Abstract
The relationship between urban tourism and crime has been well documented, although a focus on the specifics of tourist victimization is lacking. This study explores police-recorded thefts against urban tourists in the city of Barcelona (Spain). We apply Conjunctive Analysis of Case Configurations to a large data sample to uncover victimization case profiles and the context in which crimes are committed. Comparisons to resident victimization reveal that the case profiles most likely to result in a tourist being the victim of theft have predominantly female, young targets, and occur mostly at restaurants/bars and during the Summer. Future research should examine victimization trends while accounting for time spent in different settings, and compare self-protective behaviors taken by tourists and residents.
Keywords
Introduction
Tourism has become one of the most profitable industries in the last few years. In 2019, 1.5 billion international tourist arrivals were recorded worldwide (United Nations World Tourism Organization [UNWTO], 2020), and the Travel & Tourism sector contributed 10.4% to the global GDP (World Travel & Tourism Council, 2021). Spain is the second most visited country by tourists in the world, with over 83 million international tourist arrivals in 2019 (UNWTO, 2020). The city of Barcelona, located in the northeastern region of Spain (Catalonia), and with a population of around 1.62 million, is Spain’s most popular tourist destination. In the last couple of decades, the tourism sector in the city has experienced a dramatic increase: from 2010 to 2019, the number of tourists visiting Barcelona almost doubled, reaching about 14 million in 2019 (Ajuntament de Barcelona, 2020).
According to official sources, from 2015 to 2019 the city experienced an over 30% increase in crime, led mostly by a rise in the number of thefts, robberies, and burglaries (Ajuntament de Barcelona, 2020; Garrido, 2019). The media, as well as some scholars (Maldonado-Guzmán et al., 2020), partially attributed this phenomenon to the ever-growing tourism industry in the city and the fact that tourists seemed to be specially targeted by thieves (Canalis, 2019; Castan & Albalat, 2019). The situation escalated during the Summer of 2019 when a couple of robberies against foreign officials (one of them resulting in death) resonated broadly in the international press (Euro Weekly News Media, 2019; The Express Tribune, 2019). On August 21, 2019, the U.S. Consulate General in Barcelona issued a travel advisory, due to “an increase in violent crime in the city of Barcelona in the summer of 2019, specifically in popular tourist areas. Local authorities have reported a significant increase in the number of petty theft schemes that have included acts of violence, such as aggressive thefts of jewelry, watches, and purses.” (France 24, 2019).
The relationship between tourism and crime has been well documented in the literature in the past few decades, with the consistent finding that the increased presence of tourists is related to higher volumes of crime (Mawby, 2014). Some authors argue that existing research consists mostly of macro-level quantitative research and that specifics on the details and context of tourist victimization, as well as comparison to resident victimization, are largely lacking (Brunt & Shepherd, 2004; E. Cohen, 2019). The literature on this issue in Spain is notably sparse, especially considering the volume of tourists visiting the country every year. To the best of our knowledge, until very recently, existing research was exclusively focused on Andalucia, the southernmost region of the country (Aebi & Mapelli, 2003; Cerezo et al., 2022; Stangeland, 1998; Stangeland & Felson, 1995), with the only exception of the study by Montolio and Planells-Struse (2016) on the impact of tourist activity on crime rates in Spain as a whole.
Coinciding with the increased availability of official crime statistics, in the last couple of years a few studies have emerged using data from the city of Barcelona: Maldonado-Guzmán (2020, 2022) analyzes the relationship between levels of tourism and urban crime, and Buil-Gil and Mawby (2022) explore the different crime reporting propensities of tourists and locals. However, the basic characteristics of the crimes committed against tourists, as well as the victim typology and the contextual factors that surround the commission of the crime, remain largely unexplored.
Our study aims to identify unique features of tourist victimization in the city of Barcelona through the analysis and comparison of tourist and resident victimization profiles, thereby addressing some of the gaps in the literature described above. We use police-recorded data of theft incidents from 2016 to 2019 to study victim profiles and contextual characteristics of thefts such as the time and location of the crimes. Contributions to the field include: (1) the analysis of contextual factors of theft against tourists/residents applying Conjunctive Analysis of Case Configurations (CACC), (2) the comparison of thefts against tourists and residents using a unique dataset that identifies the tourist/local status of the victim, and (3) the proposal of a novel way to apply CACC to very large samples.
The Relationship Between Tourism and Crime
Although tourism can play different roles regarding crime (Ryan, 1993), most research analyzing the relationship between tourism and crime has focused on exploring tourism as a criminogenic factor, comparing crime rates in tourist destinations against other areas, and analyzing how crime volumes vary during high and low tourist seasons (Allen, 1999). Findings of research conducted in multiple locations consistently indicate that an increased presence of tourists in an area, or during certain seasons, is related to a rise in crime, especially property crime (Alleyne & Boxill, 2003 in Jamaica; Biagi & Detotto, 2014 in Italy; Jud, 1975 in Mexico; Michalko, 2004 in Hungary; Montolio & Planells-Struse, 2016 in Spain; Recher & Rubil, 2020 in Croatia; Walmsley et al., 1983 in Australia; among others).
Not all tourists experience the same victimization risk. Females, as well as young and elderly tourists, are often described in the literature as more vulnerable to property crimes, citing their riskier lifestyle when on vacation (in the case of younger tourists) and a lower capacity to protect themselves against victimization (in the case of females and the elderly) (Allen, 1999; Brunt & Shepherd; 2004; Paliska et al., 2020, Zhao & Ho, 2006; Ngo et al., 2021).
Just as tourists differ in their risk level, not all places and times are equally risky. Crime properties and contextual characteristics vary depending on the type of crime analyzed (Cornish & Clarke, 1987). For example, it is more likely that a robbery happens at a location and time where few people are passing by; meanwhile, pickpocketing tends to occur at crowded places and times, as has been well established when examining the convergence of offenders, targets, and guardians in time and space through the lens of criminological theories such as Routine Activity Theory (L. E. Cohen & Felson, 1979). For this reason, it is important to analyze violent and property crimes separately and to discriminate further even within these general categories. Analyzing property crimes, Crotts (1996) found that tourists were mostly victimized at their hotels and motels, followed by parking lots and garages, and highways and roadways. Zhao and Ho (2006) found, in a study on victimization among hotel visitors, that thefts tended to occur inside their hotel, and during the daytime. Allen’s (1999) study of theft against tourists discusses the same temporal patterns. Vakhitova et al.’s (2022) study of burglary from tourist accommodations reported that all-inclusive resorts, particularly those located within a city, are much riskier than other types of tourist accommodations. Paliska et al.’s (2020) analysis of pickpocketing against foreign victims revealed that this crime tends to happen in public places. Studying where and when different groups of tourists are most victimized, and what type of activities they were engaged in at the time of the crime when compared to the resident population, can provide much-needed insight into tourist victimization (Brunt et al., 2000; Mawby, 2010).
Few studies have focused on examining tourist victimization compared to that of the local population (Brunt & Shepherd, 2004). Some studies using police data extrapolate whether a victim is a tourist or not from their nationality since police departments rarely record whether the victim is a resident or a tourist (de Albuquerque & McElroy, 1999; Mawby, 2017; Michalko, 2004; Paliska et al., 2020; Zhao & Ho, 2006). In other studies, such status is assumed if the crime happened at a tourist resort (Recher & Rubil, 2020; Walmsley et al., 1983). This is an important limitation of existing work on tourist victimization. Whenever actual resident/tourist comparisons have been possible, often through the use of other data sources such as surveys, the evidence shows that tourists are at higher risk of becoming crime victims of theft and robbery than residents (Buil-Gil & Mawby, 2022; see Harper 2001 for a comparison of tourist/resident crime rates in five international locations) and that people are more likely to be victimized on holiday than at their home country (Mawby, 2014).
Other gaps in the literature include the lack of comparison of victimization profiles between tourists and residents beyond their varying victimization rates, as well as the contextual factors of victimization. Cerezo et al. (2022) offer descriptive data on tourist vs resident victimization in Malaga (Spain), although such comparisons are conducted on aggregate data on a variety of crimes (sexual abuse, sexual assault, murder, theft, burglary, robbery, and car theft), except for the analysis of locations of thefts against tourists and residents. As indicated by Harper (2001), “[. . .] to the extent that the crime experience of tourists and resident populations is different, both qualitatively and quantitatively, the development of a criminology of tourism that focuses on the situational context of tourism seems appropriate at this time” (p. 1055).
Our study overcomes some of the limitations of previous scholarship. Using police data where tourist/resident status is specifically recorded, we analyze and compare thefts against residents and tourists with a special focus on the profile of the victims (gender, age) together with the contextual characteristics of the crimes (date, time, type of location). We approach this research from the lens of Routine Activity Theory, described below. The research questions leading this study are:
How do tourist and resident case profiles differ?
What contexts are riskier for tourists than for residents?
Tourists as Vulnerable Victims
As seen above, there is a need “to focus on the risks experienced by different subgroups of tourists and the extent to which these are influenced by their different vacation routines” (Mawby, 2010, p. 33; the same idea is echoed by Brunt, 2010; Harper, 2001; Montolio & Planells-Struse, 2016). The main theoretical approach used in the tourism and crime literature to explain tourists’ heightened victimization risk is Routine Activity Theory (Boakye, 2010; Crotts, 1996; de Albuquerque & McElroy, 1999; Mawby, 2010; Recher & Rubil, 2020; Vakhitova et al, 2022, among others).
Routine Activity Theory (L. E. Cohen & Felson, 1979) states that, for any crime to occur, three elements are required: (1) a suitable target, (2) a likely offender, and (3) the absence of a capable guardian who can prevent the crime from happening. When applied to tourists, the three elements come into play.
Suitable Target
Tourists are particularly attractive targets due to a variety of factors:
They are easy to spot due to the way they dress and how they behave and they are relaxed and off-guard (Crotts, 1996; Harper, 2006; Ryan, 1993)
They carry valuable items (cash, tech, jewelry) and documentation (Ryan, 1993)
They may experience language barriers (Allen, 1999)
They are less likely to report the crime to the police (Buil-Gil & Mawby, 2022), to press charges (Ryan, 1993), and to stay or return to the holiday destination to follow through with the prosecution of an offender (Chesney-Lind & Lind, 1986; Harper, 2006)
They spend less time indoors and more time on the street and out at night (Brunt et al., 2000; Chesney-Lind & Lind, 1986)
Some of them engage in risky behaviors while on vacation that they would not engage in at home, and tend to visit popular and crowded areas and riskier locations such as bars and nightclubs (Brunt & Shepherd, 2004; Chesney-Lind & Lind, 1986; Harper, 2006)
Likely Offender
Studies examining target selection decisions among offenders have found that tourists are preferred targets among pickpockets (Inciardi, 1976), con artists (E. Cohen, 1996), and street robbers (Harper, 2006) for the reasons stated above.
Absence of Capable Guardian
Mawby (2010) argues that low levels of guardianship may influence tourists’ risk of victimization at three levels:
Self-guardianship. Tourists are less vigilant and more prone to find themselves in high-risk environments, knowingly or not.
Community guardianship. Tourists lack the social connections (i.e., neighbors) that could provide informal guardianship due to the transient nature of their visit. Additionally, the high turnover of tourists and workers at tourist destinations offers high levels of anonymity, making it difficult for potential guardians to know whether something is amiss, or whether somebody “doesn’t belong” (Chesney-Lind & Lind, 1986).
Formal guardianship. Local police may consider that crimes against tourists are not a priority, given that the probability of successful apprehension and conviction of perpetrators is extremely low.
Study Methods
Data and Variables
Incident data on thefts against tourists and residents were obtained from Mossos d’Esquadra (Catalonia’s Regional Police) through the Department of Interior of the Government of Catalonia. The dataset included property crimes committed from January 1, 2016, until December 16, 2019, and occurring in the five districts (out of a total of ten) in Barcelona that concentrate most tourist activity (Molinero, 2012; Moll, 2020): Ciutat Vella, Eixample, Sants-Montjuïc, Sant Martí, and Gràcia. Two of these districts (Ciutat Vella and Eixample) concentrate about two-thirds of all hotels in the city of Barcelona, and around 40% of all restaurants and retail stores (Moll, 2020). In the past few years, tourist activity has spilled over to the three adjacent districts as well: Sants-Montjuïc, Sant Martí, and Gràcia (Moll, 2020). While only 56% of all residents are registered in these five districts, 76.4% of all crimes reported in 2019 were concentrated in these areas (Ajuntament de Barcelona, 2020).
In this study, theft is defined as the unlawful taking of personal property without the consent of the owner. The dataset described above classifies property crimes into the following categories: theft, robbery, burglary, theft of vehicle, and theft from vehicle. Only the incidents recorded under “theft” are used in this analysis: this includes any taking of personal property conducted without violence or intimidation (considered robbery according to Spanish law) or breaking and entering (which would constitute burglary), and excludes thefts of vehicles or items taken from a vehicle.
Victims reporting a crime to Mossos d’Esquadra self-identify in the report as residents or tourists, as well as three other categories: visiting for work-related reasons, occasional visitor (i.e., visiting the city to go to a doctor appointment), and other/unknown. According to the Department of Interior, the last three categories, taken together, make up less than 5% of all reported crimes. This data-recording practice overcomes limitations cited in the literature regarding how tourist crime is quantified (Mawby, 2017).
A total number of 221,778 thefts against residents and 114,253 thefts against tourists were recorded during the study period. For each crime event, data recorded includes information about the victim (age, gender, nationality) and the event itself (such as date, time, and type of location).
There are several limitations to the use of official crime records when analyzing crime, including how crime is conceptualized (Buil-Gil et al., 2021). The shortcomings of using police-recorded data in criminological research are well-known and include sources of measurement error such as underreporting, recording bias, and data processing errors, among others (Buil-Gil et al., 2021; Kroneberg et al., 2022). There are additional limitations of police data relevant to our research. International tourists report crime to police at a lower rate than locals do (Buil-Gil & Mawby, 2022) and, in the case of thefts, the time of the incident may be an estimation, as victims of theft and pickpocketing do not always know when exactly the crime happened (Lisowska, 2017). Despite these limitations, police records constitute a valuable source of available data on victim and event characteristics in tourism victimization. Thus, in this study, we operationalize thefts as police-recorded incidents of theft in the five main tourist districts in the city of Barcelona.
Conjunctive Analysis of Case Configurations
We use Conjunctive Analysis of Case Configurations to uncover distinctive features of tourist victimization through the analysis of victimization case profiles and the comparison of tourist and resident victimization profile patterns (CACC; Miethe et al., 2008). CACC is a data analysis technique that focuses on the context of a particular crime through the analysis of unique combinations of variable attributes (case configurations, also called “situational profiles”—Hart, 2020) related to a specific outcome. CACC can be used for exploratory data analysis, as well as hypothesis testing, and it provides the opportunity to examine complex relationships of combinations of categorical variables. In the past few years, a growing body of literature has used this technique to identify dominant situational profiles and to explore patterns within case configurations in a variety of topics, such as violence against college students, street robbery and bus stops, terrorism, mass school shootings, sex offenses, and online harassment, among others (Cook et al., 2021; Gruenewald et al., 2019; Hart & Miethe, 2011, 2014; Moneva et al., 2021; Paez et al., 2021). To the best of our knowledge, this is the first time that CACC has been applied to the study of tourist victimization, and to a dataset with hundreds of thousands of cases.
The first step in conjunctive analysis is to construct a data matrix table (also called “truth table”), by placing all variables that we want to include in our analysis in the columns of the table (Hart et al., 2021). Each row in the matrix represents a unique combination of variable attributes and constitutes a case configuration (or situational profile). For example, if we include four dichotomous variables in our analysis, the matrix will contain a total of (2× 2× 2× 2) 16 unique case configurations. We then aggregate the observations in our dataset to the truth table, which allows us to sort our case configurations from most to least prevalent.
The second step is to identify dominant case configurations by establishing a minimum-frequency threshold (typically 10 cases when N > 1,000, and 5 when N < 1,000), which are then used to explore the patterns in the data. Examination of the level of clustering of unique combinations of variable attributes can also be used to determine the likelihood of a particular outcome (i.e., the dependent variable), but such concentration has been quantified differently in the literature applying CACC. The use of a Situational Clustering Index (SCI) has been recently suggested as a consistent and replicable way to measure the magnitude of clustering among dominant profiles (Hart, 2020).
Table 1 lists the seven variables included in our model and describes how some of them were recoded into categorical variables.
Variables Included in the Conjunctive Analysis of Case Configurations.
Victims under 15 years old (0.3% of the dataset) were excluded from the analysis for consistency with other sources of data on tourism in Barcelona (Ajuntament de Barcelona, 2020). We followed the listwise method to handle missing data, resulting in a 1% reduction of our total sample size.
Results
Table 2 compares tourist and resident victimization and crime event data. We use contingency tables and Chi-Square tests to compare frequencies of residents and tourists in each category. Demographic data for victims of theft is quite aligned between the two groups: females and younger individuals are more at risk than males and other age groups. The main differences are observed when analyzing the crime event: thefts against tourists concentrate more than those against residents in city public areas and bars/restaurants, and during the Summer. Other temporal patterns such as weekday/weekend and time of day are consistent among the two groups, with slight variations.
Comparison of Tourist and Resident Theft Victimization Characteristics, 2016 to 2019.
Conjunctive Analysis of Case Configurations
The analyses provided in Table 2 only offer a snapshot of each of the variables in isolation. To conduct a descriptive analysis of context-specific effects of the variables included regarding the probability of tourists being a victim of theft, we used Conjunctive Analysis of Case Configurations (CACC), as described in the Methods section of this paper. The terms “case configurations,” “case profiles,” and “situational profiles” all refer to case-specific combinations of variables.
A data matrix (or “truth table”) including our 7 variables was created following a 2 (tourist/resident) × 2 (gender) × 6 (age brackets) × 8 (type of place) × 4 (season) × 2 (weekday/weekend) × 3 (time of day) design, which yielded a total of 4,608 potential case profiles, that is, unique combinations of the attributes of the variables included in the model. Of those, 4,287 case configurations were observed and listed from the most to the least prevalent. The first 200 profiles (<5% of the profiles observed) accounted for 30.7% of all the observations in our dataset. This suggests that, although there is variety in the context of thefts, we also find a pattern of clustering among a reduced number of profiles, consistent with findings in similar studies (Hart & Miethe, 2011).
The next step is the identification of dominant case configurations, which is done by selecting the unique combinations of factors that cluster above a minimum-frequency threshold. As indicated above, previous literature suggests a threshold of 10 cases for samples over 1000 (considered “large samples”). But never before has CACC been used with a sample as large as the one in this study, which exceeds 300,000 cases. Applying that threshold to our data yields a total of 1,790 dominant profiles, which renders the analysis futile due to the volume of profiles to be interpreted. As indicated by Miethe et al. (2008), “Rules for minimum cell frequencies are important in conjunctive analysis so that idiosyncratic patterns from low-frequency cells do not adversely affect the interpretation of more dominant patterns of case concentration within a study.” (p. 239). For this reason, the minimum-frequency threshold had to be reconsidered in our study (Hart, personal communication on July 14, 2022).
The selection of specific thresholds to define dominant configurations has been referred to as “a rule of thumb” that must constitute a “reasonable minimum” (Miethe & Drass, 1999, pp. 10, 11), and has been compared to the choice of significance levels in traditional statistical analysis (p < .01, .05, or .10) (Hart, 2020). The threshold should be high enough to define a small number of profiles relative to the number of observed profiles while maintaining a sufficient level of clustering. We propose the use of the Situational Clustering Index (SCI), which has been recently suggested as a measure of profile clustering in CACC analysis (Hart, 2020), to help determine the most appropriate minimum-frequency threshold in this study (Figure 1).
Calculation of situational clustering index using 1 to 1,000 minimum-frequency thresholds.
We calculated all SCIs using thresholds ranging from 1 to 1,000 (only four case configurations exceeded N = 1,000) and proceeded to compute their mean and standard deviation (X̄ = 0.2056, SD = 0.1097). Three options were explored: using the mean (threshold = 359), the mean + 1 SD (threshold = 148), and the mean + 2 SD (threshold = 58). The first option was discarded due to the low SCI, and the last one did not represent a significant improvement over the use of a threshold of 10 cases, as a total of 1,426 profiles exceeded the minimum-frequency threshold of 58 cases. Additionally, a visual inspection of Figure 1 revealed that the mean + 1SD marks the point where the curve starts leveling off.
Using a minimum-frequency threshold of 148 cases, a total of 673 dominant situational profiles were identified (14.6% of all potential profiles). Figure 2 illustrates how theft victimization clusters within these profiles.
Frequency of theft victimizations within each situational profile.
A goodness-of-fit test was performed to determine clustering in the data (Hart, 2020). The findings (χ2 [672, N = 332,550] = 107643.6, p < .000) indicate that thefts concentrate among a subset of dominant profiles, that is, that thefts tend to occur when certain variable attributes are combined in particular ways. This further supports the idea that thefts against tourists in urban areas are context-dependent. We measure the magnitude of profile concentration using the Situational Clustering Index. The SCI ranges from 0 to 1, with higher values indicating higher levels of clustering. Our results show moderate clustering (SCI = 0.316, 95% level CI [0.30, 0.33]); confidence interval levels (CI) were calculated using bootstraps (with 2000 replicates for estimating the bias-corrected and accelerated, “BCa,” bootstrap interval) following Bernasco and Steenbeek’s (2017) approach when calculating Gini’s coefficient CI (see also Steenbeek & Bernasco, 2018, for the R package). Figure 3 provides a visual representation of the profile clustering.
Lorenz curve for the magnitude of the situational clustering.
Given that the focus of our study are crime events where tourists are the identified victims, we ran the conjunctive analysis using type of victim (tourist-1 vs. resident-0) as our dependent variable. This allowed us to determine what combinations of situational factors are more likely in thefts with a tourist and with a resident victim. We proceeded to consider the most/least likely case configurations in thefts with tourist victims by calculating the mean of the probability of the victim being a tourist (
Case Configurations With Victims Most Likely to Be Tourists.
Number of categories within variables in parentheses.
Case Configurations With Victims Most Likely to be Residents (Least Likely to be Tourists).
Number of categories within variables in parentheses.
The top profile in Table 3 shows that when the victim of a theft is a young male and the theft is committed during the Summer, on a weekday morning, at a restaurant or bar, the probability of that victim being a tourist is very high (0.725). Examination of the top 17 dominant situational contexts of thefts against tourist victims reveals that tourists experience the highest risk of victimization during the Summer (100%), especially during weekdays (76%), and in the morning (53%). Seventy-one percent of theft profiles feature a victim under 24 years of age, with a slight preponderance of female victims (59%). In terms of location, out of eight possible types of place, 59% of case configurations take place at restaurants or bars.
If we look at the very last profile in Table 4, we can see that when a female of 65 years of age or older is victimized during the Winter, on a weekday morning, in a retail establishment, the probability of that victim being a tourist is extremely low (0.034). The contexts in which victims are less likely to be tourists (18 situational contexts) or, in other words, in which victims are more likely to be residents, include thefts happening in seasons other than Summer (50% of least likely profiles occurred in the Winter), during a weekday (89%), and in the morning (61%). Most notably, victims in these case profiles are all over 34 years of age, with the age brackets of 35 to 44 and 65+ being most prevalent (44% and 39%, respectively). 61% of the victims in the least likely case profiles are male. Finally, 78% of these profiles take place in retail settings.
Discussion
This study set out to explore the unique features of tourism victimization, as well as to identify what contexts are related to a higher risk of thefts against tourists, by examining the differences between police-recorded tourist and resident thefts. Police recorded data on the five main tourist districts of Barcelona from 2016 to 2019 were analyzed, and Conjunctive Analysis of Case Configurations (CACC) was used to determine whether thefts against tourists are context-dependent, and which situational contexts are related to increased risk.
When comparing descriptive data of police-recorded thefts against tourists and residents, we see that they are very similar in the victim’s demographic characteristics, as well as the temporal patterns of thefts. The main difference is that thefts against tourists concentrate during the Summer (37% of all thefts occur during this season), while thefts against residents do not present such seasonal trends (25% occur during the Summer months). Interestingly, the percentage of tourists entering the city between June and August is about 28% for each of the years included in the study (Ajuntament de Barcelona, 2022), as Barcelona experiences year-round tourism with a slight (but not dramatic) increase during the Summer, which seems to indicate that tourists are disproportionately targeted during that season. However further research is needed to fully understand this phenomenon. Regarding where thefts happen, the four most common types of locations for both groups are city public areas, transportation, restaurants/bars, and retail. When comparing the two groups, the proportion of police-recorded thefts against tourists is higher than those of residents in city public areas and restaurants/bars, while resident police-recorded theft is more represented in transportation and retail.
The differences between the profiles of tourists and residents become more salient once CACC is performed. The case profiles most likely to result in a tourist being the victim of theft have predominantly female, young targets (under 34 years old), and occur mostly at restaurants/bars. These findings are consistent with Cerezo et al.’s (2022) comparison of risky locations for theft victimization among tourists and residents. Risk exposure (tourists are more likely to spend more of their time in the city eating out), as well as lack of self-protective behaviors by tourists, are plausible explanations for the clustering of thefts at restaurants and bars. For example, it is a “known fact” among residents in Barcelona that, when eating out, bags should never be placed on the back of the chair or on the floor where they are not being actively monitored (Barcelona Tourist Guide, n.d.). However, this local practice may not be known among tourists. The least likely case configurations occur mostly in retail establishments (78%), and to older victims. An examination of the number of thefts suffered in different settings accounting for the time spent in them (i.e., using public transportation) would provide finer measures of relative risk for tourists and residents (in line with the work of Lemieux & Felson, 2012). Future research should study self-protective behaviors (i.e., routine precautions—Felson & Clarke, 1995) taken by tourists and residents in different settings to test likely explanations for these results. Additionally, comparing the profile and travel habits of victimized and non-victimized tourists would also help shed light on these issues.
In terms of temporal patterns, the CACC confirms, once again, that Summer is the riskiest season for tourists, at least for our sample of police-recorded thefts. Every single case profile with the highest probability of tourist victims occurs in the Summer, while none of the ones with the lowest probability do. Assuming that Summer visitors behave similarly as visitors during other seasons do (which should be empirically confirmed), our results suggest a disproportionate targeting of tourists during the Summer. This finding could be explained by groups of “outsider-offenders” who travel to cities like Barcelona during peak tourist season to commit thefts, as suggested by some authors (Mawby, 2017).
Finally, looking at the case profiles with the highest probability of tourist victims, we see that risk is the highest for young tourists at bars and restaurants during the Summer. This is an important finding that could be used to design very specific interventions directed at preventing tourist theft victimization in the city.
Some of the limitations of this research that have been discussed throughout the paper include the use of police reports as the source of crime data and the fact that, in the case of thefts from the person, the time of the offense may be an estimation whenever the victim does not know for sure when the theft happened. It is also worth noting that this study is based only on the five main tourist districts in Barcelona that concentrate 76% of all crimes and not the whole city. Moreover, our findings may not generalize to cities with different tourism compositions. Finally, we proposed a new analytical strategy to establish minimum-frequency thresholds in CACC analyses using very large samples, which will need to be tested and validated in future research. Despite these limitations, this research was conducted with data on crimes with actual tourist victims; it addressed some of the important gaps identified in the literature and developed CACC further for its use with very large samples, thereby opening avenues for future research.
Policy Implications
Exploration of profiles of police-recorded thefts against tourists and residents provided invaluable insights into their context and patterns. Comparisons of both types of victims were only possible due to the specific recording of tourist status when reporting the crime. Law enforcement agencies around the world, particularly those in tourist destinations, should consider incorporating this recording practice to be able to analyze and compare tourist and resident victimization.
Analysis of the contextual characteristics of crime events was used to assess when and where tourists are most often victimized, according to police data. Not only are these findings relevant to the scientific understanding of this phenomenon in general, but they draw a clearer picture of the characteristics of tourist victimization in one of the world’s leading tourist destinations. This knowledge can, in turn, be used for the design of tourist-specific crime prevention initiatives that, given the volume of tourist victimization, would surely have an impact on the crime levels of the city in general.
Several interventions have been suggested to reduce and prevent crimes against tourists: some examples include establishing partnerships between the tourism industry and criminal justice agencies, facilitating crime reporting by tourists, creating specialized tourism police units or tourism victims’ support services, making an effort to collect and analyze tourist victimization data to aid in the creation of targeted crime prevention initiatives, and tourist education campaigns (Buil-Gil & Mawby, 2022; de Albuquerque & McElroy, 1999; Mawby, 2017; Mawby & Ozascilar, 2022).
Authorities in Barcelona have already taken some of these steps: in 2008, the city implemented a system that allows tourists to report crimes at their hotels; in 2014, the regional police started recording tourist/resident status in their crime reports; and in 2020 the Department of Interior created the “Pla Barcelona Ciutat Segura” (Barcelona Safe City Plan) in which it coordinated efforts from stakeholders in different sectors (transportation, retail, tourism, nightlife, etc.) to work together in the reduction of crime in the city. Other interventions at what are known as risky facilities or hotspots should be considered, for example at restaurants and bars, guided by detailed analyses of where and when crimes happen, and who are the most targeted victims. Mixed methods research designs combining validated ecometrics measures and interviews with place managers and place users can help identify potential social mechanisms at the location level (Eck, 2010).
Conclusions
This paper examined the characteristics of police-recorded theft events against tourists in urban settings using a unique dataset where the tourist/resident status of theft victims was recorded. We used Conjunctive Analysis of Case Configurations (Miethe et al., 2008) to shed light on victim profiles and contextual factors of this criminal behavior in the city of Barcelona (Spain), and we proposed an adaptation of this analytical strategy for use with large samples, which represents a relevant methodological innovation. Our findings indicate that tourist theft victimization concentrates in time (Summer) and space (in locations such as restaurants) and that these patterns are different from those of victimized residents. Policy implications derived from these findings were discussed above.
Further research on this area should examine more deeply the factors that influence the risk of victimization. To that end, it would be useful to compare profile and behavioral differences between victimized and non-victimized tourists, as well as self-protective behaviors adopted by residents and tourists in places such as restaurants or transportation. Additionally, other crimes such as robbery should also be considered and analyzed: contrasting similarities and differences with the findings of this paper could reveal alternative explanations for the patterns observed.
It is worth noting that, as we write this paper, COVID-19 travel restrictions are being lifted worldwide. It is expected that international travel will gradually recover and reach 2019 levels by 2023 or 2024 (United Nations Conference on Trade and Development, 2021). This is a unique opportunity for local authorities, along with the tourism sector, to reflect on tourist victimization in urban settings and to plan new approaches to these issues.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
