011). A case occurring in San Diego in June 2017 illustrates
this point. A Lucky Lady Casino patron was shot during the early morning while leaving
the casino with his winnings. Two men followed him out of the casino, and as he drove
out of the parking lot, the assailants intentionally rear-ended the car, causing him to
stop at a gas station across the way. During the attempted robbery the victim was shot 4
times. While the victim survived, his assailants were not apprehended (KTLA5, 2017; see
also CBS News, 2017). Given the pervasiveness of user generated electronic media,
notorious incidents can have a significant impact on casino reputations and the communities that house them.
CONTACT Gisela Bichler [email protected] Department of Criminal Justice, California State University, San
Bernardino, 5500 University Parkway, San Bernardino, CA, 92407, USA
CRIMINAL JUSTICE STUDIES
2019, VOL. 32, NO. 2, 140–164
https://doi.org/10.1080/1478601X.2019.1600820
© 2019 Informa UK Limited, trading as Taylor & Francis Group
Two objectives drive this study. First, we adopt a risky facilities perspective, seeking to
learn which property characteristics are associated with greater social media attention
about reported onsite crime and deviance problems and observed crime in the vicinity
of the property. Second, we aim to investigate the casino-crime nexus using information
about a broad set of properties located in different communities. Aside from a few
notable locations where many casinos co-locate (e.g., Las Vegas, Atlantic City, and Reno),
communities tend to have one or a small number of facilities. To move beyond a case
study approach, casino researchers must face the data challenges associated with multijurisdictional data collection. Moving the field forward, this study investigates violent
and non-violent crime occurring at and within the vicinity of Southern California casinos
using publicly available information, including: (1) incident data extracted from crimemapping.com; (2) property characteristics and reported onsite crime and deviance
problems drawn from Yelp reviews; and, (3) a travel network, linking Yelp.com reviewers’
home cities to the casino locations, to calibrate the magnetism or attractiveness of each
property.
The paper proceeds as follows. First, we describe the different types of casinos in
Southern California and review the ongoing debate about the economic benefits
generated by casinos versus the potential criminogenic impact they have on their
neighboring communities. Next, we provide a brief overview of two related theoretical
frameworks – crime pattern theory and risky facilities. We also describe eWOM by
explaining how the Yelp.com rating system functions. The sample, variables, and
methodology used in the study are explained. Finally, we report our findings and
discuss what these results suggest for investigating risky facilities and policing casino
neighborhoods.
Project background
Casinos
Adult entertainment hubs
Gambling facilities located in California run the gamut from small, privately-owned card
rooms, also known as card clubs, to Las Vegas style resort casinos located on large tracts
of Indian lands. The legal distinctions between these types of properties are linked with
facility features and gambling activities. Cardroom owners can conduct card games once
they have registered with the Attorney General and have obtained facility approval and
licensing from the local jurisdiction (Dunstan, 1997). The regulatory authority overseeing
Indian casinos is delineated by a three-tiered system implemented by the National
Indian Gaming Commission (NIGC). Class I consists of facilities hosting social games
with minimal value prizes associated with traditional tribal ceremonies or celebrations
and gaming oversight is completely under the tribe’s jurisdiction; Class II gaming
consists of limited card games, lotto, and bingo that are monitored by the tribe and
the NIGC; and, Class III includes house-banked card games, all other games, and casinostyle slot machines which are regulated by both the state and the tribe equally
(Contreras, 2006).
Gambling facilities are important to study in Southern California as their functionality
is complex. Many gambling facilities act as adult entertainment hubs (Maheshwari,
CRIMINAL JUSTICE STUDIES 141
2016). As multipurpose facilities, casinos offer an array of amenities such as restaurants,
bars, hotels, pools, spas, slot machines, gambling tables, and special event venues. By
shifting the focus of activity from gambling to other pursuits, properties support social,
physical, cultural, and entertainment activities for adults. A key factor driving casino
development is the promise of economic benefits.
Economic benefits
Tribal and non-tribal gaming operations provide substantial economic benefits. Casinos
generate considerable employment opportunities, associated with running the facility,
operating non-gaming amenities, and new construction (De Anda, Levine, Schrader, &
Thornberg, 2014; Dunstan, 1997). For example, studies show that in 2012, tribal casinos
in California generated approximately $8 billion in economic output, which supported
56,000 jobs state-wide (De Anda et al., 2014); county-level estimates show that in the
four years after an Indian casino opens there is a five percent increase in jobs (Evans &
Topoleski, 2002); and, when a new casino opens, unemployment rates tend to decrease
by 12 percent within the vicinity of that casino (Gerstein et al., 1999). Locally, communities with an Indian casino generate greater tax revenue (De Anda et al., 2014, Evans &
Topoleski 2002) – even though tribal casinos do not pay into state and local taxes, their
employees pay taxes (Dunstan, 1997). In addition, large, successful tribal casinos often
donate earnings to non-profit organizations and civic infrastructure projects, thereby
improving public services in the region.
Economic incentives may also drive the development of Las Vegas style resort
properties. Examining the locus of revenue generation, gambling is not the most
lucrative activity; rather, the amenities and other activities that a casino offers generate
the most revenue (Dunstan, 1997). It follows that facilities that offer a full range of
amenities, and attract overnight guests, will stand to generate greater economic
impacts. Despite the noted benefits offered by gaming facilities, casinos can also contribute to crime problems in their host communities.
Crime/casino impact
Table 1 summarizes research investigating the association between casinos and crime.
While the studies are diverse in scope and methodology, they are reasonably consistent
in two ways. First, studies show that the presence or introduction of casinos is often
associated with increased crime levels in the host community as recorded in official
statistics. For instance, Stitt, Nichols, and Giacopassi (2003) discovered that larceny and
liquor violations were significantly higher in casino communities. And, examining crime
over a period of five years, Grinols and Mustard (2006) argue that although crime is low
after a casino opens, it increases over time. These authors estimate that 8.6% of the
observed property crime and 12.6% of the violent crime in casino counties are due to
the presence of gambling facilities. To date, few studies refute the casino-crime connection. For example, using a mixed-method design that included focus groups with
community residents, Belanger, Williams, and Arthur (2012) found no reported increase
in crime following the introduction of casino facilities in two different First Nation’s
communities. Of note, the observed casinos were located in rural areas, operating within
a 50-kilometer catchment area with less than 35,000 local residents. Although the
142 V. SOSA ET AL.
Table 1. Selected studies investigating crime at casinos.
Citation Location Focus Crime Time Method Purpose/Issue Finding
Albanese (1985) New Jersey Atlantic City Part I crimes 1978–1982 Correlational
study
Casinos in relation to crime Casinos have no e
ffect on the
serious crime; crime rose
due to other factors
Hakim and Buck (1989)
1 New Jersey:
Atlantic, Cape
May & Ocean
Counties
64 Communities Part I crimes 1972–1984 Pre/post comparison Casinos in relation to crime
and population at risk
Crime was higher in post
casino years and
a spillover e
ffect occurred
in surrounding areas
Chang (1996) Mississippi Biloxi Part I crimes, other
crimes &
mischief
1986–1994 Pre/post comparison Measure the impact of
casinos on local crime
Decrease in overall crime
rates during the first
full year of casinos
opening; crime rates
returned to the pre-casino
level during the second
year
Grinols, Mustard, and Dilley
(2000)
United States 3,165 counties Part I crimes
(excluding
arson)
1977–1996 Correlational
study;
post-test
only
Determine the relationship
between casinos and crime
Casinos linked to higher
crime except murder;
crime increased beginning
about 3 years after the
casino opened
Gazel, Rickman, and
Thompson (2001)
Wisconsin all counties Part I & other
crimes
1981–1994 Pre/post comparison Examine the relationship
between Native American
casinos and crime levels Counties with casinos
showed an increase in
crime rates; there was also
a spillover e
ffect with
counties adjacent to
casino-counties (higher
crime rates)
(Continued)
CRIMINAL JUSTICE STUDIES 143
Table 1. (Continued).
Citation Location Focus Crime Time Method Purpose/Issue Finding
Stitt et al. (2003) Iowa, Missouri,
Illinois, &
Mississippi
8 cities Part I & II crimes 1987–1998
Min. 4 yrs.
pre & post
casino
opening
Pre/post comparison Effects of new casinos on
crime and the quality of
life in the area
Some communities
experienced an increase in
crime, others a reduction,
and some remained the
same: overall, crime rates
do not increase
Moufakkir (2005) Michigan City of Detroit & 3
counties
Part I, other
crimes, &
disorderly
conduct
1996–2002 Pre/during/
post
comparison Examine crime volume in
Detroit and its
neighboring communities
Volume of crime did not
materially increase when
the 3 casinos opened
Grinols and Mustard (2006) United States 3,165 U.S. Counties Part I crimes 1977–1996 Pre/post comparison Examines how the opening
of new casinos affect
crime The effect on crime is low
shortly after a casino
opens, then increased over
time
Barthe and Stitt (2007)1 Nevada 15 casinos in
downtown Reno
Part I crimes &
other crimes
2003 Correlational
study
Determine if casinos and their
surrounding blocks are hot
spots that generate crime
Almost 25% of Reno’s crime,
occurred within 1000 feet
of the major casinos but
factoring for population at
risk, casinos do not appear
to be ‘hot spots’ that
generate crime
Wheeler, Round, Sarre, O’Neil
(2007)
South Australia 111 local areas Income & nonincome- generating crimesb 2002–2003
(Continued)
144 V. SOSA ET AL.
Table 1. (Continued).
Citation Location Focus Crime Time Method Purpose/Issue Finding
Correlational study Compare
electronic
gaming
machine (EGM)
expenditures
to crime rates
Higher EGMs
expenditures
were significantly
related to
incomegenerating crime
rates but not
non-incomegenerating crime
rates
Barthe and Stitt (2009)
1 Nevada Reno Violent crimes,
property crimes,
and disorder
crimes
Not reported. Correlational
study
Compare casino and noncasino zones Temporal trends in casino zones are not very
di
fferent than those found
in non-casino areas
Walker (2010) 21 casino-crime
papers
locations
throughout the
United States
Overall crime rates
of some or all
Index I crimes
1985–2010 Systematic
review
Examines the relationship
between casinos and crime
Some studies suggested that
casinos cause crime, while
others argued that an
increase in crime was
more likely due to tourism
Belanger et al. (2012)1 Canada Alberta: Dene &
Eagle River (First
Nations
communities)
Casino-related
crime
2006–2010 Pre/post comparison Assess crime &
socioeconomic e
ffects on
rural communities
No reported increase in
criminal activity; residents
are aware of the financial
and social realities of
casino operations
Pontell et al. (2014) Macau & China Macau &
Hong Kong
White-collar &
economic
crimes
2009–2011 Case study Examine economic and whitecollar criminal activities The growing casino industry has created a receptive
environment for various
forms of corruption to
thrive
Falls and Thompson (2014)
1 Michigan 83 counties Robbery, burglary,
larceny & motor
vehicle theft
1994–2010 Correlational
study
Impact of casinos on crime
rates in the host and
neighboring counties
The presence or size of
a casino does not increase
property crime rates in the
host county or in the
nearby counties
1 Spatial aspect to the analysis of casino locations and crime. 2 Income-generating crimes: robbery and extortion, burglary, break and enter, fraud, forgery, false pretenses and larceny. Non-income-generating crimes: o
ffences against the person,
damage (property and environmental), o
ffences against good order, driving, and other o
ffences.
CRIMINAL JUSTICE STUDIES 145
qualitative data did not suggest that crime increased, the focus group participants did
identify an increase in drug activity in both communities.
The second take-away from the literature is that even though many of the studies
purport to examine the impact of casinos on the community, geographic precision is
missing from most analysis. Rather than investigating the immediate surrounds, the
general impact of casinos is assessed in relation to crime in large administrative units,
such as counties or cities. For instance, Falls and Thompson (2014) measured changes in
four types of crime in 83 Michigan counties and Moufakkir (2005) examined crime
volume before, during, and after a casino opened in Detroit, Michigan and 3 neighbouring counties. Again, the Belanger et al. (2012) study is the only noted exception. Rather
than using administrative units, these authors examined the impact of casinos within
45 minutes, or a 100 kilometer drive from the facility. When compared to the facilitybased research presented in studies focusing on liquor-serving establishments, a 100 km
radius is still exceptionally large and would not accurately capture facility-based crime –
a more reasonable radius is needed that accounts for the average facility size.
Casinos often house liquor serving venues. For this reason, we may be able to draw
upon the large body of research examining the effect of liquor serving establishments
on crime to determine an appropriate zone of impact. For instance, examining a range
of drinking places that prepare and serve alcoholic beverages for immediate consumption such as bars, nightclubs, and taverns, Groff (2011) found that drinking places were
associated with increased crime in the surrounding area that extended at least two
blocks (244 meters) but no more than three blocks (366 meters); and, using a 1,500 foot
radius in an urban area, Ratcliffe (2012) discovered that violence appeared to be highly
clustered within 85 feet (25.9 meters) of bars.
A notable feature of research on drinking establishments is the tendency to study
facilities located in dense urban environments. To the contrary, resort-style casinos can
be larger than a city block and situated on an expansive campus, located on a remote
parcel of land in a rural area; it is not uncommon for several bars and nightclubs to be
housed within a single facility; and, since patrons are likely to drive or be bussed to
a property, an impact radius that is set a ‘driving scale’ may be warranted. For these
reasons, we suggest using a 2-mile impact zone, as opposed to two blocks as is
commonly used in dense urban environments, to begin investigating the immediate
spillover effects of casinos that are not located on a walkable strip. While the studies
reviewed here helped us to think about the criminogenic capacity of some casinos, they
do not provide a framework for explaining the correlations observed. To explain why
high activity nodes are linked to crime we turn to crime pattern theory and the risky
facilities perspective.
High activity nodes and crime
Crime pattern theory
Crime pattern theory (CPT), as argued by Brantingham and Brantingham (2008), stipulates that crime does not occur randomly; instead, incidents are patterned in time and
space, occurring where the activity of the offender and victim (or target) intersect.
Though originally described as happening in physical space, the intersection of targets
and offenders also occurs in the digital domain, through the internet and various
146 V. SOSA ET AL.
electronic media (Brantingham & Brantingham, 2015). Since an individual’s activity is
shaped by daily routines, routine behavior contributes to the development of an
awareness space, including knowledge of the activity nodes and their surroundings
(Brantingham & Brantingham, 1981, 2008). Home, work, school, shopping areas, and
recreation areas are often listed as primary activity nodes, but activity nodes may also
include e-commerce websites, gaming sites, and various electronic forums, such as
social media. When activity nodes are common to many people, they function as hubs,
where the lives of different people intersect e.g., a local bar or Tinder® could function
equally well as recreational dating hubs.
Arguing that most people are situated within a network of family, friends and acquaintances who influence their behavior, Brantingham and Brantingham (2008) assert that
interactions with others also shape behavioral patterns, thereby exposing offenders to
opportunities and placing potential victims at risk. Social networks shape decisions about
what activities to participate in and where to conduct those activities because interactions
with others provides information. The social network also provides a reason to go to
specific places and engage in activities, often with others in the network. Since each
person is enmeshed within a dynamic set of family, friends, and acquaintances drawn from
work, school, hobbies, and the like, activity spaces will change as these relations evolve
(e.g., someone gets a new job).
One of the reasons why crime is not evenly dispersed is that the criminogenic
potential of places is not equal. For example, recreational hubs that offer entertainment
and activities to their surrounding communities can become local activity hubs linking
area residents and the local business community. Over time, successful properties will
become regional attractors, drawing people from a larger area, including tourists from
a great distance – business success causes the ‘hubness’ of the facility to expand. When
hubs are perceived to be target rich environments, crime problems may develop.
Crime generators and crime attractors are two types of locations where crimes tend
to concentrate. Crime generators produce high concentrations of people that enable
some individuals, who without any predetermined criminal motivation, act on crime
opportunities they encounter while on site. Crime attractors, on the other hand, are sites
that draw in motivated offenders because of the location’s already well-known criminal
opportunities; as a result, they become activity nodes for repeat offenders. For instance,
casino patrons often carry cash, and many casinos serve free alcohol, so patrons may be
less alert than usual: these interacting conditions could produce a target rich environment (Pontell, Fang, & Geis, 2014; Walker, 2008). The property would be classed as
a crime attractor if pickpockets and other predators frequent the location looking to take
advantage of the patrons who are intoxicated. Between the two, crime attractors are the
most criminogenic – as their reputation grows, they may evolve into a regional crime
problem, attracting even more motivated offenders from a larger area. Fortunately, not
all activity hubs become crime attractors or generators.
Risky facilities
Research shows that even among a single type of facility, crime concentrates at a small
number of properties (Eck, Clarke, & Guerette, 2007; Weisburd, 2015) and studies show
that aiming problem-solving initiatives at the subset of high-crime places stands to
generate the greatest reductions in crime (e.g., Braga, Papachristos, & Hureau, 2014). For
CRIMINAL JUSTICE STUDIES 147
example, examining crime problems occurring at motels in Chula Vista, CA, Schmerler
and colleagues discovered that only five of 27 properties exhibited high crime levels and
the high-crime properties were often located near low-crime properties (2009). An
evaluation of the Chula Vista Budget Motel Project showed that crime reduction strategies were more effective when aimed at properties hosting the most extreme problems
(Bichler, Schmerler, & Enriquez, 2013).
Efforts to identify the subset of facilities generating a disproportionate amount of
crime typically look for distributions approximating the 80–20 rule, targeting properties
(about 20%) associated with most of the crime, at times as much as 80 percent of the
problem behavior (Clarke & Eck, 2007; Eck et al., 2007). The question facing criminologists is not whether crime concentrates among a class of facility, rather, the question is
why certain facilities host more crime than others. High traffic facilities tend to be
associated with more crime, thus the volume of use can be a contributing factor to
the emergence of a problematic area (Wilcox & Eck, 2011). However, volume of patrons
is not the only factor contributing to high crime levels. Arguably, some of the observed
variation in criminogenic capacity of a set of properties is related to management and
property characteristics.
Extrapolating on the role of place managers, emerging evidence suggests that
differing facility characteristics, many of which exist due to decisions made by managers
and owners, play a role in the uneven distribution of crime (e.g., Eck, 1995; Felson, 2006;
Madensen, 2007; Madensen & Eck, 2008; Madensen & Sousa, 2008). Criminal opportunities can be influenced by everything from the physical layout of the internal and
external areas, to the presence of staff and security. For instance, with respect to casinos
this would include management decisions about the design and surveillance of parking
lots, the density and type of greenery used in landscaping, security presence at the
entrance and throughout the facility, as well as the layout and staffing of the gambling
floor. Thus, two casinos offering the same amenities and drawing similar crowds, may
have significantly different crime potential. A well-designed and managed property
should not experience the same degree of crime problems as a property overseen by
management that promotes illicit activities or permits crime problems to manifest: a
casino will become a crime attractor when management encourages, incorporates, or
tolerates illicit enterprises within their domain (Pontell et al., 2014). Individuals can learn
about management tendencies through direct experience, and indirectly, through their
contacts and media sources. Returning to the Brantinghams’ (2015) suggestion that we
need to better understand online activity and the role that interaction in this domain
plays in crime, we argue that an important source of information about business
characteristics is social media, specifically Yelp.com.
Yelping about a good time
Modern communication systems provide mechanisms for understanding how information from personal networks formed by linkages with family, friends, and acquaintances,
intertwines with social media. Individuals select places to visit based on their own
experience, as well as on recommendations from others. With the advent of Web 2.0
and the introduction of public websites with web-based search engines that allow users
to interact and communicate directly, i.e., Wikipedia, Facebook, etc., the experiences and
opinions of individuals well removed from our social network influence our behavior. For
148 V. SOSA ET AL.
example, a group of friends living in different communities select a city, equidistant from
each person, at which to meet for dinner. Someone nominates a restaurant recommended by a third party. Not knowing anyone in the city, and having no personal
experience, the rest consult Yelp.com. Based on anonymous reviews, the suggested
restaurant is approved and the group convenes. In this scenario, the final decision is
contingent upon information flowing through an electronic communications network,
not a personal network. This marks a shift toward electronic word-of-mouth information
exchange.
Word-of-mouth (WOM) communication refers to the ‘informal communications directed at other consumers about the ownership, usage or characteristics of particular goods
or their seller,’ (Westbrook, 1987). The digitization of WOM communication through
internet channels provides public information about business operations, such as statements about a product or company, transmitted via the internet that are made by
potential, actual, or former customers (Dellarocas, 2003; Hennig-Thurau, Gwinner, Walsh,
& Gremler, 2004). Of relevance to the present study, consumers have come to rely more
heavily on eWOM communication from strangers than purchased advertisements
(Hennig-Thurau et al., 2004; Steffes & Burgee, 2009).
Founded in 2004, Yelp.com is a prime example of eWOM communication. This forum
permits anyone to freely access reviews of businesses; however, to rate businesses on
a scale from 1 to 5, with 5 being the highest, individuals must register for a free account.
While Yelp allows members to rate an unlimited number of businesses, it restricts review
comments to a maximum of 5,000 characters (Tucker, 2011). Some information about
the reviewer is posted with the review, including pseudonym and city of residence (as
described by the reviewer).
Yelp prides itself in the validity of the reviews by claiming that no business can pay
any amount of money to change, alter, rearrange, or remove any unwanted reviews that
appear on their page. Notably, reviewers are not all equal. Termed ‘opinion leaders’,
some consumers have more influence on others’ decision-making (Lyons & Henderson,
2005). In 2012, Yelp developers introduced a mechanism to establish the caliber of each
review posted. To achieve Elite status, a person must frequently provide good quality
reviews. Good quality reviews are those that are rated as useful by other consumers.
Individuals who are actively posting good quality reviews can submit an application to
the Elite Council; otherwise, another member of the Yelp community could nominate
them. Preliminary research confirms that Elite status is shown to be an indicator of an
opinion leader. For example, Tucker (2011) examined 106 reviews to explore the
correlation between the characteristics of reviews and star ratings, concluding that
fellow consumers ranked the validity of Yelp reviews based on reviewers’ Yelp status.
To ensure information is current, algorithms inspect millions of reviews submitted
each day. The algorithms decipher which reviews to post with a formula that considers
the overall review quality, and reliability of the post, as well as the reviewer’s activity and
status in the Yelp community. As a result, Yelp only posts about three-quarters of the
reviews they receive which are predominantly selected from the more active users (Yelp.
com, n.d.). With increased accessibility through smartphone technology, it is not surprising that by the end of the third quarter in 2018 ‘yelpers’ had written more than
171 million reviews (Yelp.com, n.d.).
CRIMINAL JUSTICE STUDIES 149
With so many members of the public visiting and writing these reviews, Franquez, Hagala,
Lim, and Bichler (2013) investigated the utility of Yelp.com and showed that comments could
be used to gauge the criminogenic capacity of facilities. As will be discussed shortly, we
extend this argument by suggesting that since reviewers report a city of residence, it is
possible to calibrate the magnetism or attractiveness of a property with metrics available
through social network analysis (Bichler, Malm, & Enriquez, 2014).
Current study
The current study advances casino crime research in two ways. First, while it is important
to capture the local impact of criminogenic properties, current social trends and the
expanding influence of eWOM highlight the importance of online communities. Drawing
upon Brantingham and Brantingham’s argument, we must rethink how we conceptualize routine activities, and the factors influencing offender and victim behavior (2015).
Extending the work of Franquez et al. (2013), we use Yelp.com to obtain information
about facility management and site characteristics. Online communities have the capacity to overcome geographic boundaries, by sharing information with a much broader
user base. As a consequence, star ratings and comments about site characteristics
generate a wealth of public information about properties. Harvesting from Yelp.com
provides an opportunity to study site characteristics in a new way. It is hypothesized that
property magnetism, Yelp ratings, and site characteristics will be significantly related to Yelp
posts about onsite crime and deviance.
Second, we seek to build on the work of Barthe and Stitt (2007, 2009) and Belanger
et al. (2012), by continuing to examine the local crime impact of risky-facilities, rather
than the broader social and economic impact of gaming. Using a radius of 2 miles, this
study investigates which property characteristics are most associated with high-crime
casinos. In addition, by linking reviewers’ home city to the property site, it is possible to
develop a standardized metric gauging property magnetism which could be used to
examine the hubness of facilities in relation to crime. Extending the work of Bichler et al.
(2014) we use social network metrics to calibrate the draw of an activity node. In doing
so, we advance inquiry into the utility of a tenet of crime pattern theory: investigating
casinos through a CPT/risky facilities framework provides a theoretically-driven argument for exploring the distribution in criminogenic capacity of properties and classification of activity nodes. It is hypothesized that facility magnetism will be a significant
predictor of observed crime in the vicinity of the casino.
Methods
Sample
Currently, there are 29 operational casinos within the five-county study region – Los
Angeles, Riverside, San Bernardino, San Diego, and Ventura counties.1 The majority of
the casinos are located on reservation land and are owned or operated by local tribes (see
Table 2). The facilities vary greatly with regard to amenities and services offered to patrons.
Properties range from small gaming sites to those offering a complete resort experience.
150 V. SOSA ET AL.
Many properties offer multifaceted entertainment: 59% of the casinos have their own
music venue; 28% have at least one nightclub; and 97% have a bar. Of interest, 52% of
the properties include onsite hotels and 52% have a multilevel parking structure. Missing
data reduced the sample used in the multivariate models to 19 casinos. While we will
elaborate on this point later, the final two columns of Table 2 suggest that sample attrition
did not compromise the representativeness of the properties examined.
Variables
Observed crime
Since properties were located in different cities distributed across five counties, we used
the most viable, publicly available crime data – Crimemapping.com.2 Using a search
radius of 2 miles, we extracted all listed incidents occurring at or near each property
from 08/10/2015 to 11/15/2015. We used a 2-mile buffer to ensure that the catchment
included all areas under the facility’s control, as well as adjacent land. For instance,
37.9% of the casinos are located in rural areas, and while the average size of these
buildings was small, only 185,866 square feet, some properties had golf courses and
other large amenities. Adding parking, landscaped areas, and access roads, significantly
extends the facility size. Admittedly, this is a limitation of the study, as there was a lot of
variation within the sample. Properties were also located in urban areas (24.1%), industrial zones (24.1%), and residential neighborhoods (13.8%).
While 6 months of data were available, we set the boundaries of the observation
period to three months to avoid season fluctuations caused by summer and winter
holidays. We classified crime into two categories: violent crime and non-violent crime.
Violent crimes included: assault, homicide, robbery, and sex crimes including rape. Nonviolent crimes included 11 types of crime – arson, disturbing the peace, drugs and/or
alcohol, driving under the influence, motor vehicle theft, theft/larceny, vandalism, vehicle break in, and other. On average, the impact zones around casinos hosted 42 violent
crimes (SD 75) and 139 non-violent crimes (SD 261) during the three months observed
(see Table 3). We converted all crime counts to z-scores to explore the effects of being
above the mean level of observed crime.
Table 2. Casinos characteristics.
Study Population
(N = 29)
Multivariate Model Sample
(N = 19)
Facility Characteristics Frequency Percent Frequency Percent
Tribal Ownership or Management 20 69% 10 53%
Age Requirement
18 & Over 5 17% 4 21%
21 & Over 24 83% 15 79%
Property Amenities
Concert Music Venue 17 59% 9 47%
At least 1 Nightclub 8 28% 2 11%
At least 1 Bar 28 97% 18 95%
Hotel 15 52% 10 53%
Pool 11 38% 7 37%
Parking Structure 15 52% 10 53%
CRIMINAL JUSTICE STUDIES 151
Facility characteristics
Magnetism. One way to gauge the degree to which a property is a crime generator or
attractor is to assess its drawing power or magnetism. Coding magnetism involved a twostep process. First, we captured information about the home or residential city listed by
reviewers posting on Yelp. For each review, we generated the Euclidean distance in miles
between the location of the casino and the centroid of the reviewers’ residential city, and
then, we recoded distances to reduce the extreme variability: 1 = lives in the city where the
casino is located, 2 = lives in a city located up to 10 miles away from the casino, 3 = 11–-
20 miles, . . ., 16 = 141 miles and farther. Since tourists from abroad or other states would
unduly inflate the magnetic pull of properties, reviews from people living outside of
Southern California were absorbed in the final category. We discovered that the 29 casinos
were visited by residents of 523 different cities.
The second step involved using the distance codes to weight the calculation of
indegree centrality. To begin, we generated networks for each property by linking each
casino to the residential city of each reviewer, weighting the link by the distance scale
described above. Linking residential city to casinos generated 1,632 unique ties (3,334
total ties). The network was a single component as people from the same residential
city frequented different casinos. Then, we calculated indegree centrality for each
casino.3
Indegree centrality is a conventional social network metric that indicates how many
connections one node or actor receives from other nodes or actors in the network. The
basic idea is that casinos that receive Yelp reviews from people living in a greater variety
of cities are characterized as being more attractive. Meaning, a casino drawing reviewers
from five cities is more attractive than a casino with reviews from residents of 2 cities. In
our study, however, we used a valued score, which adds a bit of a complication. We
valued each link by a scaled distance. This means that an indegree centrality score of 26
could mean that at least 26 locals yelped about a casino located in their residential city,
Table 3. Descriptive statistics for study variables.
Variables N Mean SD Min. – Max.
Observed Crime in the Vicinity
Violent Crimea 19 41.70 74.68 0–307
Non-violent Crimea 19 139.13 260.98 0–1,206
Facility Characteristics
Magnetism 29 966.52 1,037.72 0–3,433
Amenities
Hotel 29 0.52 0.51 0–1
Nightclub 29 0.28 0.45 0–1
Slot Machines 29 0.69 0.47 0–1
Resort Property 29 0.41 0.50 0–1
YELP Variables
Star Rating 29 3.05 .50 2.00–4.50
Reported Onsite Crime and Deviance Problemsb 29 13.99 19.70 0–87.50
Gambling Commentsb 29 45.61 24.85 8.57–83.64
Food Commentsb 29 43.88 19.80 8.57–83.33
Bar Commentsb 29 10.73 7.27 0–28.85
Rudeness of Staffb 29 13.28 8.96 0–41.67
Friendliness of Staffb 29 22.99 17.14 4.17–100.00
a Violent Crime Z scores range from −0.56 to 3.55; Non-violent crime Z scores range from −0.53 to 4.09. b
All comments rates are calculated per 100 reviewers.
152 V. SOSA ET AL.
or 13 yelpers traveled up to 10 miles to get to the casino, and so on. There are different
permutations which could sum to 26.
Considering the distance weighting, not only are high scoring properties more attractive
but they draw from a wider geographic area, suggesting they are more magnetic (Bichler
et al., 2014). As expected, not all casinos were equal – some had greater magnetism than
others. For example, people from 96 cities posted comments about their visit to Barona
Casino, the highest scoring facility (as a point of reference the least magnetic property had
a score of 8). San Diego residents contributed the most to Barona’s overall indegree
centrality score of 2,309. Drilling down to the strongest city-casino link, a score of 508,
tells us that, 127 reviewers traveled from San Diego to enjoy themselves at Barona –
a distance of less than 30 miles, centroid to centroid. Of note, the average indegree
centrality score was 966.52 (SD 1,037.72).
Amenities. Casinos offer patrons a range of gambling and entertainment services, such
as nightclubs, event centers, and hotels; and the presence of specific amenities may
contribute to, or mitigate, crime and public safety issues. Information about four
amenities – hotel onsite, nightclub, slot machines, and resort-style facility – was
obtained from casino websites (coded in November 2015). If the website indicated
that any of the previously mentioned amenities were present, we assigned a value of 1,
otherwise a value of 0 was used. Two amenities were common among casinos: 52% of
the casinos had a hotel onsite and 69% had slot machines.
Yelp Variables. As discussed earlier, Yelp.com is a public website developed to enable
consumers to share their reviews of businesses. With few prior studies to guide our use
of Yelp ratings, we generated two sets of variables. First, reviewers rank properties with
a 5 star-rating system (5 being the highest possible score) and provide a brief description or comment of why that business earned the star rating. To measure star rating, we
averaged ratings for each property. Second, we captured information about site conditions with eight dichotomous variables from the comments about: crime and deviance
problems (this is a measure of onsite reported crime and deviance), food quality or
availability, the main gambling floor, onsite hotel, resort property, alcohol outlets (bars,
nightclubs, event venues, etc.), and instances of good and rude customer service. If the
reviewer commented on any of the above, we assigned a value of 1, else 0. In total, we
extracted information from 3,336 Yelp reviews submitted from 1 January 2013 to
11 November 2015. Since reviews were not distributed evenly among properties, we
converted sums for each condition into a rate per 100 posts. Reviewers more frequently
commented about three issues – gambling, food, and friendliness of staff.
Results and discussion
Reported crime – bivariate analysis
Modernizing our conceptualization of CPT and the risky facilities debate to include
eWOM, we asserted that property magnetism, Yelp star ratings, and site characteristics would be significantly related to onsite crime and deviance issues. Facilities
garnering high rates of Yelp posts commenting on instances of crime or deviance are
CRIMINAL JUSTICE STUDIES 153
defined for the purposes of this study as having bad reputations that may attract
more crime – crime attractors. Table 4 reports the results of the exploration into the
viability of using Yelp reports of crime and deviance to study the criminogenic
characteristics of risky facilities. In addition, to validate this investigation we explored
whether Yelp posts about onsite problems performed similarly to crime observed
within a 2-mile impact zone: these correlations are also reported in the table. Three
sets of correlations are reported because the Pearson Correlation Coefficients
between reported crime and deviance (Yelp posts indicating a bad reputation) and
observed crime in the impact zone were less than perfect (Pearson with violent
crime = .476; p <.01; Pearson with non-violent crime = .219; n.s.), indicating each
measure captured unique phenomena.4
Several findings are noteworthy. First, higher property magnetism was significantly
associated with fewer reported crime and deviance problems: meaning that properties
attracting more Yelpers from a greater distance (larger hubness) were associated with
fewer posts about onsite problems. In other words, higher magnetism is associated
with better reputations. Second, two amenities stood out as being significantly associated with lower reported problems – having slot machines and being a resort property. This means that generally speaking, larger Class III gaming facilities draw less
negative social media attention regarding crime and deviance problems. Third, with
one exception, all Yelp variables were significantly associated with crime and deviance
posts. Of note, star ratings were significantly lower when problems were reported
(Pearson = −.629; p <.01); and, perceived staff rudeness was strongly correlated with
reported problems (Pearson = .828; p <.01).
Finally, despite the lack of strong correlations between reported onsite crime and
deviance (Yelp comments) and observed crime occurring within 2 miles of facilities
(observed crime recorded by law enforcement), all three crime measures preformed
similarly in terms of the overall strength and direction of correlations, with one exception.
Higher Yelp star ratings were strongly associated with lower reported crime; whereas
Table 4. Pearson correlation coefficients of facility characteristics and crime.
Reported Crime/Deviance
(onsite)
(N = 29)
Observed Crime Problems
(2-mile impact zone)
(N = 19)
Facility Characteristics
Problems p/100
Yelp posts Violent Crime Non-violent Crime
Magnetism −.355* −.269 −.258
Amenities
Hotel −.284^ −.347* −.329*
Nightclub −.090 −.190 −.161
Slot Machines −.462** −.663** −.581**
Resort Property −.445** −.458** −.432*
Yelp Variables
Star Rating −.629** −.164 .157
Gambling Yelps p/100 posting .253^ .300 .176
Food Yelps p/100 posting −.353* −.093 .068
Bar Yelps p/100 posting .035 .279 .402*
Staff Rudeness p/100 posting .828** .431** .092
Staff Friendliness p/100 posting −.287^ .311* .609**
Note: Significance levels for the 1-tailed tests are as follows: ^ indicates p < .10; * indicates p < .05; and
** indicates p < .01.
154 V. SOSA ET AL.
ratings were only weakly related to observed crime (not significant) and in unexpected
directions – the association was positive for non-violent crime. Based on these results we
tentatively assert that Yelp may be a viable public source of information about the
differential characteristics of crime-prone risky facilities.
Observed crime – multivariate analysis
To further explore the casino-crime nexus, three sets of OLS regression models are
reported in Table 5. 5 Each set includes a model to account for observed violent crime
occurring at or in the vicinity of casinos and a separate model investigating non-violent
crime. The first set of models examines the explanatory influence of amenities controlling for property magnetism. Variables were entered as a block. Model set 2, the Yelp
models, investigate the relative importance of star ratings and the nature of comments
when accounting for violent and non-violent crime. Again, variables were entered as
a block. Significant or substantively important variables (since we have a small sample
size and significance is elusive) from model sets 1 and 2 were retained for the final,
parsimonious models.
Adjusted R2 values are reported because missing data reduced the sample to 19
properties, thereby reducing our power and restricting the number of explanatory
variables we could enter into each model. Given this limitation, we tested each block
of variables with magnetism included to account for property attractiveness. The final
set of results present the most parsimonious models which included the best performing variables from each block.
The amenities models revealed that three of the four types of amenities were
significantly related to lower crime near casinos. Notably, the presence of slot machines
was the most influential factor associated with lower violent and non-violent crime, with
a standardized beta coefficient that was more than double that of having a hotel onsite
for violent crime and more than triple that of being a resort property for both crime
types.
Examining the Yelp models, we found that the variation in violent and non-violent
crime was significantly related to reviews commenting about staff friendliness and
comments about crime and deviance problems. Comparatively, staff friendliness was
a more important predictor of higher violent and non-violent crime. Of note, a high star
rating was associated with a reduction in both violent and non-violent crime, but this
association failed to reach significance with the small sample.
The parsimonious models perform the best, accounting for about 53% of the variation found in violent crime (adjusted R2 = .533, F (6, 18) = 6.333, p <.001) and nearly 55%
of the variance in non-violent crime (adjusted R2 = .548, F (6, 18) = 6.661, p <.000). These
models include the best performing variables found in each of the other models – the
small sample size limited the number of explanatory variables that could be included in
each analysis. Overall, holding all other factors constant, three variables best account for
varying crime levels at or within the vicinity of casinos in Southern California – (1)
comments about friendly staff are associated with higher crime, (2) the presence of slot
machines dampens crime problems, and (3) property magnetism is linked to higher
violence.
CRIMINAL JUSTICE STUDIES 155
Table 5. Three sets of OLS regression models.
Facility Characteristics
Amenities Model Yelp Model Parsimonious Model
Violent Crime Non-violent Crime Violent Crime Non-violent Crime Violent Crime Non-violent Crime
STD Beta (S.E.) STD Beta (S.E.) STD Beta (S.E.) STD Beta (S.E.) STD Beta (S.E.) STD Beta (S.E.)
Magnetism (indegree centrality valued by distance) .283 (.000) .229 (.000) .191 (.000) .183 (.000) .331^ (.000) .278 (.000)
Amenities
Hotel −.326^ (.366) −.301 (.406) −.207 (.341) −.163 (.336)
Nightclub .042 (.377) .051 (.418)
Slot Machines −.662** (.405) −.556** (.449) −.497** (.387) −.368* (.381)
Resort Property −.143 (.404) −.158 (.448)
Yelp Variables
Star Rating −.407 (.610) −.347 (.564) −.377 (.486) −.271 (.478)
Gambling Yelps (p/100 posting) .143 (.008) .029 (.007)
Food Yelps (p/100 posting) .091 (.010) .080 (.009)
Bar Yelps (p/100 posting) .074 (.029) .167 (.026)
Crime & Deviance Problems Noted (p/100 posting) .491* (.011) .341^ (.010) .220 (.010) .152 (.010)
Staff Friendliness (p/100 posting) .729** (.015) .903** (.014) .528* (.013) .767** (.013)
Model Fit
R2/Adjusted R2 .528/.425 .419/.293 .522/.363 .592/.456 .633/.533 .645/.548
F 5.147 3.316 3.283 4.348 6.333 6.661
Significance .003 .021 .016 .004 .001 .000
Notes: These models have an N = 19 and ^ p < .10, *p < .05; **p < .01
156 V. SOSA ET AL.
Discussion
Yelp ratings
Results for the parsimonious models indicate that patron experiences recorded on
public forums (eWOM), used in tandem with information about amenities, contribute
to our ability to understand patterns of violent and non-violent crime occurring at or in
the vicinity of casinos. The performance of variables created from Yelp comments was
interesting. As expected, higher star ratings were generally associated with less crime.
Although not significant in the final model, the effects were large enough to be
noteworthy. At first glance, our findings support our hypothesis, but they do not concur
with prior research. For instance, Franquez et al. (2013) found that highly rated bars had
significantly more problems. Our study found the opposite, higher casino ratings were
associated with less crime. We argue that this discrepancy is actually evidence of the
discriminant validity of the star rating. Despite the fact that both types of commercial
properties cater to the recreational needs of adults, the star ratings for bars and nightclubs might tap into something different than casino ratings. For instance, the objectives
of the clientele may vary by facility type, and this may be reflected in how people rate
properties. When going to a bar, excessive pouring may constitute exceptional service
warranting a high rating, but this ‘star quality’ might lead to greater intoxication and
associated crime issues; whereas high casino ratings may be linked to the overall patron
experience based on value for money of the entertainment and facility services. As such,
we argue that Yelp ratings may be useful for facility-based research, but not for mixedfacility research. In addition, the richness of comments may also provide a viable source
of publicly available information about the management of commercial facilities.
Place management
While Yelp reviews commenting on crime and deviance problems performed as
expected, albeit poorly, casino staff friendliness was the most significant and perplexing
factor associated with higher crime. Reexamining the original text associated with
a ‘friendly’ rating we found two potential explanations. (1) Reviewers were more inclined
to comment on the friendliness of staff when they were experiencing something good,
such as winning, cheap or strong drinks served quickly, excellent food, or a short line at
check-in. (2) ‘Friendly’ ratings were also associated with how the staff made guests feel
through various actions, such as having immediate contact with the guest upon arrival
or entry into a room, being approachable, acknowledging and remembering the patron,
and having a sense of humor. Given the link with higher crime, it is possible that within
the context of a gambling facility, we tentatively speculate that overly friendly staff/
patron interactions may lead patrons to feel so comfortable that they let their guard
down to potential offenders. Future research is needed to determine whether the
finding is robust and whether our speculation is sound. Meanwhile, this finding is of
great interest as many of the site conditions captured by this measure represent aspects
of good place management, prompting us to ask whether it is more useful to think of
place management as a continuous phenomenon, as opposed to something that can be
measured as a categorical variable.
Madensen and colleagues (Madensen, 2007; Madensen & Eck, 2008; Madensen & Sousa,
2008) argue that management style is a critical facet of the criminogenic capacity of
CRIMINAL JUSTICE STUDIES 157
places. Management makes decisions about the use of space, maintenance, and choice of
activities permitted on the property; they establish behavioral norms, set goals and
policies; and, they control access to the facility and direct the use of resources. Using
a two-by-two matrix, these authors suggest a typology of four management styles, each
with varying criminogenic capacity. Active, engaged managers with a proclivity to proactively eliminate crime opportunities are classified as suppressors, in contrast, promoters
are managers who recognize and exploit crime opportunities for personal or business
advantages. Passive managers who modify behavior or the environment following crime
incidents are referred to as reactors, whereas enablers describe place managers who fail to
recognize or respond to problems (Madensen & Sousa, 2008).
Management has many facets. Given all of the elements that constitute effective place
management, maybe the management styles can be operationalized with a set of indices.
At a minimum, casino-crime studies should use an indices to rate cleanliness and facility
appearance, entry and exit screening practices, alcohol and drug controls, and patron/staff
engagement. Using several indices would permit greater exploration of the association
between management styles and crime. For instance, considering patron/staff engagement, which seems to be the root of what our measure captures, there may be
a threshold, beyond which increasing the number of friendly, positive interactions
becomes counter-productive. There may be a point at which patrons become overly
comfortable and relax their use of routine crime precautions, i.e., leaving valuables
unattended, being overly friendly with strangers, and drinking and gambling alone after
friends or family go to bed.
Turning to the amenities that management supports onsite, slot machines were associated with significantly less violent and non-violent crime. It is plausible that this can be
explained by the legal and practical requirements of managing a Class III gaming floor. The
gaming floor becomes more complex with slot machines due to their size and floor pattern
layout – slot machines necessitate increased security measures, such as the number of
casino employees required to monitor the floor. Slot machine attendants work in close
proximity to machine users, providing a high level of surveillance that may deter victimization of patrons, while fulfilling other non-crime related management functions (Austrin &
West, 2005). Greater supervision may play a role in reducing crime in the vicinity of casinos
as the stronger management presence may disrupt precursor conditions, e.g., alcohol
consumption can be better monitored to prevent over intoxication and people playing
slots might win smaller amounts thereby reducing the number of people leaving the
property with large sums of money. Where facilities lack the resources to heavily staff
the casino floor, highly visible security cameras placed throughout the property may offer
dual-functions – deterring crime and satisfying the stipulations of regulatory oversight
(Kruegle, 2007). Camera systems can be placed indoors and outdoors, thereby increasing
surveillance in the areas immediately adjacent to the facility where vehicles are parked and
significantly extending the monitoring capacity of security personnel.
Property attractiveness
The findings pertaining to casino magnetism suggest that we may have developed
a way of differentiating between crime generators and attractors. Magnetism varied
widely, and when used in tandem with a criminogenic Yelp reputation, it may be
possible to differentiate properties likely to draw more motivated offenders. Our
158 V. SOSA ET AL.
findings support this supposition. At the bivariate level, magnetic properties were
associated with significantly fewer problems – as measured by reports of onsite crime
and deviance. However, holding all other factors constant, including reported onsite
issues, greater magnetism was significantly associated with higher crime rates in the
vicinity of the property. Meaning, that casinos attracting more patrons from a greater
distance expose the surrounding community to higher levels of opportunistic crime,
specifically, violent crime. Where magnetism is high, and there is a high reporting of
onsite crime and deviance, the surrounding community is expected to experience even
higher levels of violent and non-violent crimes – such a property would be classed
a crime attractor. Exploring our findings further, facilities with high magnetism, but low
reported onsite crime problems, may constitute crime generators – theses properties
have good reputations but a high level of hubness. These findings are consistent with
tenets of CPT. Crimes can concentrate where a greater number of offenders and victims
interact in the same target rich environments; however, overlapping activity space is
a necessary precursor to crime, but is not sufficient to account for differential crime
levels. Instead, it may be possible to differentiate crime attractors or crime generators
incorporating information about the public reputation of the property.
Why is this important? Because crime control responses should differ depending on
property classification. Crime generators have many unprotected targets, so property
management would need to invoke strategies that increase the likelihood that people
would engage in routine precautions (Felson & Clarke, 2010). To determine the best
preventative solutions future research is needed to identify the circumstances that
make targets vulnerable, i.e., time of day, location, etc., known as spatio-temporal
characteristics of hotspots (Ratcliffe, 2004). Regarding crime attractors, since facilities
draw offenders, crime control might focus more on discouraging offenders from going
to the location (Clarke & Eck, 2005). Offender interviews may be needed to determine
why the offenders are attracted to the location and how the situation could be
remedied.
Impact zone
Using a 2-mile buffer around casinos to measure the impact zone around facilities, raises
the issue of transition zones between land-uses. Edges between different types of areas
may experience high crime rates due to the mixture of physical features associated with
the connecting land types which may provide ideal criminal opportunities (Brantingham
& Brantingham, 1993). The crimes could be in part due to the 24-hour nature of casinos
which provide potential offenders access to nearby urban neighborhoods and adjacent
businesses at all hours of the day. Jane Jacobs (1961) drew attention to boundaries and
divisions separating urban neighborhoods from commercial centers by emphasizing the
importance of localities having several functions so that there is a constant influx of
diverse people on the streets throughout the day. She alleged that crime will not be
a viable option if there are enough watchful people, able to instil a feeling of collective
security.
While the 2-mile zone proved useful in this study, advancing this line of inquiry will
require exploration of alternate operationalization of impact zones to better capture
spillover effects. We argue the type of facility and land use should be used to determine
the buffer size best suited to capture appropriate crime data. Businesses located in or
CRIMINAL JUSTICE STUDIES 159
near urban centers may not require very large buffers, whereas, properties located on
larger tracts of land in suburban and rural environments may require a substantially
larger distance buffer (e.g., 2 miles) in order to encapsulate associated crimes. For
instance, Ratcliffe’s (2012) finding that violence is highest 85 feet from a bar is instrumental for thinking about crime within a Las Vegas styled casino, but this operationalization would not work when investigating community impacts. Alternatively, Belanger
et al.’s (2012) use of a large community-based impact zone (100 km) is too vast for urban
locations. In the future, to better approximate impact zones, studies should consider the
type of facility and land use in the immediate vicinity.
Study limitations
Several study limitations must be acknowledged – the quality and quantity of information
posted to Crimemapping.com, the use of a 2-mile radius, and the lack of patron information. First, drawing crime data from Crimemapping.com limited the study in two ways. (1)
Lack of reporting for casinos located in Riverside County, specifically the Coachella Valley
not only reduced our sample size by 10 but also left a region comprised entirely of Indian
gaming casinos underrepresented in the final analysis. (2) This source provides few details
about the incident (restricted to temporal aspects and general crime type) and limited
geographic precision (street block level), thereby prohibiting detailed study of the context
of crime incidents. The second limitation of the study is that using a 2-mile radius
prohibited us from differentiating between crimes occurring on the casino grounds, as
opposed to crimes happening in the neighboring communities. As noted above, all crime
occurring at or near each property was lumped together. Future research should investigate the difference by partitioning the data. Finally, a true measure of patronage would be
useful to better model target richness/opportunity. For starters, conducting interviews with
casino staff would reveal context to better understand activity that is going on from their
standpoint. Similarly, patron interviews would provide invaluable insight into their perceptions of what is occurring, and this information would better inform crime prevention and
policing efforts (Haberman, Groff, Ratcliffe, & Sorg, 2015).
Conclusion
This study investigated covariates that might account for crime occurring within the
vicinity of Southern California casinos using only publicly available resources. Using data
extracted from Crimemapping.com and Yelp.com, our study was able to determine that
casinos that have slot machines were associated with less crime. Also, casinos with more
Yelp.com reviews that included comments about staff friendliness were associated with
higher crime; whereas higher star ratings were associated with lower crime. Lastly,
property magnetism (measured with indegree centrality calculated on a travel network)
proved valuable in detecting high crime facilities and demonstrated how this standardized measure, used in conjunction with public posts about onsite crime and deviance,
could be used to differentiate types of crime hotspots – crime generators and attractors.
And, while further research is needed to better understand casino-related crime from
160 V. SOSA ET AL.
the perspective of staff and community members, we can also learn a lot about the
criminogenic capacity of places from people yelping about a good time.
Notes
1. A list of properties is available upon request.
2. What is the alternative to crimemapping.com? At the time this study was conducted, there
were none. Obtaining data from 29 different agencies or divisions, since each casino was
located in a different jurisdiction, was untenable in this region. The difficulty associated with
getting facilities-based data from law enforcement agencies in Southern California was
documented by Marteache, Jimenez, and Lizarraga (2018) – when 50 agencies were asked
for crime incidents occurring at and around commuter train stations, only two agencies
provided data. One of the most frequent responses from agencies was that the data was
posted to crimemapping.com, so the information was already available.
3. Notably, casino size in square feet and indegree centrality were highly correlated (Pearson
.840; p < .01), suggesting that larger facilities draw more clientele from an expansive area.
Notably, the correlation is not perfect suggesting that indegree centrality captures something unique. Since our measure of indegree centrality captures a more complex concept,
we retained it for the analysis.
4. Two additional inter-item correlations are reportable. Staff rudeness was inversely correlated
with star rating (Pearson = −.710; p < .01): fearing multicollinearity staff rudeness was
dropped for the multivariate model. Additionally, the presence of onsite hotels was correlated with indegree centrality (Pearson = .609; p < .01); meaning that casinos with a hotel
attract clientele from greater distances. Both variables were retained because the correlation was not large enough to be prohibitive.
5. With no prior regression-based findings regarding the intersection of casinos-crime in
multiple cities, this study sought to begin the investigation with simple OLS models.
From this foundation, subsequent research will be able to assess the applicability of
Poisson or negative binomial regression.
Acknowledgments
The author(s) presented a version of this manuscript at the annual meeting of the Western Society
of Criminology (WSC), February 2018, Long Beach, CA. They would like to thank Melanie Aguayo,
Kayla Arroyo, Stephanie Castro, and Andres Serrato for their assistance with data collection and
helpful comments on earlier drafts.
Disclosure statement
No potential conflict of interest was reported by the authors.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this
manuscript.
Notes on contributors
Virginia Sosa, BA, is a research assistant in the Center for Criminal Justice Research. Her academic
area of specialization is crime analysis.
CRIMINAL JUSTICE STUDIES 161
Gisela Bichler, PhD, is professor of criminal justice at California State University, San Bernardino.
Her recent scholarship explores the interplay between the environment and offending behavior,
with an emphasis on the influence of social networks. Recent publications include Journal of
Research in Crime and Delinquency, Crime and Delinquency, Crime Patterns and Analysis, Global
Crime, and the Security Journal.
Lianna Quintero, BA, is a research assistant in the Center for Criminal Justice Research. Her
academic area of specialization is crime analysis.
References
Albanese, J.S. (1985). Effect of casino gambling on crime. Fed. Probation, 49, 39.
Austrin, T., & West, J. (2005). Skills and surveillance in casino gaming. Work, Employment and
Society, 19(2), 305–326.
Barthe, E., & Stitt, B.G. (2007). Casinos as “hot spots” and the generation of crime. Journal of Crime
and Justice, 30(2), 115–140.
Barthe, E., & Stitt, B.G. (2009). Temporal distributions of crime and disorder in casino and
non-casino zones. Journal of Gambling Studies, 25(2), 139–152.
Belanger, Y.D., Williams, R.J., & Arthur, J.N. (2012). Assessing the impact of the introduction of
casinos in two northern alberta first nation communities. American Review of Canadian Studies,
42(1), 1–19.
Bichler, G., Malm, A., & Enriquez, J. (2014). Magnetic facilities: identifying convergence settings of
juvenile delinquents. Crime & Delinquency, 60(7), 971–998.
Bichler, G., Schmerler, K., & Enriquez, J. (2013). Curbing nuisance motels: an evaluation of police as
place regulators. Policing: an International Journal of Police Strategies & Management, 36(2),
437–462.
Braga, A.A., Papachristos, A.V., & Hureau, D.M. (2014). The effects of hot spots policing on crime: an
updated systematic review and meta-analysis. Justice Quarterly, 31(4), 633–663.
Brantingham, P.J., & Brantingham, P.L. (1981). Environmental criminology. Beverly Hills: Sage
Publications.
Brantingham, P.J., & Brantingham, P.L. (2008). Crime pattern theory. In R. Wortley & L. Mazerolle
(Eds.), Environmental criminology and crime analysis (pp. 78–93). UK: Willan.
Brantingham, P.L., & Brantingham, P.J. (1993). Nodes, paths and edges: Considerations on the
complexity of crime and the physical environment. Journal of Environmental Psychology, 13(1),
3–28.
Brantingham, P.L., & Brantingham, P.J. (2015). Understanding crime with computational criminology. In M.A. Andresen & G. Farrell (Eds.), The criminal act: the role and influence of routine activity
theory (pp. 131–145). Basingstoke, UK: Palgrave Macmillan.
CBS News. (2017, June 7). Man leaving gardena casino attacked, shot by robbers

.
Retrieved from: https://www.youtube.com/watch?v=eqRABEH8RzY).
Chang, S. (1996). Impact of casinos on crime: The case of Biloxi, Mississippi. Journal of Criminal
Justice, 24(5), 431–436.
Clarke, R.V., & Eck, J. (2005). Crime analysis for problem solvers in 60 small steps. Washington, DC:
Office of Community Oriented Policing Services, United States Department of Justice.
Clarke, R.V., & Eck, J. (2007), Understanding risky facilities. Problem-Oriented Guides for Police
Problem-Solving Tools Series, Guide. No. 6, Office of Community Oriented Policing, US
Department of Justice, Washington, DC.
Contreras, K.S. (2006). Cultivating new opportunities: tribal government gaming on the pechanga
reservation. American Behavioral Scientist, 50, 315–352.
De Anda, R., Levine, J.G., Schrader, D., & Thornberg, C., Ph.D. (2014). 2014 California tribal gaming
impact study: an updated analysis of tribal gaming economic and social impacts, with
expanded study of rstf and charitable effects. Beacon Economics. LLC
162 V. SOSA ET AL.
Dellarocas, C. (2003). The digitization of word of mouth: Promise and challenges of online feedback mechanisms. Management Science, 49(10), 1407–1427.
Dunstan, R. (1997). Gambling in California. IX. Economic Impacts of Gambling. Retrieved from
https://www.library.ca.gov/crb/97/03/Chapt9.html
Eck, J.E. (1995). A general model of the geography of illicit retail market places. In P.S. Vol. 4., J.
E. Eck, & D. Weisburd (Eds.), Crime and Place, Crime (pp. 67–94). Monsey, NY: Criminal Justice
Press.
Eck, J.E., Clarke, R.V., & Guerette, R.T. (2007). Risky facilities: crime concentration in homogeneous
sets of establishments and facilities. Crime Prevention Studies, 21, 225–264.
Evans, W.N., & Topoleski, J.H., (2002). The national bureau of economic research. The Social and
Economic Impact of Native American Casinos. Retrieved from http://www.nber.org/papers/
w9198.pdf
Falls, G.A., & Thompson, P.B. (2014). Casinos, casino size, and crime: A panel data analysis of
Michigan counties. The Quarterly Review of Economics and Finance, 54(1), 123–132.
Felson, M. (2006). Crime and Nature. Thousand Oaks, CA: Sage.
Felson, M., & Clarke, R.V. (2010). Routine precautions, criminology, and crime prevention. In H.
D. Barlow & S. Decker (Eds.), Criminology and Public Policy (pp. 106–120). Philadelphia, PA:
Temple University Press.
Franquez, J., Hagala, J., Lim, S., & Bichler, G. (2013). We be drinkin’: a study of place management
and premise notoriety among risky bars and nightclubs. Western Criminology Review, 14(3),
34–52.
Gazel, R.C., Rickman, D.S., & Thompson, W.N. (2001). Casino gambling and crime: A panel study of
Wisconsin counties. Managerial and Decision Economics, 22(1–3), 65–75.
Gerstein, D. Murphy, S., Toce, M., Hoffmann, J., Palmer, A., Johnson, R, … Hill, M. (1999). Gambling
Impact and Behavior Study: Report to the National Gambling Impact Study Commission. Chicago,
IL: National Opinion Research Center at the University of Chicago.
Grinols, E., & Mustard, D. (2006). Casinos, crime, and community costs. The Review of Economics and
Statistics, 88(1), 28–45.
Grinols, E.L., Mustard, D.B., & Dilley, C.H. (2000). Casinos, crime and community costs. SSRN
Electronic Journal. doi:10.2139/ssrn.233792
Groff, E. (2011). Exploring ‘near‘: Characterizing the spatial extent of drinking place influence on
crime. Australian & New Zealand Journal of Criminology, 44(2) Retrieved January 17, 2016,
156–179.
Haberman, C.P., Groff, E.R., Ratcliffe, J.H., & Sorg, E.T. (2015). Satisfaction with police in violent crime
hot spots: using community surveys as a guide for selecting hot spots policing tactics. Crime &
Delinquency, 62(4), 525–557.
Hakim, S., & Buck, A.J. (1989). Do casinos enhance crime? Journal of Criminal Justice, 17(5), 409–416.
Hennig-Thurau, T., Gwinner, K., Walsh, G., & Gremler, D. (2004). Electronic word-of-mouth via
consumer-opinions platforms: What motivates consumers to articulate themselves on the
Internet? Journal of Interactive Marketing, 18(1), 38–52.
Jacobs, J. (1961). The death and life of great American cities. New York: Modern Library.
Kruegle, H. (2007). CCTV surveillance: Analog and digital video practices and technology. Amsterdam:
Elsevier Butterworth Heinemann.
KTLA5 (7 June 2017). Man Leaving Gardena Casino with Winnings Is Shot 4 Times in Robbery;
Assailants at Large. http://ktla.com/2017/06/07/1-shot-in-possible-robbery-near-gardena-casinolapd/.
Lyons, B., & Henderson, K. (2005). Opinion leadership in a computer-mediated environment.
Journal of Consumer Behavior, 4, 319–329.
Madensen, T. (2007). Bar management and crime: Toward a dynamic theory of place management
and crime hotspots. PhD. dissertation, University of Cincinnati. Cincinnati, OH.
Madensen, T., & Eck, J.E. (2008). Violence in bars: Exploring the impact of place manager
decision-making. Crime Prevention and Community Safety, 10, 111–125.
CRIMINAL JUSTICE STUDIES 163
Madensen, T., & Sousa, W. (2008, July). Integrating place managers and broken windows theory.
Presented at the 17th International Symposium on Environmental Criminology and Crime
Analysis, Anchorage, AlaskaJuly).
Marteache, N., Jimenez, I., & Lizarraga, M., (2018, February). Crimemapping.com: The Holy Grail?
Presentation at the 45th annual meeting of the Western Society of Criminology Conference,
Long Beach, CA.
Moufakkir, O. (2005). An assessment of crime volume following casino gaming development in the
City of Detroit. UNLV Gaming Research & Review Journal, 9(1), 15.
Pontell, H., Fang, Q., & Geis, G. (2014). Economic crime and casinos: China’s wager on Macau. Asian
Journal of Criminology, 9(1), 1–13.
Ratcliffe, J.H. (2004). The hotspot matrix: a framework for the spatio temporal targeting of crime
reduction. Police Practice and Research, 5(1), 5–23.
Ratcliffe, J.H. (2012). The spatial extent of criminogenic places: a changepoint regression of
violence around bars. Geographical Analysis, 44(4), 302–320.
Schmerler, K., Hunter, D., Eisenberg, D., & Jones, M. (2009), “Reducing crime and disorder at motels
and hotels in Chula Vista, CA”, Submission for the Herman Goldstein Award for Excellence in
Problem-Oriented Policing, available at: www.popcenter.org/library/awards/goldstein/2009.
Steffes, E., & Burgee, L. (2009). Social ties and online word of mouth. Internet Research, 19, 42–59.
Stitt, B., Nichols, M., & Giacopassi, D. (2003). Does the presence of casinos increase crime? An
examination of casino and control communities. Crime & Delinquency, 49(2), 253–284.
Tucker, T. (2011). Online word of mouth: characteristics of yelp.com reviews. Elon University,
Communication Science. Retrieved January 16, 2016, from Google Scholar: http://www.elon.
edu/docs/eweb/academics/communications/research/vol2no1/04tucker.pdf
Walker, D. (2008). Do casinos really cause crime?. Econ Journal Watch, 5(1), 4–U1.
Walker, D.M. (2010). Casinos and crime in the USA. Chapter 19 of Benson. In L. Bruce &
P. Zimmerman (Eds.), Handbook on the economics of crime (pp. 488–517). Northampton, MA:
Edward Elgar (November).
Weisburd, D. (2015). The Law of Crime Concentration and the Criminology of Place. Criminology,
53(2), 133–157.
Westbrook, R. (1987). Product/consumption-based affective responses and post purchase
processes. Journal of Marketing Research, 24, 258–270.
Wilcox, P., & Eck, J.E. (2011). Criminology of the unpopular: implications for policy aimed at payday
lending facilities. Criminology & Public Policy, 10(2), 473–482.
Williams, R.J., Rehm, J., & Stevens, R.M.G. (March 11, 2011). The Social and Economic Impacts of
Gambling. Final Report prepared for the Canadian Consortium for Gambling Research. http://
hdl.handle.net/10133/1286
Yelp.com. (n.d.). About Us. Retrieved January 16, 2016, from Yelp.com: http://www.yelp.com/about
164 V. SOSA ET AL.