Vulnerability and perceived insecurity in the public spaces of Barcelona / Vulnerabilidad y percepción de inseguridad en el espacio público de la ciudad de Barcelona

Abstract The concept of vulnerability is directly linked with perceived insecurity and fear of crime, especially with regard to women and older people in our cities. However, it is not the only paradigm contemplated in the analysis of this phenomenon. In addition to a model of vulnerability, some scholars propose a model of disorder and a model of social integration. The aim of this article is to set out some of the results obtained from a questionnaire about perceived insecurity in eight public spaces in the city of Barcelona. It analyses the relationship between the variables of gender and age, and factors in the questionnaire linked to the model of vulnerability (coping, prior experiences and social representation), disorder (perceived environmental quality and tolerance towards uncivil behaviours) and social integration (satisfaction/identification and social cohesion). The findings indicate that the models of vulnerability and disorder have a greater impact on perceived insecurity in the case of women, whereas older people displayed no differences from other groups with regard to perceived insecurity, associated with a greater influence of the model of social integration.

Palabras clave: vulnerabilidad; miedo al delito; espacio público; género; edad A great deal of research from different fields and disciplines has attempted to identify factors capable of predicting citizen insecurity or fear of crime. One sociological strand of research relates urban insecurity to a broader experience of other types of insecurities such as working, financial or emotional (Hollway & Jefferson, 1997), in addition to insecurities derived from change in social welfare policies (Hummelsheim, Hirtenlehner, Jackson, & Oberwittler, 2011). At an urban level, the increase in social inequalities resulting from the global economic crisis, along with increased ethnic diversification and the appearance of new social groups, seems to be making coexistence in cities much more complex, sometimes leading to an increase in citizen insecurity as a social construct (Brunton-Smith & Sturgis, 2011;Di Masso, Dixon, & Pol, 2011;Henry, 2007). Urban insecurity, understood as a perceived risk of victimization, is also related to social segregation (Vilalta, 2011), urban stigmatization (Quillian & Pager, 2010) and the loss of public space as social space (Finol, 2005;Low, 2005;Valera, 2008).

Explanatory models of fear of crime
The multiple proposals developed to explain fear of crime have been summarized by Franklin et al. (2008) on the basis of three models: vulnerability, environmental disorder and social integration.
With regard to the model of vulnerability, authors such as Van der Wurff, Van Staalduinen, and Stringer (1989) have identified variables capable of predicting psychological insecurity, such as the perception of being an attractive target for possible criminals, the attribution of criminal intentions to another individual or group, the feeling of being in control and the confidence of being unhurt during a possible threatening situation, and the perceived likelihood of being the victim of crime in a specific place. In short, vulnerability can become one of the principal factors that influence a person's perceived wellbeing (Sulemana, 2015) or which may account for the withdrawal from public spaces of certain collectives who, perceiving themselves to be vulnerable, avoid contact with potentially threatening or dangerous situations (Low, 2003;San Juan, Vozmediano, & Vergara, 2012;Vesselinov, Cazessus, & Falk, 2006).
Consequently, numerous studies have stressed sociodemographic variables linked to vulnerability (Rader, Cossman, & Porter, 2012), especially age and gender. In spite of certain critical positions (Reid & Konrad, 2004), many studies indicate that women, older people and younger people are more likely to experience fear of crime (Amerio & Rocatto, 2005;Gardner, 1990;Jackson, 2009;Lawton & Yaffe, 1980;Mesch, 2000;Roman & Chalfin, 2008;Saldívar, Ramos, & Saltijeral, 1998;Warr, 1984). Furthermore, people who perceive themselves to be physically more vulnerable tend to develop more acute feelings of insecurity (Cossman & Rader, 2011). The relationship between fear of crime and vulnerability (Jackson, 2004), perceptions regarding limited capacity to cope with dangerous situations (Moser & Levy-Leboyer, 1985), and social levels of tolerance with regard to antisocial behaviour or risk (Torrente, 2001) complement this framework. Risk theories, on the other hand, consider fear of crime to be the interaction between emotion and cognition (Jackson, 2006), and highlight perceived control and perceived consequences as important factors that moderate the relationship between perceived likelihood and concern with crime (Jackson, 2011), even among victims, as an effect of resilience (Shippee, 2012). The concept of Defensible Space (Brown & Werner, 1985;Newman, 1996) talks about social strategies and spatial configurations that allow a community to control its environment securely (Marzbali, Abdullah, Razak, & Tilaki, 2012). In addition, the social and structural characteristics of a neighbourhood can influence efficacy and mechanisms of collective social control (Sampson, 2009).
The model of disorder, on the other hand, emphasizes perceptions of a high correlation between environmental disorder and social disorder. It frames antisocial behaviour (incivilities), understood as an absence of civil order (Ferraro, 1995;Hunter, 1978), to the extent that this phenomenon reflects social degradation or potential threat (LaGrange, Ferraro, & Supancic, 1992;Roché, 1993). In the words of Hunter: '"erosions" and "accumulations" in the physical environment, burned-out buildings and rubbish on the streets lead people to make inferences about an area and, more specifically, about the type of people who live there. In the absence of personal encounters, these signs transmit an image of "disorder" and, specifically, the "loss of civil society" ' (1978, p. 7).
This relationship between social disorder and environmental disorder, developed broadly on the back of the Broken Windows Theory (Wilson & Kelling, 1982), proposes one of the most fruitful hypotheses in relation to the subject of this article: people who perceive a neighbourhood to be more disordered will tend to worry more about their safety in that neighbourhood (Swatt, Varano, Uchida, & Solomon, 2013;Wyant, 2008). However, when analysing the impact of physical and social disorder on perceived insecurity, recent studies have questioned Broken Windows Theory, showing the importance of perceived social disorder over environmental disorder in explanations of fear (Acuña-Rivera, Uzzell, & Brown, 2011;Hinkle, 2015). In addition, research has recently been calling for a reinterpretation of the concept of civility in cities, not as an imposition but rather as a symptom of democratic consideration of the public space (Bannister, Fyfe, & Kearns, 2006;Boyd, 2006;Groth & Corijn, 2005;Phillips & Smith, 2006). Finally, the model of social integration encompasses psychosocial variables that are linked with social cohesion, a sense of community and place attachment. These variables, which can be grouped into what Skogan and Maxfield (1981) have termed urban life, include aspects such as density-overcrowding (Thomé & Torrente, 2003), difficulties of social integration, size of buildings (Newman & Franck, 1982), presence of potentially dangerous or threatening social groups, fear of social difference (Sandercock, 2000;Vieno, Rocatto, & Russo, 2013) and fear of the unknown, of outsiders (Amin, 2006;Quillan & Prager, 2010;Sandercock & Attili, 2009). Other authors have considered other types of variables, such as residential satisfaction, place attachment or place identification (especially at the level of the neighbourhood), to explain modulating effects of perceived insecurity (Taylor, Shumaker, & Gottfredson, 1985;Di Masso et al., 2011;Scanell & Gifford, 2010;Taylor, 1996;Tester, Ruel, Anderson, Reitzes, & Oakley, 2011;Vidal, Valera, & Peró, 2010). The underlying logic links feeling safe and secure in a neighbourhood and feelings of place attachment and identification with that neighbourhood to positive repercussions on subjective wellbeing and psychosocial health (Hill, Ross, & Angel, 2005;Kitchen & Williams, 2010;Sulemana, 2015). In this regard, the concept of collective efficacy becomes particularly relevant as a predictor of victimization risk (Pitner, Yu, & Brown, 2012;Swatt et al., 2013), although it is not always linked to the emotional aspects of fear of crime (Yuan & McNeeley, 2015) The dimensions of perceived insecurity Among all the attempts to develop integrating models of fear of crime, Carro, Valera, and Vidal (2010) have proposed one based on three factors: profile of personal skills; representation of the space; and dangerous environment. Starting with this model, and continuing along the same lines as previous articles (Guardia, Valera, Carro, & De La Fuente, 2009;Valera & Guàrdia, 2012, 2014, this paper approaches the phenomenon of perceived insecurity on the basis of four dimensions with their associated factors or hypothetical constructs ( Figure 1): • Level of perceived coping skills. This dimension includes perceived control and the capacity to cope with potentially threatening situations, as well as  the level of social skills linked to variables such as cohesion or social integration.

PERCEIVED INSECURITY
• Representation of the unsafe space. This dimension includes experiences of victimization and the social representation of dangerous places, as well as the level of satisfaction and identification with the neighbourhood. • Characterization of the dangerous environment, through an assessment of socio-physical aspects linked to spatial design, which may affect feelings of insecurity. • Tolerance and civility, including a factor regarding levels of tolerance towards behaviours that could potentially be labelled antisocial, or incivilities, in addition to a global perception factor for incivilities in the urban context in question.
As a result of the above, it is possible to establish a relationship between the hypothesized factors and the theoretical models presented, so that the factors 'control/coping', 'prior experiences' and 'social representation of the dangerous place' can be linked to the model of vulnerability; 'perceived environmental quality', 'tolerance' and 'perceived civility' can be linked to the model of disorder; and finally, the factors 'satisfaction/identification' and 'integration/social cohesion' would correspond to the model of social integration.

Objectives
The first aim of this article is to analyse in depth the instrument utilized and its capacity to gather relevant information regarding the different factors that make up our explanatory model of insecurity in urban environments.
It will also seek to profile the relationships that exist between dimensions linked to perceived insecurity and the classic variables of gender and age examined in vulnerability studies. In turn, by allocating the dimensions proposed to the different models explaining fear of crime (vulnerability, disorder and social integration), we will analyse which of them has the greatest influence over gender and age with a view to glimpsing any potential modulating effects with regard to global perceived insecurity.

Participants
The sample comprised a total of N = 1,320 people. 75% of the questionnaires were administered on working days and 25% at the weekend. Efforts were made to balance the times at which they were conducted (as far as possible, since some of these spaces close at night), differentiating between morning (34%), afternoon (36%) and evening (30%).
A similar number of questionnaires were conducted in each space as there were people in said space or in its area of influence, as shown in Table 1.
As for the sample characteristics, the gender variable was balanced, and adequate distribution was sought in relation to age, with 40% of respondents aged 25-45, 27% aged 46-64, 18% aged 18-24 and 15% of respondents aged over 64. A total of 71% of those surveyed lived in the same neighbourhood as each space, and 23% stated that they had been born in that neighbourhood (Table 2).

Instrument
The questionnaire utilized is an updated version of an instrument developed previously by the research team (Carro et al., 2010;Guardia et al., 2009;Valera & Guàrdia, 2012, 2014. In relation to the previous version, some of the items were adjusted in order to relate them more clearly to the underlying dimensions, and in turn, a new measurement scale was tested out in relation to the items about perceived environmental quality. The final result was a questionnaire made up of 65 items, the majority of which were presented in a Likert-type scale with response options ranging from 1 to 6. It included questions about the habitual usage made of the space (items 1-10) as well as factors contemplated in the theoretical model defined previously: perceived safety and security in the space (items 11-16, 22-34); direct or indirect experience of victimization (items 17-20); social representation of the dangerous environment (items 23, 24, 35); residential satisfaction and perceived social identity (items 36, 37, 41, 42); integration and social cohesion (items 38, 39, 40); assessment of environmental quality (items 43-52); and finally, levels of tolerance and civility (items 53-65). The questionnaire also included the customary sociodemographic data. The measurement scale for the items pertaining to perceived environmental quality consisted of a 10 cm line inserted in a differential semantic type format in which the respondent had to mark a point according to their degree of proximity to either of the poles of the differential. The result was then measured in millimetres and, if required, inverted in order to standardize the meaning of the assessment. The internal reliability of each factor based on its indicators (items) is shown by means of Cronbach's Alpha coefficients. According to Muñiz (2005), good reliability is accepted as of .700, and excellent reliability as of .850. In our case, having excluded items 35 and 65 from the analysis, only the factor 'Control and coping capacity' yielded a reliability index that we might consider to be on the threshold of goodness (Cronbach α = .700). Two factors ('Social cohesion' and 'Prior experiences') give indexes of above .750, and the remainder are over .800. Of particular note are the factors 'Social representation of the dangerous environment' and the principal factor of this study 'Perceived insecurity', which yielded values of .905 and .916 respectively (see Table 3).

Selection of public spaces
When selecting the spaces to conduct the survey, we proceeded in a way that would ensure maximum diversification. To this end, we used data taken from a previous study of 40 parks and squares in Barcelona using the observational analysis instrument EXOdES (Pérez, Valera, & Anguera, 2011). Three indicators were selected from the set of observational variables that, according to the literature, are sensitive to perceived insecurity: social diversity; conflictive uses in the space; and signs of poverty in the space. The 40 spaces were ordered according to these indicators and then divided into quartiles. We then proceeded to select spaces with a view to obtaining different configurations according to different levels of each indicator: for example, spaces located in the first quartile for the three indicators (minimal social diversity, minimal levels of conflict and minimal levels of signs of poverty) or others that were located at the opposite pole (maximum social diversity, maximum levels of conflictive uses and signs of poverty), but also spaces with significant combinations of the first and last quartiles. Finally, eight spaces were selected, as shown in Figure 2 below.

Results
Following the internal reliability analysis of the factors, global scores were obtained for each subscale, turning them into variables with the values shown in Table 4. In general terms, a trend is observed towards perceived insecurity in the public space (  However, the data are not as homogeneous when analysed on the basis of the gender and age of the people surveyed. Hence, the results with regard to the gender variable, in line with much of the literature in this regard (Ferraro, 1995;Franklin & Franklin, 2009;Snedker, 2012), show that there are statistically significant differences in many of the items on the questionnaire. To put it another way, men and women tend to disagree clearly with regard to questions about security and safety in public spaces. In our case, they do so for seven of the nine factors contemplated. Figure 3 shows these effects.
The most clearly and directly linked factors with perceived security (an individual's own perceived insecurity, level of perceived control, perception attributed to others that a space is dangerous and the degree of experiences of victimization) show significant differences between women and men. Women perceived greater insecurity (t = 4.939; df = 1,264.8; p < .05; r = .137), have lower perceptions of control and the capacity to cope with potentially threatening situations (t = 9.762; df = 1,275.828; p < .05; r = .263) and greater sensitivity towards the social representation of insecurity, in other words a greater conviction that others believe the place to be unsafe (t = 2.014; df = 1,293.499; p < .05; r = .05). Women also claim to have more access than men to directbut above all indirectexperiences of victimization (t = 4.294; df = 1,235.476; p < .05; r = .121).
Furthermore, women are more sensitive and critical than men with regard to the quality of the public space, yielding statistically significant differences on scores pertaining to perceived environmental quality of the place (t = −4.510; df = 1,266.934; p < .05; r = .125) as well as greater levels of intolerance towards possible antisocial behaviour (t = −6.512; df = 1,288.596; p < .05; r = .178) and, in general, a greater perception of incivilities in behaviour than men (t = 3.412; df = 1,282.893; p < .05; r = .094).
Unlike the effects detected for the gender variable, age appears not to yield great sensitivity to the majority of variables directly related to the subject at hand, such as one's own perception of insecurity, access to experiences of victimization, the perceptions of others that a place is dangerous and perceived level of environmental quality or degradation. However, the hypothesis of vulnerability does hold insofar as older people perceive themselves to be less capable of controlling potentially threatening situations (F = 7.819; df = 3.1; p < .05; η 2 = .216). Furthermore, age does display sensitivity towards more 'psychosocial' aspects of the subject; hence, older people have a stronger perception of social cohesion in the context of the space analysed (F = 3.090; df = 3.1; p < .05; η 2 = .156), there is an increase in the level of satisfaction with the neighbourhood that correlates to an increase in age (F = 19.464; df = 3.1; p < .05; η 2 = .337), and the same is true of the level of intolerance to apparently uncivil behaviours (F = 42.776; df = 3.1; p < .05; η 2 = .472). In short, the older the person, the greater the levels of intolerance and perceived vulnerability, but also the greater their appreciation of the 'social quality' offered by the space in question or the neighbourhood where the space is located. Figure 4 summarizes the effects described.

Discussion
In light of the results set out above, there appear to be interesting effects in the analysis of the principal variables associated with vulnerability and, consequently, with the perception of fear and insecurity; genderin the case of womenand agein the case of older people.  Factors that are most directly linked to perceived security (one's own perception of insecurity, control and coping, prior experiences and social representation) yield significant differences according to gender. In line with the literature, women tend to perceive greater insecurity (Ferraro, 1995), less control in the face of threatening situations (Killias & Clerici, 2000), a greater perception that people consider the environment to be unsafe and greater access to experiencesparticularly indirect onesof victimization (Jackson, 2011). In short, it would seem that women tend to develop what Jackson (2011) refers to as a greater sensitivity towards fear of crime. This is coherent with the role developed by the concept of vulnerability not only from an individual perspective, but also a sociostructuring one (Snedker, 2012), and also with the model of vulnerability proposed by Franklin et al. (2008).

PERCEIVED SAFETY
Furthermore, following these latter authors, the model of disorder also has a strong influence on the assessments made by men and women. Women are much more sensitive and critical in their assessment of the environmental quality (physical variables) and social quality (behaviours within the public space) of the parks and squares analysed. In contrast, no significant differences were detected in the third of the models proposed: social integration. In fact, the factors 'integration/cohesion' and 'satisfaction/identification' linked with the model do not display differential tendencies in the responses given.
With regard to the age variable, the findings are not as conclusive, and in line with Lloyd-Sherlock, Agrawal, and Minicuci (2016)   vulnerability does not appear to be clearly applicable in the case of age: there are no significant differences with regard to perceived insecurity, level of experiences of victimization or social representation of the dangerous environment. This effect could be due, as Tullock (2000) explains, to the fact that older people choose a lifestyle that permits them to carry out their daily activity without being exposed to more risk than strictly necessary, avoiding potentially dangerous situations. This explanation is backed up, in our case, by the fact that older people differ significantly from the other age groups with regard to their perceived capacity to cope with potentially threatening situations.
The influence of the model of disorder also seems to be complex, since although there are no significant differences between age groups with regard to environmental quality, as the age of the groups increases, the level of intolerance towards potential uncivil behaviours also increases. Finally, the model of social integration appears relevant since the factors 'integration and social cohesion' and 'satisfaction and identity' display significant differences, as older people score the highest on them. These results corroborate previous studies that have highlighted the importance of confidence, social cohesion and place attachment as moderator variables of fear of crime (Brown, Perkins, & Brown, 2004;Oh & Kim, 2009;Pitner, Yu, & Brown, 2012), which would be along the lines of the effect of collective efficacy on perceived insecurity (Swatt et al., 2013). In short, we could conclude that the greater perception of vulnerability detected among older people is compensated by a positive perception of the social qualities of the surroundings. This would explain why, in spite of feeling more vulnerable, there are no significant differences in their perceived insecurity. In this case, the model of social integration seems to act as an attenuating factor for perceived insecurity.

Conclusions
Public spaces are diverse, not only in terms of their structural nature but also the types of users that frequent and use them. To ignore this diversity would be to ignore the democratic nature that should be attributed to any space considered public. In this article, we have analysed this diversity with regard to the variables of gender and age. The conclusions reached about what makes an environmental safe and secure or not are substantially different in the two cases. With regard to gender, the differences observed in the assessment of environmental order and tolerance towards behaviours likely to be considered uncivil or antisocial require a gender perspective to be adopted when analysing, designing and managing public spaces, possibly in accordance with somewhat different parameters to those habitually used.
On the other hand, in relation to age, fostering cohesion and social integration, favouring positive identification with the neighbourhood and increasing perceived residential satisfaction are, in our contexts, the most powerful tools for improving perceived safety and security in public urban spaces.

Discusión
A la luz de los resultados expuestos, podemos detectar efectos interesantes en el análisis de las principales variables asociadas a la vulnerabilidad y, por consiguiente, a la percepción de miedo e inseguridad; el géneroen el caso de las mujeresy la edaden el caso de las personas mayores.
Por otro lado, y en relación con la edad, la potenciación de la cohesión y la integración social, y el favorecer una identificación positiva con el barrio y un incremento de la percepción de satisfacción residencial son, en nuestros contextos, las herramientas más potentes para mejorar la percepción de seguridad en el espacio público urbano.