( UN ) SAFE AND (IN)SECURE AT HOME: VARIATIONS IN RESIDENTIAL SECURITY IN BRAZIL By SARAH RUTH BENTON A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIR EMENTS FOR THE DEGREE OF MASTER OF ARTS UNIVERSITY OF FLORIDA 2014
2014 Sarah Ruth Benton
4 ACKNOWLEDGMENTS I thank my family for their continued love, prayers, and support. I also thank my friends and c olleagues for their support through these last four years inasmuch as I am thankful to them for the many distractions that have made my time in graduate school special I am extremely thankful for the advising, encouragement and patience of my supervisor y committee. Additionally, I am eternally indebted to the Center for Latin American Studies and the Tinker Foundation for the funding opportunities given to me throughout the years that have contributed to my research and academic development.
5 TABLE OF C ONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGURES ................................ ................................ ................................ .......... 8 LIST OF ABBREVIATIONS ................................ ................................ ............................. 9 ABSTRACT ................................ ................................ ................................ ................... 10 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 11 Study Area ................................ ................................ ................................ .............. 17 Outline of Chapters ................................ ................................ ................................ 19 2 LITERATURE REVIEW ................................ ................................ .......................... 20 Fear of Crime ................................ ................................ ................................ .......... 2 1 Crime Victimization and Vulnerabilities in Brazil ................................ ............... 24 A Paradox: Actual Crime and Fear of Crime ................................ .................... 27 Responses to F ear of Crime ................................ ................................ ................... 28 Gated Communities ................................ ................................ .......................... 29 Household Security Measures ................................ ................................ .......... 35 3 METHODOLOGY ................................ ................................ ................................ ... 43 Data ................................ ................................ ................................ ........................ 43 Variables ................................ ................................ ................................ ................. 44 Primary Independent Vari able: Fear of Crime ................................ .................. 45 Dependent Variable: Index of Residential Fortification ................................ ..... 45 Control Variables ................................ ................................ .............................. 47 Data Analysis ................................ ................................ ................................ .......... 51 Tables ................................ ................................ ................................ ..................... 53 4 FINDINGS ................................ ................................ ................................ ............... 54 General Results ................................ ................................ ................................ ...... 54 Multivariate Analysis ................................ ................................ ............................... 59 Figures ................................ ................................ ................................ .................... 62 Tables ................................ ................................ ................................ ..................... 65 5 DISCUSSION AND CONCLUSION ................................ ................................ ........ 73
6 LIST OF REFERENCES ................................ ................................ ............................... 83 BIOGRAPHICAL SKETCH ................................ ................................ ............................ 87
7 LIST OF TABLES Table page 3 1 Variable Descriptions ................................ ................................ .......................... 53 4 1 Me an Years Education Completed, Brazil and Major Regions, 2009 ................. 65 4 2 Mean Monthly Income Per Capita in Reais Brazil and Major Regions, 2009 ..... 65 4 3 Type of Residence, Brazil and Major Regions, 2009 ................................ .......... 66 4 4 Victimization, Brazil and Major Regions, 2009 ................................ .................... 67 4 5 Distribution of Fear, Brazil and Major Regions, 2009 ................................ ......... 68 4 6 Distribution of Security Mechanisms, Brazil and Major Regions, 2009 ............... 69 4 7 Means of Security Device Index by Region, 2009 ................................ .............. 70 4 8 Level of Residential Fortification Regressed on Fear of Crime in the City and Selected Indicators: Brazil, 2009 (OLS regression co efficients) ......................... 71
8 LIST OF FIGURES Figure page 4 1 Education, Brazil and Major Regions, 2009 ................................ ........................ 62 4 2 Income Distribution by Deciles, Brazil and Major Regions, 2009 ........................ 63 4 3 Mean Security Device Index by Tenancy and Housing Typology, 2009 ............. 64
9 LIST O F ABBREVIATIONS CPTED Crime Prevention Through Environmental Design IADB Inter American Development Bank I BGE Brazilian Institute of Geography and Statistics PNAD Brazilian National Household Sample Survey
10 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Arts (UN)SAFE AND (IN)SECURE AT HOME: VARIATIONS IN RESIDENTIAL SECURITY IN BRAZIL By Sarah Ruth Benton May 2014 Chair: Charles Wood Major: Latin American Studies Fear of crime, violence, and security are major issues for Brazil. A common response to fear of crime is through fortifying/securing their place of residence. This manipulation of the built environment is a physical and need or desire to secure self and family from perceived threats. The well documented proliferation of walls and high security gated communities in Brazil, in addition to the rest of Latin America, is evidence of a prevaili ng fear of crime. These well fortified homes and communities are both inspired by and inspire a fear of crime. This study describes the state of household security and fear of crime in urban areas in Brazil to create a baseline of data for use in future co mparative research, while also establish ing a relationship between fear of crime and r esidential fortification. Data from the 2009 Brazilian National Household Sample Survey (PNAD) is used and analyzed employing a variety of statistical techniques includin g ordinary least squares regression modeling. This research concludes that by controlling for sociodemographic variables, victimization, and housing related variables, fear of crime in the city has a positive and significant effect on degree of usage of re sidential security devices by heads of household s in urban areas of Brazil.
11 CHAPTER 1 INTRODUCTION Through my personal experiences and observations across many different cities in Brazil, I have seen how residents have implemented various security mechanis ms to fortify their homes against threats. As a frequent student and tourist without an automobile in Brazil, walking was my main form of transportation. I sometimes found walking around the city and its surrounding neighborhoods to be intimidating and sca ry due to the omnipresence of high walls, tall security fences, bars and security grills protruding from windows, and sometimes even walls lined with shards of glass to ward off intruders or would be criminals. Not only was I in a foreign country where the y spoke a foreign language, but the urban form was foreign too. I was not used to what I was seeing. Inasmuch as I was translating what people were saying to me in Portuguese, I ty Without a doubt, my impressions of the built environment around me altered th e way I experienced these cities and Brazil in general. These elements, highly visible in the built environment and experienced on a daily basis in Brazil, influence the thoughts and perceptions of the passerby, leading one to think there might be a crime problem or something to fear in these environs. They create a sense of anomie and isolation in addition to this fear, ultimately restricting me to several questions: Is crime really such an issue that these security measures need to be put in place? Do these residents feel safe? Are they living in fear? Was it
12 fear that made them secure their home? What else are Brazilians doing to protect their homes? Do certain group s of people feel more threatened, or have different responses to crime? why they are bot h essential to this study. In this case, the word segurana is central to the survey data on which this research is based. Curiously enough, the Brazilian Portuguese word segurana translates as either safety or security. Like many Portuguese words, there are no direct, one word translations into English which fully encompass the true meaning of the word. According to Merriam loss ). ) In my opinion, using the full English translation in the title of this study expresses a more nuanced understanding of the Brazilian Po rtuguese word segurana In English, a phrase that is used often is meanings, which is something that might be missing, or lost in translation, if I were to only use the wo rd segurana they may also have certain value and meaning attached to them by a native English speaker, and may offer a better understanding of the term segurana to a reader whose first language mig ht be Portuguese. The feelings of insecurity and lack of safety, along with the fear of crime (or perception of it) I experienced is not unique to tourists or foreigners, but is something
13 Brazilians experience as part of their daily lives as well. This fea r of crime stems from a multitude of factors and also spurs certain behavioral responses. As a tourist (not having a permanent home), I opted for avoidance strategies not going out alone, not walking on the street at night, avoiding certain streets or ne ighborhoods, etc. But, for a resident who has a home, a common, instinctive behavioral response or coping mechanism to this fear is to also use protective measures, that is, to fortify and secure their place of residence. This manipulation of the built env ironment is a physical and from perceived threats. The well documented proliferation of walls and high security gated communities in Brazil, in addition to the rest of L atin America, is evidence of the prevailing fear of crime felt by residents in urban areas. These well fortified homes and communities are both inspired by and inspire a fear of crime. gained by taking a look at the 2010 AmericasBarometer Survey 1 (LAPOP) results. Based on this public opinion poll from 2010, around one third of Brazilians believe that issues related to violence and personal security are major problems for their country, with violence being the most common response to this survey item. This is high compared to the United States, which, using the same survey from 2010, shows that only one tenth of Americans feel this way about their country. These polls further reveal that around 90% 1 The AmericasBarometer is a well known survey conducted throughout the Americas by the Latin American Public Opinion Project hosted by Vanderbilt University. I chose to use this data because it is available through and used frequently by students and faculty in the Center for Latin American Studies at the Univ ersity of Florida.
14 compared to 74% of Americans. When asked about issues or problems that worried and gangs (15%). Additionally, about one third of the Brazilian population fears being a victim of a crime in their neighborhood; while only 8% of Americans experience this same fear. Clearly, violence and crime are of relatively great concern in Brazil, with many people worrying and experiencing fear related to it. The general climate of insecurity is space. This disconnection and isolation, facilitated by the design of space and place, has led to certain attitudes. Data suggest that attitudes toward crime are increasingly hardening in Latin America, with more support for hard lined crime fighting (Prillaman, 2003). The traditional methods of preventing crime, for example by increasing the presence of the police, is not effective in Brazil due to widespread corruptio n 2 and requires more innovative and comprehensive measures. Nonetheless, in order to make recommendations for said measures, it is important to take a look at what is being done now, and by whom. Fear of crime is as much an issue as crime itself, as both have deleterious effects both for people and the state. The costs of crime, according to Prillaman (2003), conomic growth, undermining irrespective of actual incidence of crime, has also shown to negatively affect quality of 2 Transparency International ranked Brazil as the 69th most corrupt country in 2010, with a rating of 3.7 on a scale of 0 to 10 (with 0 being highly corrupt and 10 being very clean). For perspective, the United States was ranked 22 with a score of 7.1.
15 life due to its emotional, environmental, socia l, and economic impacts. Fear may cause stress and anxiety, a lack of trust in others, increased isolation and diminishing social connections. Not only can people suffer emotionally and mentally, but their behavioral responses can lead to changes in daily routines and activities, as well as in the environment (design of the built environment and defensive residential architecture). Economic effects of both crime and fear of crime are vast, including increasing expenditures by the public and private sector o n private security and security infrastructure. Moreover, individuals are willing to pay more for housing in order to feel safe, according to a recent IADB study (Vetter et al, 2013) This is money and capital foregone that could have been spent or invested elsewhere, like on education and health. Fear of crime, oftentimes unrelated to actual crime rates, may result in an over fortified built environment which is only reproducing these fears and feeding a fear cycle. The majority of the available studies and literature focuses on how fortified residences and gated communities inspire or reproduce fear, but this study suggests that fear is an important predictor of the need to fortify and secure the home in the first place. Although this has been suggested im plicitly, this relationship has not been tested explicitly. This study comes at a time when crime and human security in Brazil are more important than ever with the approaching World Cup and 2016 Olympic Games. These impending international mega events are major draws for international tourists and barricade themselves in their private fortresses, how are the multitudes of international visitors supposed to trust they wil l be safe? The world will be watching and listening for
16 their perceptions and impressions of Brazil and its cities. Clearly, addressing the issue of fear of crime and attempting to reduce it may improve the quality of life for Brazilians. It may also have positive social and economic effects, in addition to long term effects on the character of the built environment in Brazilian cities. Improving human security could lead to positive growth and development of the country in a time when Brazil is under the s potlight on the world stage. With these observations in mind, this study aims to achieve two goals. One goal is to paint a picture of the state of household security and fear of crime in urban areas in Brazil. Little research has been done on this topic, a nd it is useful to create a baseline of data for use in future comparative research. The final goal of the study aims to establish a relationship between fear of crime and residential fortification. The hypotheses to be tested are as follows: 1. A strong asso ciation exists between fear of crime and degree of home security/fortification when the net effect of victimization and various socio economic factors are controlled; 2. Brazilians use tall walls/fences more than any other security measure for their residence s; 3. Brazilians feel safest in their home, and less safe the further away they get from their home; 4. Differences exist in household choice of home security measures by location, demographics, victimization, and household characteristics. The following section gives context to the study area, introducing the reader to Brazil with a brief description of general characteristics followed by some background and history. This chapter closes with a section outlining the organization of the remaining chapters in this thesis.
17 Study Area Brazil is the largest country in South America, as well as the fifth largest country in the entire world (by territory and population). It is in a unique position to be bordered by all South American countries except for Chile and Ecuado r, while also containing the largest water resources in the world as well as some of the most biodiverse lands. Brazil covers 8,514,880 square kilometers, and is divided into 5 major regions: North, Northeast, Center West, Southeast, and South. The country has a population of 203,429,773 (a 2011 estimate), with a n annual population growth rate near 1.1%. The average life expectancy is 73.5 years. While 26% of the population is under 14 years of age, only 6.7% is over 65. On average, Brazilians attend 7.2 ye ars of school. The gender ratio is on average 1 to 1 (Central Intelligence Agency, 2012) Brazil has many large cities, with its largest cities being So Paulo and Rio de Janeiro in the highly populated Southeast region. The percentage of the population l iving in urban areas is at 87%. An overwhelming majority of urban populations have access to improved drinking water sources, at 99%, while only 87% have proper sanitation facilities access. The majority of the population is white, at 53.7% percent, with t he next largest category being shades of brown to black (Central Intelligence Agency, 2012) Brazil is a melting pot, with European, African, and Native Indian heritages mixed together as a result of hundreds of years of miscegenation starting in colonial and slavery times. This, along with the history and geography of the conquest, colonization, and internal migrations of Brazil, has resulted in a rich and varied culture across different regions of Brazil.
18 Brazil was established as a Portuguese colony for more than three hundred years. It remained under colonial rule until its independence in 1822, from whence it was ruled as a monarchy until two major events happened in the history of Brazil. The slave trade was officially and completely abolished in 1888, followed by the military establishing a republic in 1889. The demise of the monarchy and the rise of the Old Republic (in theory a constitutional democracy, yet effectively only the wealthy and literate few could vote) were paired with the rise of the cof fee trade and the rule of the wealthy land owning few. Later in the twentieth century, Getlio Vargas led a populist revolt in 1930. This marked the beginning of industrialization and of several decades of populist and military rule/dictatorships. In 1985, Brazilian military rule ended when power transferred into the hands of civilian leaders This transition to democracy was culminated by the establishment of the New Constitution of1988 which declared Brazil a Federative Republic. Brazil, like many Latin A merican countries, experienced a period of rapid growth and urbanization as a result of increasing industrialization since about the middle of the 20 th century. Many people migrated to commercial centers, like Rio de Janeiro and So Paulo, to find work. Th is exacerbated regional inequalities, while also resulting in rapid led to many problems related to health, housing, transportation, environmental degradation, educatio n, inequality, and last but not least, crime and violence. These are some of the many problems the Brazilian government has struggled to keep up with as the urban population grows. Crime rates have risen concurrently with increasing urbanization. These pro blems all pose threats to development, and crime is important
1 9 because human security is a key aspect of development, and greatly affects quality of life. Outline of Chapters This thesis contains five chapters. Following the Introduction, Chapter 2 is a lit erature review of pertinent studies and findings on fear of crime and responses to fear of crime, and the relationship between the built environment and fear of crime. Chapter 3 explains the methodology used to test the hypotheses, based on a variety of st atistical measures and tests. Chapter 4 presents the results of the data analysis, finally to be followed by Chapter 5, the Discussion and Conclusion. This final chapter discusses the results in further detail, explaining the implications they have for Bra zilian citizens and policymakers, as well as impacts in the fields of crime prevention and criminological theory, not to mention urban planning. Based on these analyses and results, recommendations are made for future research and policymaking.
20 CHAPTER 2 LITERATURE REVIEW Much of the extant literature on the relationship between fear of crime and the built environment in Latin America focuses on the building of walls, residential fortification, and the rise of gated communities. It is often stated that f ear of crime is the cause of this trend as far as the Latin American case is concerned (Caldeira, 1996; Caldeira, 2000; Coy & Pohler, 2002; Borsdorf et al 2007; Giglia, 2008; Silva, 2007; Vilalta, 2011a; Vilalta, 2012). There have been several qualitative studies carried out across the globe, in North America, Europe, South Africa, and Australia. These studies link fear with self protective home security measures, where residents of gated communities were interviewed or surveyed to find out their motivatio n for moving into the main motivation (Blakely & Snyder, 1997; Wilkinson, 1998; Atkinson & Flint, 2004; Lemanski, 2006; Blandy, 2007; Low, 2008). But, what about th ose whose action is socially and economically constrained? Much study focuses on those who can and do do so, be it because of income constraints, the potential loss of a social network, location of work from home, etc. There are many factors that can weigh in on this decision. Those that do not have the means or will to move, who also experience this same fear of crime because of increasing urban crime and disorder, would logically respond to this fear with easier, more accessible routes toward self protection. Home protection measures may be the more affordable and convenient response to fear of crime in urban areas. It is important to look more comprehensively at the meas ures people are taking in response to fear of crime as it pertains to crime prevention
21 measures at the household level. This is the gap in the literature that I suppose to address here in this study. Furthermore, is the relationship between fear of crime a nd home securitization being taken for granted? Is it just that the fear of crime is the logical and intuitive connection between high rates of crime and choosing to live in a secure residence? It is reasonable to expect that this relationship varies acros s cultures and place. The recent growing concern about the built environment and how it is shaping public and private lives, as well as reinforcing fear in our cities, furthers the need for this study. I aim to take a step back, to re analyze the original claim that fear of crime is what motivates people to secure and protect their home and loved ones, while also exploring the different types of protections used by different households. I use Brazil as my case study to add to the body of international liter ature on this topic. The literature review will cover the main issues, fear of crime and residential security and fortification, defining both and exploring the research done so far. Fear of Crime It has been well established that fear of crime is difficul t to define and operationalize, given that it is such a broad encompassing term that is influenced by many things. There is a debate on how to best measure it, and this is one of the main limitations to being able to draw general conclusions from the studi es available. Different operationalizations and measurements can lead to varying results. What has been agreed on is that there is no definitive answer, and the nature of survey questions and different factors influence fear of crime in different ways acro ss tim e, space, and cultures (Gerber et al 2010; Koskela & Pain, 2010; Ferraro, 1995). Nevertheless, researchers have consistently pointed to individual demographic characteristics,
22 (perceptions of) neighborhood social and environmental conditions, and vi ctimization as main influencers of fear of crime (Bannister & Fyfe, 2001; Scarborough et al 2010). The definition of fear of crime is often taken for granted, allowing for ambiguities e or dread or anxiety to crime or symbols that a person associates with crime. This definition of fear implies that some recognition of potential danger, what we may call perceived risk, is o (1995), different types of survey questions measure fear of crime based on different perceptions. People may perceive crime on a general or personal level in how it affects them, and across a cognitive affective perception continuum, with judgment, value and emotional perceptions of safety, risk, and crime. For example, a survey question asking how safe a respondent feels is related to their personal/cognitive judgment of risk. These perceptions highlight the many different ways in which fear of crime ma y be measured. Fear of crime is an urban issue according to Skogan and Maxfield (1981). They use a basic measure of fear of crime in their study on responses to crime, surveying respondents on their feelings of safety. The survey question does not even use the to be multi dimensional and reflecting other anxieties about urban life and the future independent issue Lab (2000) makes clear that fear of crime is a real emotion experienced by humans and warrants attentio n, no matter what spawns this fear or where it comes
23 from. He states that whatever the influences on this fear are, they are what influence the fear of crime is often treated as something negative, to be dealt with and reduced. It is generally accepted as something that negatively affects quality of life (Bannister & Fyfe, 2001; Foster & Giles Corti, 2008), but Jackson and Gray (2010) bring to light an important point that fear of crime can be a healthy emotional response to crime which agenc protective and precautionary behaviors. has shaped social relations and the direction of the built environment in Brazil (Caldeira, 2000). A need to self prot ect, and for some, retreat to gated communities and high rise condominiums, has been often discussed in the literature as a prevailing response to crime and the fear of crime, especially in Latin America. Nonetheless, the current trend of walls and boundar ies and private security guards to keep out both imagined and real crime and violence, as evidenced by the trend of highly secure gated communities and condos, simultaneously is inspired by and inspires a fear of crime (Davis, 1990; Caldeira, 2000; Atkinso n & Flint, 2004). It would be neglectful at this point to not mention the fear of crime feedback loop as it is one of the reasons fear of crime research can be so complicated. Fear of crime is not an emotional or cognitive response existing in a vacuum. Fe ar of crime is both an exogenous and endogenous self defeating factor in the processes related to the shaping of the built environment (Lee, 2001; Wilkinson, 1998). The origin of fear of
24 ifically locatable This can also be known as the causality dilemma, also which came first, and is especially pertinent to the relationship b etween residential for tification and fear of crime. Were people fearful first, leading to residential fortification? Or did pre existing residential fortification cause people to be more fearful and perceive a lack of safety? Consequently it can be unders tandable that causal relationships cannot be established H owever, the impacts of fear of crime and its interaction with other factors alone make it worthy of study. 1 The following section will explore actual victimization and vulnerabilities in contempora Crime Victimization and V ulnerabilities in Brazil Brazil has gained increasing global attention in recent years as an emerging power. However, crime and violence persist as probl ems plaguing the nation as well as individuals as they carry out their daily lives. The incidence of violence is real, and part p. 695). The AmericasBarometer (LAP OP, 2010) has information on victimization that is useful when exploring this topic. As far as victimization is concerned, nearly a third of Brazilians reported that either they or someone in their home had been victim of a crime in the previous year, 1 Because of the causality dilemma inherent in the relationships between the concepts used in this research, it is important to reemphasize that there is no causal ordering being established by the models used in this study. Associations an d correlations can be established between these concepts, but the shape and direction of causality cannot. It is both possible that fear of crime may result in residential fortification or that residential fortification may influence or lead to fear of cri me.
25 comp ared to about one fifth of Americans reporting the same (LAPOP, 2010). As far as the location of crimes is concerned, results were similar between Brazil and the US, with a little over one third of crimes in Brazil from the previous year occurring in the h ome; while a little under one third occurred in the home in the United States. Together, this data from 2010 identifies that crime and violence are prevailing problems in Brazil. But, different groups of people experience varying degrees of risk as it pert ains to victimization and a recent study done by Wood and Ribeiro (2013) sheds some light on this and their findings are discussed in the following paragraphs. Due to the nature of fear of crime and the multitude of factors influencing it, it is helpful t o look at current rates of various types of victimization. Historically, poor, young black men have been more likely to be a victim of crime than any other demographic group in Brazil. And crime victimization is, in effect, segregated in the same way as re sidential segregation in Brazil, between rich and poor, black and white (Rial & Grossi, 2002). vulnerable members of Brazilian society currently. In their study they analyze both P an American Health Organization data and Brazilian PNAD data on homicide, crime, and victimization. Reported homicide rates since 1995 have generally risen in Brazil, despite the existence of some troughs and crests. To give some perspective, since 1995, r ates began to spike back upwards starting in 2007. Regardless of these trends, the homicide rate in Brazil remains significantly higher than in these other countries, at a bove 25 per 100,000 people. From 1988 to 2009, assault rates have increased 0.4% nationally.
26 Aside from violent crime, which has impacts on fear of crime, property crime is se in urban property crimes than in rural property crimes. Property crimes suffered by victims, to include robbery and theft, have risen both nationally and within each region of Brazil, from 6.3% to 7.9% of the general population. The greatest rate increa ses are in the Northeast, North, and Center West. Much lower rates and slower growth are found in the South and Southeast regions. Their study also shows a reversing trend in vulnerabilities (Wood & Ribeiro, 2013) whereas in 1988 youth aged 16 24 were lea st likely to be victims of property crimes, in 2009 they are more likely to be victims. Currently, the least likely victims are over the age of 45. The trend also show that the more education one achieves, the less their likelihood of being the victim of a property crime. However, it is important to note that this remains the category (over 13 years of schooling) with the highest rate of victimization over time. This is paralleled by the 2009 findings on income, which are expected since income and education are often highly correlated with each other. Within the realms of gender and race, men are consistently more victimized than women, and from 1988 data on victimization Nonetheless, the rates of victimization across race and age appear to be evening out. All in all, the current portrait of crime vulnerability in Brazil is that of a wealthy, highly educated male, living in an urban area in the North region. Conversely, the least vulnerable Brazilian would be a white female over the age of 45, with less income and education living in a rural area of the Southeast (Wood & Ribeiro, 2013) These
27 vulnerabilities are important to identify as they may influence who is more likely to experience varying levels of fear of crime. These current victimization rates in Brazil are also instructive, as they support the claim that crime in urban areas is increasing in Brazil, and also have in fluence on the spread of fear. A Paradox: Actual Crime and Fear of C rime The focus of this study is on fear of crime, and the issue of the relationship between actual crime and fear of crime is complex and beyond the scope of this study. Nonetheless, it wa rrants brief attention to give the reader some context of the debate with Crime (1981), they acknowledge the existing relationship between crime and fear of crime, in that fear is an indirect consequence of actual crime. Actual crime is experienced by some, but the indirect experience of crime, or the vicarious experience of crime (knowing a friend or family member, or even a community member who was a victim), is experience d by or exposed to many. Beyond this, exposure to various sources of media (which often sensationalize crime and violence for ratings purposes) and perceptions of physical disorder in the environment are also indirect experiences of crime (Skogan & Maxfiel d, 1981). For this reason, fear of crime is experienced by more than just those who are in fact victimized, and this is why those who have not been victimized also take precautionary steps to prevent crime. In fact, in Brazil, most criminal activity and vi olence occur away from middle and upper class neighborhoods, yet security services and equipment are concentrated in these area s marked by less crime (Rial & Grossi, 2002). Interestingly enough, they also show that many people who are more vulnerable stil l do not take the precautions that they should (Skogan & Maxfield, 1981). Their
28 findings show that the relationship between crime and fear is not a simple one, and an area warranting further research and attention. Responses to Fear of Crime Fear of crime affects urban daily life. The varied responses have been extensively covered in the literature and across disciplines ( Blandy, 2007; Caldeira, 2000; Giblin, 2008; Lab, 1990; Lab, 2000; Lemanski, 2006 ). The fields of criminology, urban planning, sociology a nd anthropology have all delved into the topic of fear of crime in cities, how people have responded, and what feeds it. Fear of crime leads to changes and adaptations within the physical and social realms, like adding or increasing home security, moving i nto gated communities, avoidance behavior and changes in daily routine. Caldeira (2000) discusses crime and fear of crime in the Brazilian context, and their impacts on the built environment, how they daily/routine behavior and activities. She criticizes the recent trend of walled fortresses and daily life. People feel restricted in their movements, afraid, and controlled; the y go Encounters in public space become increasingly tense, even violent, because they are framed by people separation, discrimination, and suspicion are the new hallmarks of public lif e. (Caldeira, 2000, p.297) Beyond this, the UNDP Global R eport on Human Settlements (UN Habitat, 2007) supports this view, and reported on count walking the streets at night. A survey of 35 countries showed Brazil as the countr y with the highest percentage (around 70%) of respondents that felt unsafe walking home at
29 night. The United States ranked secon d to last, with only around 15% of people feeling unsafe. Clearly, urban fear and the responses people have to it affect the way they experience daily life. The focus of this research, however, is on those responses that affect the physical realm of the ho me, those which aid in protecting self and property. Gated Communities Gated communities in Latin America represent the extreme case of residential fortification. A gated community is defined as a walled or bounded residential area, either explicitly desig ned or retro fitted, with access control measures, a privatization of normally public spaces, and an underlying self governance structure (like an HOA or COA) (Blakely & Snyder, 1997). These walled, secure residential areas of today mimic the walled fortre 1990; Wilkinson, 1998), characterized by a ubiquitous obsession with and consumption Gated communities are not the exclusive focus of this study, but they do indeed fall into the general category of home security measures taken by Brazilian households. The trend towards gated communities, or condomnios fechados in Brazil, is not particular to Latin America. The rise o f gated communities started in the US and Great Britain, spreading across the globe and becoming popular in countries with urban crime problems. They became an easy way for developers to prey on those urban fears for their own profit, while providing a pro duct to consumers who sought security and safety, status, and certain lifestyles and amenities (Blakely & Snyder, 1997). The proliferation of gated communities in Latin America has been well documented, and what the literature consistently has in common is their citing of fear as the main motivator and
30 reason for people moving to gated communities ( Caldeira, 2000 ; Coy & Pohler, 2002; Giglia, 200 8; Vilalta, 2011a ). In Brazil, this began in the 1970s and since the 1980s has spread quickly and replicated thems elves across cities, big and small (Silva, 2007). changing patterns in the built environment in So Paulo, Brazil. The city was segregated along the center periphery model as a resu lt of the legacy of colonial times and geography of contemporary industrialization. After the economic restructuring of the 1980s and an increase in crime and violence, however, this pattern starts diffusing. Overlaid on the center periphery model is the d evelopment of highly walled urban clusters of the upper and middle classes dispersed throughout the lower income areas. There are still, however, those elite that refuse to abandon the traditionally elite center. The dispersed urban clusters tend to be in closed condominium developments which city opt for high security luxury high rises (Caldeira, 1996). The fear of crime is a strong motivator for segregation, and the upp er and middle classes attempt to maintain their separateness with boundaries from their social unequals, manifested in their choice of social groups and, thus, to enforce seg Caldeira studies segregation as an implication of this trend towards residential fortification, and does so in terms of income. Implied, however, is the racialized nature of this segregation. If fear of crime is a motivato r for fortification of residence, then it could be assumed that this occurs along monetary and racial lines. Caldeira explains that this new form of segregation displayed in fortified residences decreases the quality
31 of life for the nonwhite poor of Brazil and encourages fear and isolation for everyone. these things, or are not deserving. Ot her consequences of this new spatial order and segregation in the built environment are suggested to include incivility, crime, and a lack of a sense of responsibility. This loss of connection, or lack of physical interaction with others, is leading to an This loss of connection can be seen in the abandonment of public spaces by both the elite and the poor. The process of gating communities turns street private, controls new and old spaces, and redu Caldeira provides insights into the importance of studying the built environment, emocratization, social equalization, and society and a modern, developed city should be characterized by openness, freedom, and accessibility in both the social and physical form of urbanity. The organization of urban space is a result of social differentiation and separation according to Caldeira. Methods of social differentiation historically have changed from type of housing, to distance, and now to security and high walls The importance of security as a measure of social differentiation also promotes new forms of social discrimination. Caldeira would argue that, although physical distances are decreasing and therefore giving the impression of a more heterogeneous urban fo rm, social inequalities and differences are actually being reinforced by both physical and
32 imaginary walls (Caldeira, 1996). Increasing social inequalities and a prevailing fear of ation. newspapers, interviews, and observations in four major Brazilian cities. They analyze how the urban fear experienced by people of different social classes and ethnicities has transformed itself into a determining factor for the spatial and ethnic organization of the city, as well as the use of public space. With increasing urbanization and gro wing urban densities, the spatial proximity between rich and poor, white and black, living right next to each other in the urban landscape, instills fear in the white population, leading them to build walls, install gates and cameras, and hire private secu rity forces. They have abandoned public spaces, and even privatized streets for fear of exposure to crime. This urban fear has even led some elites to flee, or move away, to cities such as Florianpolis where crime rates are lower, yet they still live in t heir secure, gated condominiums. These abandoned public spaces have become stigmatized and dangerous for not only the elite, but for the nonwhites as well, because in some cases, not even the police will enter these public spaces, spaces that have become i gnored, like the poor, oppressed blacks that live there (Rial & Grossi, 2002). De Souza e Silva (2010) also comments on this increasing spatial proximity between urban residents and emphasizes how the architectural form of gated communities impacts the urb an form and the immediate neighborhood character and its residents. In her study on gated communities in Natal, Brazil, she finds that the tall walls
33 that are built for those seeking security and safety only encourage fear of crime in those who live outsid e and in close proximity to those walls. While these walls are socially divisive, they are not unique to the wealthy; gated communities have become a staple of some middle and low income communities. Their communities may not be a sophisticated as others, but they seek to control access and protect and secure from urban fears. De Souza e Silva supports the notion that these communities are created Coy and Pohler (2002) emphasize the impact like Brazil and Latin America. They note that gated communities in Brazil and Latin America tend to be much more fortified than their equi valents in the United States, with more advanced security technology. They expand beyond fear of crime as a reason for fortification, however, and cite other social and economic reasons for this popular desire to fortify. Some other social reasons include social differentiation and self segregation along with a desire to leave the chaos and disorder of cities. Economic reasons are rooted in both the supply side and the demand side, where developers seek to profit on ir ability to find cheap land on which to build these communities. Consumers desire a place where efficiency is maximized and all of the amenities of daily life are located in one place. In the end, however, these high tech fortified communities produce a false sense of security (Coy & Pohler, 2002). urban occupation are linked to race, and conceptuali zations of race are linked to urban
34 space, thereby forming a cyclical and self reinforcing structure. He uses ethnographic data to analyze a 2001 event when activists in the favela of Jacarezinho installed gates and security cameras, and the non favelans a nd police were outraged because they felt threatened. These types of security measures were reserved for the middle and upper classes for their gated condominiums to keep the poor and the blacks out. How dare a favela try to equal itself with an elite resi dential area? How dare they monitor the actions of the police in their favela? The media coverage of this event racialized the event, criminalized poor blacks, and encouraged typical racist stereotypes, and reinforced ates and cameras constituted a frontal challenge political will, action, autonom y, something the white elite reserve for themselves. mostly of a poor, working class black population. Residents of favelas have been excluded from the labor market quality education, and participation in the public and political spheres... Favelas are the historical and spatial product of this racialized argas, 2006, p.60). Vargas (2006) argues that the media have perpetuated negative racial stereotypes, and in effect, maintained the status quo of and fed the fears of an elite dominated society who reserve the right to residential Although race and segregation are not the focus of the research at hand, the ideas suggested in these works point at the need to look at how different groups within the population react differently to fear, how they simultaneously protect themselves,
35 distance themselve s, and socially differentiate themselves by accessing and employing residential security measures within the Brazilian urban built environment. Condomnios fechados may represent one extreme, the move to a fortified residence, but as an extreme it represen ts one end of a spectrum or continuum of residential fortification. Are security mechanisms and fortification just a reaction to fear, or are these being used by the elite to maintain dominance and power over other disadvantaged groups? How is this shaping the built environment? Are whites with higher incomes who have never been victimized more likely to use more security mechanisms, or more specifically, high walls and gates? Are security mechanisms used for status and power, and a result of feelings of in or re inspiring crime? Household Security Measures It may seem intuitive and therefore easy to accept on face value that fear (of crime) leads to certain self protective responses and behaviors. This relationship is often taken for granted within the literature, despite the fact that empirical evidence exists. This relationship has been well documented in the available research, also identifying victimization and demographic characteri stics as typical predictors of crime preventive behavior. People secure their homes based on an instinctive response to crime and fear of crime and an almost intuitive awareness or knowledge of place based crime prevention. to fear have been recognized within the urban planning field, mainly in North American and Western Europe, and become their own sub field of place based crime prevention, combining urban planning with crime and housing studies. The identification of differ ent crime preventive measures within the realm of the
36 household and the physical environment is rooted in the introduction of defensible space and CPTED. The principles of Crime Prevention Through Environmental Design, also known as CPTED arise out of Osca that certain kinds of space and spatial layout favour the clandestine activities of criminals. An architect [or urban planner], armed with some understanding of the structure of criminal encounter, can s (Newman, 1972, p.12). Newman focused mostly on target hardening of the residential environment where the residents can maintain a sense of ownership over their own safety and security, while highlighting f our important dimensions of crime prevention techniques through design: image, milieu, territoriality, and natural surveillance. He recognized that crime prevention measures in homes frequently were taken after the fact by residents, and expressed the need for architects, engineers and urban planners to be active in the process of creating defensible spaces in new housing developments to prevent crime from the get go (Newman, 1972). Since the 1970s, defensible space hasizing maintenance (Wilson & Kelling, 1982), environmen tal criminology (Brantingham & Brantingham, vironmental Design in 1991. According to Schneider and Kitchen (2002), design principles and strategies have the potential to reduce both crime and the fear of crime. The goal of CPTED is to deter crime by reducing opportunities for delinquency. The main p rinciples of contemporary CPTED draw from defensible space theory and updated into the following
37 dimensions of preventive measures: natural surveillance, boundary definition, access control, ownership of territory, maintenance, and the relation between lan d use and activity locations ( Clarke, 1995; Schneider & Kitchen, 2002 ). The conversation has been ongoing for years, but the literature has focused more on the effects of CPTED and not The above conceptualization of crime preventive measures is theoretical, and Lavrakas and Lewis (1980) set out to empirically test these dimensions, among others, for validity. Lavrakas and Lewis (1980) conducted a study examining citizen crime prevention Lewis, 1980, p. 255). These behaviors were pulled from various previous theoretical studies including those dimension of prevention suggested in defensible space and CPTED theory. Using frequencies, factor analysis and reliability testing, they analyzed results of four different surveys to create a reliable basis for conceptualization of crime preventive actions. Their findings support four different categories of crime preventive behavior, but to varying degrees: avoidance, access control, surveillance, and territoriality. The access control category includes techniques relevant to this current study, including home security and target hardening measures, like special window locks/bars and a door peephole. The take away from this study is the evidence of the need for treating home security measures together and not as independent and separate de pendent variables. Lavrakas and Lewis (1980) conclude that the use of a composite index of crime preventive measures is a more reliable and stable measure of
38 the concept of crime prevention as long as their internal consistency and reliability is measured first. Crime preventive actions can be divided into four categories according to Skogan and Maxfield (1981). These four categories form the basis of their research, being personal precaution, household protection, community involvement, and flight to the s uburbs. Of particular relevance to this study are their findings on household protection and its relation to fear and victimization. Their US study, focusing on households in urban areas, looked at several household level protections: target hardening tech niques (only including the existence of bars or special locks on windows), surveillance measures (leaving lights on, notification of the police when away, stop mail delivery or have someone bring it in, have a neighbor keep an eye on your home), and loss r eduction measures (property marking/engraving and insurance against property loss/theft). Correlations measured between household protections techniques and vulnerability (race, income, home ownership, and building size) and crime conditions (previous bur glary victimization, neighborhood problems, knew a crime burglary victim) find more vulnerable households do less to protect their household in response to fear. Less vulnerable households (in this case, home owning whites with higher incomes in single fam ily homes) take more precautions. These findings were consistent across the four cities in which this survey was conducted. Crime conditions, which includes past burglary victimization, is the only category which is a significant predictor of target harden ing techniques. Ultimately, Skogan and Maxfield (1981) suggest that socio economic factors may be better explanatory variables of household protective measures
39 since these present conditions which may limit or promote ability to take preventive action in t he household. Although this study only looked at two types of target hardening techniques, it represents an important point of departure for the development of research in this field. Often cited for his work on citizen participation in crime prevention, S teven Lab (1990) focuses his research on identifying the different domains and dimensions of crime prevention and the factors within each domain. He also focuses on who participates in crime prevention activities and how crime related factors like victimiz ation and fear influence this participation. He separates crime prevention techniques into five different categories or domains, isolating them to show the demographic differences between who participates in which types of preventive techniques. These five domains are personal access control, target hardening, personal security, sur veillance, and avoidance (Lab 1990). Target hardening, in this case, includes burglar alarms and property marking, while personal access control includes multiple locks and door peepholes. His findings depart from those of Lavrakas and Lewis (1980) in that territoriality is not a dimension, and target hardening and personal access control become separate dimensions of crime prevention. These dimensions are clarified for the purpo ses of indicating that different groups of people react to fear of crime in different ways. Additionally, Lab (1990) explores the relationships between these domains and demographics, fear, and victimization. He finds that demographic variables have minima l impact on victimization and fear of crime, while victimization and fear of crime also have minimal impact on crime prevention domains. He further states that it is these
40 demographic characteristics of respondents that actually have the direct effects on crime prevention activities. With respect to target hardening, the most likely type of person to take these precautions is an older, non homeowning, highly educated, female in a lower income bracket who has been the victim of a property crime. Personal acc ess behavior is most likely taken by wealthier, non homeowning, highly educated white people. The consistencies across dimensions and demographics are age, education, and homeownership as the strongest predictors for most prevention measures, with age bein g the strongest predictor. Race and marital status are found to have little influence on crime prevention participation. The fact that income was not a strong predictor, and that it was surprisingly not highly correlated with either education or home owner ship, economic variables are key influencers of crime prevention activity. It has been established that crime prevention practice is multidimensional, and Giblin (2008) goes on to test t he predictive ability of different measures on self protective behavior (avoidance and personal security measures). He accurately points studies, protective behavior is more often included as an independent variable in models explaining criminal victimization (or lack thereof); it is less frequently included as past studies to include awar eness of community policing, lifestyles, victimization, neighborhood conditions, and demographic characters as independent variables, to property crime. His findings ar e still instructive, in that they support those of Lab
41 (1990), showing that crime preventive behavior is multi dimensional. His findings show that predictors vary depending on the type of crime and type of behavior being studied. In an international study from Spain, researchers San Juan et al (2012) focus on two outlooks on fear of crime the influences of socio demographic influences (or vulnerability) and victimization. They recognize the environmental or ecological influences (incivility and disorder), but choose to focus on victimization and vulnerability, finding that age and sex do have effects on different self protection behaviors, as does recent victimization. Income and education were found to have no relationship with crime prevention techniques increasing security measures of the home). Older men were more likely to use active safety measures alone, and men were more likely to take no precautions at all. An overall preference for a combination of active safety measures with avoidance strategies was found in women. Their findings support the view that gender is an important consideration in studies on fear of crime and its response in self protective measures (San Juan et al 2012) Carlos Vilalt a has conducted several studies on fear of crime and its relationship to the use of public transportation, gated communities and housing typologies, and home security systems in Mexico City (Vilalta, 2011a; Vilalta, 2011b; Vilalta, 2012). He analyzes offic ial crime data and shows a lack of relationship between fear of crime at home and the use of home security systems. For Vilalta, home security systems consist of burglar alarm systems, special doors/locks, reinforced windows, dogs, tall walls, a security g uard or doorman, neighborhood surveillance systems, and informal surveillance systems with neighbors (Vilalta, 2012). He also finds that fear of crime
42 when at home alone is more likely to be experienced by poor young females. Although sts the reverse of this study, looking at the impact of security systems on fear at home, it is still instructive in fleshing out the relationship between home security and fortification and fear of crime. The directionality of the relationship cannot be d etermined, nor can causality. Vilalta also suggests that people may be pointlessly over fortifying their homes (Vilalta, 2012), especially in light of the fact that in an earlier study, he found that gated communities do not solve the problem of fear of cr ime. People who moved to gated communities and apartment buildings in Mexico City for the purposes of increasing safety and security did not experience lower levels of fear of crime than those that do not live in gated communities or apartments (Vilalta, 2 011a). In conclusion, the existing literature and research are helpful in guiding the current research on its endeavor to further flesh out the relationship between fear of crime and residential security and fortification in Brazil. It is especially useful when creating a reliable index of home security, as well as with determining the pertinent independent variables which may influence self protective behavior separate from fear of crime. The following section will explain the methodology for this study in detail, followed by a presentation of the results. The methodology chosen takes into consideration the literature discussed herein, while also taking into account the limitations of data available.
43 CHAPTER 3 METHODOLOGY Data To answer the research quest ions and to test the hypotheses posed require data on crime, fear of crime, victimization, behavior and the built environment. Fortunately, there are data available that is conducive to exploring this relationship further, specifically that of the Brazilia n National Household Sample Survey (PNAD Pesquisa Nacional por Amostra Domiciliar ) carried out by the Brazilian Institute of Geography and Statistics (IBGE). IBGE conducts national sample surveys every year in Brazil that include extensive information on socio economic indicators of individuals and households. The PNAD collects information on households in order to study and monitor socioeconomic development in the country. For the purposes of this research, the PNAD for 2009 provided the data needed on f ear of crime and home security, along with pertinent socio economi c 1 indicators. In 2009, the PNAD included not only questions on victimization and justice, but also, for the first time, a series of questions on household security devices. This dataset fro m 2009 is advantageous due to the large number of cases available, with 399,387 respondents, and 153,837 households. The sampling frame for the PNAD follows three stages of selection based on municipality, census tract, and housing units. Using proportiona l probabilities derived from the 2000 population census, housing units were selected at random to form the sample for this survey. Because security devices are a household level indicator, I 1 The terms socioeconomic and sociodemographic are used here interchangeably.
44 limited the analysis to heads of households in urban areas in ord er to avoid the overrepresentation of larger households in the results. This data are available at the national, regional, and state levels, in addition to being available for nine metropolitan regions in select states. This data set includes a plethora of variables that are useful when exploring socio demographic factors affecting complex issues. The study design for this research is cross sectional and non experimental, using secondary data available from the 2009 PNAD. In order to limit the sample to ho useholds rather than individuals, the variable the household 2 filtered by selecting only the cases where heads of households in urban areas of Brazil to be used for further analyses. Variables The key concepts in this study are reside ntial fortification and fear of crime in the city In order to test the hypothesis that residential fortification is influenced by fear of crime in the city it was necessary to find a way to directly measure these concepts. Fear of crime in the city is th e primary independent variable, and using the information 2 possible ambiguities in the wording chosen for this question in the survey are recognized but the researcher has faith in the expertise and thoroughness of the and offends femini sts.
45 from the PNAD 2009 Survey, was logically operationalized as the converse of Feelings of Safety in the City. Primary Independent Variable: Fear of Crime The PNAD survey did not ask respondents explic itly about their fear of crime. However, an acceptable surrogate for this concept is the question of feelings of safety. The survey did ask if respondents felt safe in their home, in their neighborhood, and in their city. By recoding the responses to these questions into Safety variables (1 = feels safe, 0 = does not feel safe), it was then possible to recode these into their opposites in order to represent fear. It is logical to assume that if a person does not feel safe in their city, then they feel fearf ul of their city. If a person does not feel safe in their neighborhood, then they must feel unsafe or fearful of crime in their neighborhood. Therefore, the aforementioned variables of Safe_home_01, Safe_neighborhood_01, and Safe_City_01 were recoded into their opposites: Fear_home_01, Fear_neighborhood_01, and Fear_city_01, where equaling 1 meant they did experience the fear in the specified location/area. An additional variable was then created to represent respondents who felt fearful in every location Fear_Everywhere_01, where the value of 1 was assigned if the respondent had answered that they did not feel safe in any of the locations. Fear_city_01 is as the primary independent variable in this study. Dependent Variable: Index of Residential Fortifica tion In the case of residential fortification, the dependent variable, there is no single indicator or variable that can fully measure it. In order to operationalize residential fortification, it was therefore necessary to combine several variables from th e PNAD
46 survey into a composite index of residential fortification, which will be explained henceforth. The PNAD survey from 2009 included a new section on Security Mechanisms/Devices, within which was asked one question with seven yes/no sub questions, all relating to whether or not the residence had a certain security mechanism or mechanisms in order to increase security/safety. 3 These seven security device(s) options are, in general terms, the following: peephole, extra locks/bars, grates/grills, tall or electric fence/wall or security alarm, video camera, private security guard or gated entry, and other. Table 3 1 lists and describes in more detail the security devices actually used in the analyses, along with the other variables. 4 After choosing these se ven variables for the index, it was important to perform a 3 Some might argue that the validity of the index is in question due to the installation and usage of certain security measures as privacy enhancers/providers. The question in Para aumentar a segurana, existe neste dom iclio... ( Instituto Brasileiro de Geografia e Estatistica, 2010). The English translation for this is It is true that many of these devices have multiple uses; however, the survey question is ph rased in a way that reduces any bias or questions of validity. 4 It is important at this juncture to discuss the limitations of the questions on security devices posed in the PNAD survey. As can be noted in the list above, each sub question under the secu rity device category on the survey asked about a different the items in each list t he respondents has, or how many they have. This is inherently problematic, because it allows for ambiguous data, and undermines the use of the security index. The security device index assumes that each sub question represents one security device. It is di fficult to do this any other way since the data is not available or measurable at the level of detail necessary within each question. Having more data would provide a more accurate count or index of security devices. However, this study makes use of the da ta available despite its inherent biases, and works within the constraints of the survey since it is the only data available on this subject.
47 coefficients range from 0 to 1, with v alues closer to 1 having greater internal consistency reliability. for this index equals 0.558 for the 7 item index, but improves to 0.614 if Grates_01 and sec_other_01 are excluded from the index. This allowed for the highe The most widely the acceptable range goes lower, since there is no standard. Generally speaking, anything below 0.5 would be unacceptab le, and anything greater than 0.60 could be acceptable (Peterson, 1994) Regardless of the indications of this measure, the face validity of these variables as an indication of residential fortification within the specialist community and the literature is reason enough to employ this index for this study. The residential fortification index was therefore computed into a 5 item index by adding the aforementioned security device variables (Olho_01, Ext_lock_01, Fence_Wall_01, Camera_01, and Pvt_security_01) together into an index variable entitled Sec_devices_index. This index is a count of the security devices in a household 5 Factor analysis was conducted in order to ascertain that these items in the index did indeed measure only one, not more, concepts or underlying constructs. Control Variables The regression models include control variables in order to remove the effects of other variables thought to influence residential fortification. The control variables comprise of age, sex, education, income, race, geographic location, type of residence, 5 A weighted index of security devices was created and tested as a result of factor analysis. This index produced similar results to the non weighted index, and therefore was determined to be unnecessary.
48 tenancy/type of ownership, and three variables on actual experience of victimization. I chose these variables based on the literature and past research, as these are likely to prot ect and their vulnerability, their likelihood of feeling unsafe or being a victim. Age is a continuous variable, operationalized as how many years a person has lived. V8005 is the survey variable for age which was copied into a new variable, Age. Sex is a categorical variable, operationalized as the respondent self reporting as Sex of the respondent, male or female, is found in V8005, which is recoded into the new variable Gender. A dummy variable was then created for later use in regres sion analysis, named Gender_01, where male = 1 and 0 = female (reference category). Education is a continuous variable, operationalized as years of schooling completed by the respondent (V4803 ). This variable was copied into a new variable named Years_Scho ol, to represent education. Income is a continuous variable, operationalized as monthly household income per capita measured in reais (Brazilian currency). Househo ld income per capita was obtained from the variable V4621, which was then computed into the n ew variable Income_percapitaHsld. The PNAD Survey offers five categories of race operationalized as color, by which the respondent may choose to self identify 6 The five categories are White, 6 Some might argue that Color (of skin) is an unjustifiable or inaccurate measure of ey, and Cor ou raa argued does indeed carry the most face validity as, an operational social (rather than genetic) definition of race.
49 Brown, Black, Yellow, and Indian. The variable Color was comput ed into these five categories from V0404. It was necessary to recode these into multinomial dummy variables, so that each Color category could be compared in reference to the White category during regression analyses. White was recoded into white=1 and not white=0. Black was recoded into Black=1, and not black=0. This pattern was repeated for the other colors Geographic location is operationalized as region of residence, which is a categorical variable with five options. The survey does not directly ask the respondent what region they live in, so this must be computed based on the other variables. The survey includes data on the state where the respondent lives, with the variable UF. This variable is recoded into the new variable State, which is then reco ded into Regions_5Cats by dividing the 27 federal units into their respective regions (1=North, 2=Northeast, 3=Southeast, 4=South, and 5=Center West). Next, five multinomial dummy variables were created for these regions so that the North region could be u sed as a reference category when performing regression analyses (e.g. Northeast=0 is respondent does not live in the Northeast and Northeast=1 if respondent does live in the Northeast). Type of residence is operationalized as whether or not a respondent l ives in a house or in an apartment. The choices for the survey question associated with variable V0202 were house, apartment, or room. Room is excluded here because, besides only representing a mere 0.4% of the sub sample respondents, it also is not the ty pe of residence under study here. V0202 was copied into the new variable House_Apart_Room, and then this variable was used to compute two new dichotomous
50 variables, House_01 (where house=1, and not in a house=0) and Apartment_01 (where apartment=1, and not in an apartment=0). Tenancy was also an important variable because it represents investment in the home, operationalized by the type of ownership of the residence, found in variable V0207. The options for this survey question included Owned and Paid, Owne d and still paying (interpreted as mortgaged), rented, gifted by employer, gifted other, and other. categories owned/paid, owned/mortgaged, rented, gifted (the two original gi fted categories are combined here), and other. These were then recoded into multinomial dummy variables (OwnPaid_01 = 1 is owned and paid, OwnPaid_01=0 not owned and paid, etc.) so that owned/paid could be used as the reference category when carrying out t he regression analyses. The last variable to be created was victimization. Victimization is a key variable in ear and safety. There were three variables from the survey chosen to be used in this research for the purposes of operationalizing victimization: V2903 which represented respondents who were victims of attempted robbery in the previous year, V2911 which re presented respondents who were victims of theft (no violence/threat) in the previous year, and V2904 which represented respondents who were the victim of a robbery which employed violence or threat in the previous year. Each of these variables was recoded into a dichotomous variable, where 1 = victim and 0 = not a victim.
51 It is important to note here that these questions only took into account victimization events that occurred in the previous year, and not before. The form of the question introduces a pot ential bias inasmuch as respondents may still experience the emotional effects of being a victim of a crime even though the incident may have occurred in an earlier time period. It also does not take into account whether the respondent knew someone, be it a family member or close friend, who had been a and fear. Notwithstanding, this is another shortcoming to this data set that must be acknowledged and considered wh ile carrying out this study and analyzing the results. Data Analysis Using SPSS statistical software version 17.0, I performed several descriptive and frequency analyses on several variables in order to create a context within which to couch further analys es. Furthermore, the method of analysis I used to test the main hypothesis was ord inary least squares regression. I chose this method because it allow s the simultaneous inclusion of several independent variables, both categorical and continuous This flexi bility was key in order to observe the predicted relationship between and effect of Fear of Crime in the City on Residential Fortification wh ile controlling for several other indicators. Several ordinary least squares regressions were performed to test th e ability of the primary independent variable to predict residential fortification net the effects of socioeconomic factors. Model 1 regresses the level of residential fortification on fear of crime in the city to ascertain its predictive validity alone. Model 2 adds victimization variables to the model. In Model 3, several socioeconomic control variables were added to the regression model to determine if the primary independent variable (fear)
52 continues to have an effect on residential fortification net of the effects of social status. The c ontrol variables included in this model were age, gender, income education, geographic location (region), and race. Dummy variables were used for gender (with female as reference category), race (with White as refer ence category), and geographic location (with North as reference category). Finally, Model 4 added the variables related to the home tenancy/type of home ownership and type of residence. Dummy variables were used for both of these variables with Owned/pa id and apartment as reference categories, respectively. To test the hypothesis regarding the most prevalent form of residential fortification (the most prevalent choice of residential security device), the data was analyzed with a simple frequency of respo nses to the variables regarding existence of different security measures in the home. To address the hypothesis on where fear is experienced the most, a frequency distribution was done on the Fear variables for in the home, in the neighborhood, in the city and everywhere. The results of these tests will be discussed in the next chapter
53 Tables Table 3 1 Variable Descriptions Concept Variables description Range Dependent Security devices Composite index Peep hole; an opening in the door for viewing; a security chain on the door; intercom (yes=1)(no=0) Extra locks, security bars on door/window (yes=1)(no=0) El e ctric fence; wall or grate/fence higher than 2 meters or with shards of glass or barbed wire; electronic security alarm (yes=1)(no=0) Video cam era (yes=1)(no=0) Private security; gated entry (yes=1)(no=0) =0.614 0 5 Primary independent Fear of crime Do you feel safe in your city? (yes=1)(no=0) 0 1 Control v ariables Age In years 14 109 Sex Gender (male=1)(female=0) 0 1 Education Years of school completed 1 17 Income Per capita household income in reais 0 94,669 Race By color: White, Black, Brown, Yellow, Indian (yes=1)(no=0) 0 1 Geographic location By region: North, Northeast, Center West, Southeast, South (yes=1)( no=0) 0 1 Type of residence House or Apartment (yes=1)(no=0) 0 1 Tenancy Type of ownership/ownership status: owned, mortgaged, rented, gifted, other (yes=1)(no=0) 0 1 Victimization Were you the victim of [an attempted robbery / theft / robbery with viol ence or threat] in previous year? (yes=1)(no=0) 0 1
54 CHAPTER 4 FINDINGS General Results This chapter present s the results of the data analysis as set forth in the previous section The sample included 103,963 valid cases of heads of households in urb an areas of Brazil. The age range of respondents (all heads of households) was 14 109 years of age, with the average respondent being nearly 48 years old. Males constituted 63.7% of respondents, with females only representing 36.3%. Education of the respon dent ranged from 1 to 17 years of school completed, with 8.5 years being the average. A comparison of means for years of school completed can be found in Table 4 1. This table shows the South and the Southeast as the regions with the highest average years of school completed, and the North and the Northeast with the lowest average years of school completed Figure 4 1 shows that, following regional demographic trends, the North and Northeast have greater percentages of their populations with little to no ed ucation completed (these regions have high rates of illiteracy), while the South and Southeast report higher percentages of their populations completing more years of school. The Center West reports a high percentage of highly educated respondents, but thi s is likely due to the influence of the capital, Braslia, which is located in the Center West region and is the center of governmental and administrative powers for the country. Monthly income per capita in reais ranged from R$0 to R$94,669. The mean mont hly income per capita was R$832, which was between 1 and 2 minimum salaries in
55 2009 Figure 4 2 shows the regional distribution of income by deciles, demonstrating that wealth is concentrated in the South and Southeast, and poverty is concentrated in the N orth and Northeast. A comparison of means of monthly income per capita are shown in Table 4 2, where the highest average is reported in the South, and the lowest in the Northeast. The majority of respondents were White, representing 51.1% of the population Afro Brazilians represented 48.1% of the population (combined, 8.5% Black and 39.6% Brown), with Yellow representing 0.6% and Indian representing 0.3%. The majority of respondents lived in the Southeast at 47.8% of respondents, followed by 22.8% living i n the Northeast. This is consistent with the pattern of population distribution in Brazil, with the Southeast being the most populous region followed by the Northeast. 15.3% of respondents lived in the South, 7.6% in the Center West (including the Federal District of Braslia), and 6.4% in the least populated region of the North. As shown in Table 4 3, the majority of respondents lived in houses as opposed to apartments, at 87.5% and 12.1%, respectively. The regions of the Southeast and the South stand out, as is evidenced by observing columns 7 through 10 in the table. The percentage of respondents living in houses in the South and Southeast is lower than the national average, and the percentage of households living in apartments is higher than the national average. Furthermore, the majority of respondents (68.4%) also owned and had already fully paid for their residence. The next largest group rented their residence, at 19.4%, followed by 7% who had been gifted their residence and not paid for it. Those who owned but were still paying on the residence (assumed to mean mortgaging their
56 residence, whether through a private or public loan) were 4.7%, followed by an Table 4 4 shows that the maj ority of respondents had not been a victim of any type of property crime. 86.8% were not victims of an attempted robbery, a robbery with threat or force, or a theft, in the year previous to the survey. Of those, nearly half, 49.2%, still felt unsafe in the ir city. Nonetheless, 13.2% had been victims, of which 70.7% did feel unsafe in their city. 7.9% had been victims of an attempted robbery in the previous year, 4.6% had been a victim of a robbery where threat or force was used, and 6.2% had been a victim o f theft. Not surprisingly, the North had the highest percentage of victims of property crimes, followed by the Center West (column 1), both above the national average. However, curiously, the South had mostly above average victimization rates (row 5) acros s types of property crimes, while the Southeast had percentages below the national average for victimization within each type of property crime. It would seem that actual victimization increases the likelihood of a person feeling unsafe in a city, which co uld be associated with their usage of security devices. This relationship is explored further in the later discussion of the regression analyses. Regarding where Brazilians feel the least safe or fear their environment the most, the most common response wa s, as predicted, in their city Table 4 5 shows the distribution of the responses to the safety questions in the PNAD survey, broken down by region. I nterestingly, 43.9% of respondents reported not feeling any fear in their home, neighborhood, or city, rep resenting the largest category of where fear is experienced (column 2, row 5) However, as can be seen in column 2 across the board, people are more fearful the further they are away from their home. Most respondents
57 reported feeling fear in the city, les s in the neighborhood, and less in the home Regardless of the fact that only around 13% of the population was actually victim of a property crime in the previous year, the percentage of the population experiencing fear somewhere is much higher than this p ercentage. As distance from the home increases, so does fear of crime. People are less fearful in their home, and more fearful in the city, therefore supporting the theory that fear of crime has latent and residual effects. Table 4 6 shows the distribution of usage of security mechanisms. It is curious to note that 38.9% of respondents report not using any sort of security mechanism or device to make them feel safer at home 1 As far as security device preference is concerned, there is an overall preference for grates/grills on windows and doors of residences. At both the national level as well as individually within each major region (row 3), around 40 % of respondents reported having this category of security devices (grates/grills). Grates are followed by a magic eye or peephole at 23.3% nationally (row 1, column 2). However, this is only the second highest category in the South and Southeast. In the North and Northeast, the second highest frequency of choice of security devices is the extra locks category. Tall walls and fences is the third most frequent category, at 21.4% nationally (row 4, column 2). Interestingly enough, this is actually the second highest ranked category in the Center West, at 24.8% (column 12, row 4). Proportionately, the highest perce ntage of a regional population using walls and 1 It is possible that the respondent owns/us es some other form of residential security device. However, the security mechanisms posed in the questions in the 2009 PNAD survey are fairly comprehensive. It does not include questions about whether the head of the household owns a weapon or a gun that is kept in the home. These are other forms of security that are not taken into account here due to the fact that this research is focused on physical/structural security elements and the manipulation of the built environment and the residence.
58 fences for security purposes is in the Center West region. The Southeast is the only regi on matching the national trend. P rivate security guards, devices nati onally, all below 10% frequency of positive responses, reason being that security cameras and private security guards are higher end security mechanisms. These cost the household more money and higher investment in security infrastructure, and typically wo uld require a household within a certain socioeconomic level. Consequently, the South and the Southeast are above the national average for population percentage (within region) employing private security guards, gated entries, and video cameras. The securi ty device index allows values on a scale of 0 to 5, given that there are 5 items in the index. 38.9 % of respondents had no security devices according to the categories offered in the survey, which is fairly comprehensive in its coverage of security device options, in the opinion of the researcher. The average security device index value indicates that households had 0.7755 categories of security devices at their residence; however, this value is depressed due to the number of respondents scoring 0 on the i ndex. Table 4 7 shows a comparison of means of security device index scores across regions, which is useful for assessing differences in general residential fortification trends The South has the highest mean of 0.8977, followed by the Southeast, both be ing above the national average. The North has the lowest mean score, at 0.5324, which is somewhat surprising given that the North has the highest reported victimization rate. However, income and education play a role in this. The interaction of all of thes e
59 independent variables is explored in the following discussion of the results of the ordinary least squared regression modeling. Multivariate Analysis A strong association exists between perceptions of safety and level of home security/fortification when the net effect of various socio economic factors and related indicators are controlled. In other words, fear of crime in the city is a strong predictor of intensi ty of residential fortification, as shown through the results of the following regression mode ls in Table 4 8. In Model 1, the proportion of variance explained in level of residential fortification by fear of crime in the city is only 1.3%. By adding victimization variables to the equation in Model 2, the proportion of variance explained increases to 1.5% (an increase in 0.2%). 0.2% is the amount of increase in explanatory power associated with victimization, indicating that fear of crime in the city is a stronger explanatory variable in the model than victimization, however, both are statistically significant and have a positive effect on the dependent variable. In Model 3, socioeconomic variables are added to the equation as control variables. Adding these increases the proportion of variance explained to 21.8%, which means 20.3% is the explanatory power associated with these variables. Finally, by adding house related control variables (tenancy and type of residence), the variance explained increases to 29.9% by this model. The strength of the model was improved by 8.2, which is the amount of incre ase in explanatory power associated with these housing related variables. Considering the fourth model, the results are as expected. Increases in age, years of education, and income increase the index of residential fortification, which
60 corresponds with th e findings on the average victim of a crime in Brazil. It could be logical that since the type of person most at risk in Brazil is a male with increasing age, income, and education, that this would also indicate the type of person to increase fortification in the home. The type of residence is the best explanatory variable in the equation. By comparing standardized Beta coefficients, living in a house decreases the residential fortification index the most. Income and education are the next best explanatory variables. Not surprisingly, Color is also a telling factor. Blacks have fewer security devices than Whites, and Browns have less but not as few as Blacks, while Yellows have more. Because a greater percentage of the population in the South and Southeast l ives in apartments, and given that the type of residence is a strong predictor of residential fortification, it would be expected that there would be a higher incidence of security device usage in these regions as well. In fact, as previously mentioned, a comparison of means for the index of residential fortification demonstrates the highest mean values for the South and Southeast (see Table 4 7). Further investigation into the type of residence shows that the index of residential fortification varies acros s tenancy and housing typology combinations. A breakdown of mean values for the security device index by housing typology in conjunction with tenancy status, at the national level, is demonstrated in Figure 4 3. Quadrant II shows the lowest mean index valu e for residential fortification (0.5056) by those who rent a home, compared to Quadrant III which shows the highest mean value of 2.2439 for owned apartments. Households in apartments have higher
61 mean index values than those in houses. These trends are ref lected consistently at the regional level. Model 4 shows a positive relationship between residential fortification and fear of crime in the city. This model is strong and statistically significant, explaining 29.9% of the variance in residential fortificat ion. Each independent variable in this model, except model. It can be concluded then, that the null hypothesis can be rejected (fear of crime in the city has no influence or effect on residential fortification), and the proposed hypothesis can be accepted: controlling for sociodemographic variables, victimization, and housing related variables, fear of crime in the city has a positive and significant effect on de gree of usage of residential security devices by heads of household s in urban areas of Brazil.
62 Figures Figure 4 1. Education, Brazil and Major Regions, 2009 0 5 10 15 20 25 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 % Population Years of School Completed North Northeast Southeast South Center-West Brazil
63 Figure 4 2. Income Distribution by Deciles, Brazil and Major Regions, 2009 0 5 10 15 20 25 1st 2nd 3rd 4th 5th 6th 7th 8th 9th 10th North Northeast Southeast South Center-West Brazil
64 Figure 4 3. Mean Security Device Index by Tenancy and Housing Typology, 2009
65 Tables Table 4 1. Mean Years Education Completed, Brazil and Major Regions, 2009 Location Mean N Std. Deviation Brazil 8.4548 103963 4.66187 North 8.1481 6699 4.65075 Northeast 7.3816 23745 4.81467 Southeast 8.8180 49707 4.54667 South 8.9667 15868 4.50099 Center West 8.6259 7944 4.71020 Source: PNAD Survey, 2009 Note: Sample restricted to Heads of Household in Urban Areas Table 4 2. Mean Monthly Income Per Capita in Reais Brazi l and Major Regions, 2009 Location Mean N Std. Deviation Brazil 832.63 100541 1411.08944 North 584.37 6546 945.43332 Northeast 568.48 23400 1099.49888 Southeast 935.08 47290 1491.70172 South 965.97 15550 1734.92236 Center West 947.07 7756 1734.92236 Source: PNAD Survey, 2009 Note: Sample restricted to Heads of Household in Urban Areas
66 Table 4 3. Type of Residence Brazil and Major Regions, 2009 Brazil North Northeast Southeast South Center West Type of Residence N % of S.P. N % N % N % N % N % (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) House (1) 90749 87.5 6358 95.1 21404 90.4 42221 85.1 13635 86.1 7131 89.9 Apartment (2) 12585 12.1 256 3.8 2191 9.3 7238 14.6 2187 13.8 713 9.0 Source: PNAD Survey, 2009 Note: S ample restricted to Heads of Household in Urban Areas Note: S ample Population = 103,963; Valid cases 103,758; 204 missing Note: N = Number of Cases, % of S.P. =% of Sample Population %=% of N for region
67 Table 4 4 Victimization, Brazil and Major Regions, 2009 Victim (N =103964) Victim Attempted Robbery (N=103963) Victim Robbery with Force/Threat (N=103962) Victim Theft (N=103964) (1) (2) (3) (4) (5) (6) (7) (8) Location %Yes %No %Yes %No %Yes %No %Yes %No Brazil (1) 13.2 86.8 7.9 92.1 4.6 95.4 6.2 93.8 North (2) 1 8.1 81.9 10.9 89.1 7.6 92.4 8.2 91.8 Northeast (3) 14.5 85.5 7.8 92.2 5.8 94.2 6.1 93.9 Southeast (4) 11.0 89.0 6.9 93.1 4.0 96.0 4.9 95.1 South (5) 14.4 85.6 9.0 91.0 3.5 96.5 8.1 91.9 Center West (6) 15.9 84.1 9.9 90.1 4.5 95.5 8.7 91.3 Source: PNAD Survey, 2009 Note: S ample restricted to Heads of Household in Urban Areas
68 Table 4 5. Distribution of Fear, Brazil and Major Regions, 2009 Brazil North Northeast Southeast South Center West Fear N % of S.P. N % N % N % N % N % (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Home (1) 24703 23.8 2168 32.4 6069 25.6 11193 22.5 3170 20.0 2103 26.5 Neighborhood (2) 37537 36.1 2999 44.8 9548 40.2 17082 34.4 4831 30.4 3077 38.7 City (3) 54063 52.0 3800 56.7 13506 56.9 25814 51.9 7057 44.5 3885 48.9 Everywhere (4) 20649 19.9 1767 26.4 5244 22.1 9421 19.0 2465 15.5 1752 22.1 Nowhere (5) 45600 43.9 2519 37.6 9372 39.5 22037 44.3 8028 50.6 3645 45.9 Source: PNAD Survey, 2009 Note: Sample restricted to Heads of Household in Urban Areas Note: Sample Po pulation = 103,963 Note: N = Number of Cases, % of S.P.=% of Sample Population, %=% of N for region
69 Table 4 6. Distribution of Security Mechanisms, Brazil and Major Regions, 2009 Brazil North Northeast Southeast South Center West Sec. Mechs. N % of S.P. N % N % N % N % N % (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Peephole (1) 24165 23.3 609 9.1 4185 17.7 13538 27.3 4231 26.7 1602 20.2 Ext_lock (2) 21108 20.3 1571 23.5 4786 20.2 9346 18.8 3896 24.6 1509 19.0 Grates (3) 42014 40.5 2683 40.1 9315 39.3 21095 42.5 5437 34.3 3484 43.9 Fence_wall (4) 22197 21.4 1049 15.7 4195 17.7 11230 22.6 3759 23.7 1964 24.8 Camera (5) 5119 4.9 101 1.5 576 2.4 3151 6.3 888 5.6 403 5.1 Pvt_Sec (6) 7881 7.6 230 3.4 1621 6.8 4062 8.2 1444 9.1 525 6.6 oth er (7) 1518 1.5 47 .7 281 1.2 809 1.6 305 1.9 76 1.0 None (8) 57288 55.2 4155 62.2 13717 57.9 26803 54.0 8200 51.8 4414 55.7 Source: PNAD Survey, 2009 Note: Sample restricted to Heads of Household in Urban Areas Note: Sample Population=103,759 ; 204 syste m missing Note: N = Number of Cases, % of S.P. =% of Sample Population %=% of N for region Note: Peephole = peephole; an opening in the door for viewing; a security chain on the door; intercom ; Ext_lock = extra locks, security bars (of iron or wood) on the door/window against break ins ; Grates = Grates bars/grills on window/door ; Fence_wall = Electric fence; wall or grate/fence higher than 2 meters or with shards of glass or barbed wire, electronic security alarm ; Camera = Video camera ; Pvt_Sec = Private se curity; gated entry
70 Table 4 7. Means of Security Device Index by Region 2009 Regions Mean N Std. Deviation North .5324 6684 .81478 Northeast .6487 23683 .94217 Southeast .8328 49622 1.13558 South .8977 15839 1.17563 Center West .7571 7929 1.0764 4 Total .7755 103759 1.08303 Source: PNAD Survey 2009 Note: Sample limited to Heads of Household in urban areas
71 Table 4 8. Level of Residential Fortification Regressed on Fear of Crime in the City and Selected Indicators: Brazil, 2009 (OLS regression coefficients) Independent V ariables Model 1 Model 2 Model 3 Model 4 (1) (2) (3) (4) Constant (1) .636 .626 .394 .765 Age (2) .006 .005 Sex Female (ref) (3) Male (4) .006* .022 Years of School (5) .065 .050 Per Capi ta Household Income (6) .000 .000 Race White (ref) (7) Brown (8) .151 .128 Black (9) .182 .169 Yellow (10) .445 .367 Indian (11) .077* .068* Region of Residence North (ref) (12) Nor theast (13) .152 .104 Southeast (14) .120 .065 South (15) .159 .108 Center West (16) .101 .092 Type of Residence Apartment (ref) (17) House (18) .982 Tenancy/Type of Ownership Owned/Paid in Full (ref) (19) Mortgaged (20) .201 Rented (21) .151 Gifted (22) .131 Other (23) .086 Theft Not Victim (ref) (24) Victim (25) .036 .015* .032
72 Table 4 8. Continued Independent Variables Mod el 1 Model 2 Model 3 Model 4 (1) (2) (3) (4) Robbery with Force Not Victim (ref) (26) Victim (27) .071 .059 .046 Attempted Robbery Not Victim (ref) (28) Victim (29) .159 .102 .094 Crime in the City No Fear (ref) (30) Fear (31) .244 .230 .168 .127 R 2 (32) .013 .015 .218 .299 Source: PNAD Survey 2009 *Not statistically significant. Note: Sample limited to Heads of Household in urban areas Note: Coefficients statistically significant at less t han .05 unless otherwise noted.
73 CHAPTER 5 DISCUSSION AND CONCLUSION This study supports the findings of Lab (1990) and Giblin (2008) by empirically showing that crime preventive behavior is multi dimensional. This study tests the dimensions of fear of cr ime, victimization, and socio economic variables on crime preventive behavior in the household, finding that socio economic factors are the strongest predictor of this type of behavior, in support of past findings by Skogan and Maxfield (1981) and Lab (199 0). Moreover, while fear of crime and victimization are positively correlated with this behavior, they are of minimal yet significant impact, paralleling the findings of Lab (1990). Due to the nature of the survey questions, it is difficult to categorize t he different security measures listed. The security measure questions encompass various measures that sometimes overlap dimensions of access control, surveillance, and target hardening. Many of the measures in the security device index are target hardening techniques, and all three major correlates of fortification were found to be research that found victimization to be the only statistically significant predictor. This ma y highlight the difficulty of research in this area due to the fact that all surveys are different, using different groupings of security measures as well as oftentimes different operationalizations of fear of crime. Many Brazilians do not use any security devices for their home, representing 38.9% of the population. This could be due to the fact that some people feel safe, and others may not have the resources or means to access these devices. Conversely, some people may have some of these measures, but th e survey questions may be
74 asked in a way that is confusing. If I live in an apartment, I do not have a fence by nature of the fact that an apartment is a unit within a building. But, the apartment building may have its own fence or wall, or even a video se curity camera. And, one person may say in this case that they do have a fence or wall or camera, and another might have the perspective that they do not. It is difficult to ascertain consistency of responses as survey questions may be interpreted different ly by different people. So, environment as well as of the survey. Nevertheless, it is essential to mention that tall walls/fences and security alarms are the third most preferred group of security devices within the index. 21.4% of Brazilians have either a tall wall/fence over 2 meters tall and/or an electronic security alarm. The liter ature on the matter, in combination with my personal experiences and observations in cities in Brazil, led me to believe walls and fences were more ubiquitous than this. The Center West has the highest rates of tall walls/fences and security alarms, and th is could be due to the impact of the federal capital, Braslia, where important government officials may feel the need to have more privacy and security at their residence. The most ubiquitous security devices are in fact grills/grates on windows and peeph oles and security chains on doors. This may not actually represent a preference, but the fact that these devices are cheaper and more accessible, and possibly have become a normative part of building and home housing design. Private security guards, gated entries, and video cameras, all presumed to be part of what might constitute a gated condominium, each represent less than 10% of the
75 attention in the literature to forti fication and the spread of gated condominiums. Problematic is the fact that there is no previous number or rate to which to compare. Given that these last categories represent more expensive modes of self protection, it is not surprising that these are hig hly represented in the South and Southeast regions where wealth is concentrated. The findings of this study within the dimension of socio economic variables somewhat contrast with the findings of the study done by San Juan et al (2012) in Spain. My study f inds increases in age, education, and income to be strongly and positively correlated with levels of home security. These were the strongest socio economic predictors. In contrast, San Juan et al (2012) found victimization and age and sex to be the only va riables positively correlated with crime preventive behavior, while income and education had no relationship. Lab (1990) found age to be the strongest socio economic factor, along with education and tenancy (non homeowners were more prone to self protectiv e measures in the home). This current study distinctively finds housing variables to be important and significant factors (including tenancy), while type of housing is the strongest factor. This is followed by socioeconomic indicators, with the strongest p redictors being income and education, which may give credence to the views of Caldeira (2000), Vargas (2006), and Rial and Grossi (2002) that residential fortification is a means by which the elite socially differentiate themselves. This is further suppor ted by the findings of the security devices related to race. Blacks are more disadvantaged than Browns, who are more disadvantaged than Whites in terms of household crime prevention measures. The fact that wealthier, whiter, and more
76 educated heads of hous ehold choose to have more home security may reflect on the theme of residential segregation. Residential segregation could go beyond spatial divisiveness to differences in housing quality and typology, and even in access to safety and security at home. Bec ause of the social inequality in Brazil and the history of socio spatial segregation, it is not surprising here that income and education, highly correlated, both have a positive relationship with security measures. The fact that type of residence is the b est explanatory variable in this model has a possible explanation. The highest populations are in the Southeast where the largest cities are, and with increasing urban densities of more people per hectare, the problem of scarcity of space begins. High rise apartments and condos may have become a necessity in this region. The fact that apartments are more likely to have security devices than homes, coupled with more people living in apartments than homes in the South and Southeast, may have an impact on the fact that the highest security device index measures are in the South and Southeast. I give attention to this issue since victimization rates are highest in the North where people are more vulnerable to crime 1 Using the data presented by Wood and Ribeiro (2013), the vulnerability theory of predicting crime preventive measures would indicate that the highest levels of home fortification would be associated with males in the North who are wealthier and more 1 It may ap pear that certain results, upon comparison, seem contradictory (for example, Tables 4 5 and 4 analyzed and interpretations are then made (generalized) as though they automatically d to ( Lavralcas, in mind when interpreting results of social science research and statistic al analysis.
77 educated. The findings in this study diverge from t his slightly, because the wealthier and more highly educated populations reside in the South and Southeast and those are the regions with the highest security device index means. Victimization was found to play a small role in predicting fortification, the refore this finding is not surprising. Still, consistent with vulnerability theory is that males are more likely to be victims of crime and also more likely to have increased security than females. This could possibly be avoidance strategies over protective measures (San Juan et al, 2012), and also by the gender ratio in that there are more male headed findings found minimal differences in victi mization by race, the category of Brown was still found to be more vulnerable. So, it would have been expected that there would be either no difference in security measures across race, or that Browns would be more likely to secure their homes more than Bl acks and Whites. But, in fact the Yellow and White categories are more likely to have increasing home security. While crime and fear of crime are still spreading in Brazil, it is important to note that fear of crime is a stronger predictor of crime prevent ive measures of the household in this model than actual victimization, reinforcing the concept that more people experience fear of crime than are actual victims of crime (Skogan & Maxfield, 1981). This may explain the minimal increase in variance explained when the victimization variables are added to the model. This, coupled with the findings that a majority of Brazilians have not been the victim of a recent property crime yet nearly half of them experience a fear of crime in their city, also supports the view that a culture of fear prevails in Brazil (Caldeira & Holston 1999; Caldeira, 2000).
78 Although 43.9% of Brazilians do not experience any fear anywhere, the majority (52%) experience fear of crime in their city, feeling more unsafe here than in their ho mes or neighborhoods. The lower number of people experiencing fear of crime at security de vices, they might be given a sense of safety within their home (this is logically the desired effects of fortification and home security). Likewise, neighborhood safety could be influenced by social cohesion and levels of community involvement, or if the p erson lives in an expansive gated community and their neighborhood is bounded and controlled, thereby giving a false sense of security (Coy & Pohler, 2002). Looking at effectiveness of security devices, and other indicators like social cohesion and environ mental disorder and their impacts on fortification, suggests areas for future research (Bannister & Fyfe, 2001; Scarborough et al 2010). It should be noted that this research is limited by the fact that it is a cross sectional study based on data from one point in time. Therefore, one of the main drawbacks to this study is the inability to include in the analysis the influence of time order in crime prevention responses. The results of this study are likely susceptible to the implications of the victimizat ion fear crime prevention relationship, which could alter the influence of both fear of crime and victimization on crime prevention measures of the home (Lab 1990). The nature of one survey done at one point in time without supplementary questions does no t allow for the researcher to know the sequence of events. Since only the previous twelve months were included for victimization, it is possible that there were previous victimizations and fear beyond this time frame that
79 instigated the installation of sec urity devices. Resultantly, these devices might have lessened future victimizations and fear. Another aspect of this time order conundrum is the relationship between the household head and the actual installation of security devices. It is entirely plausib le that a respondent bought or rented a home which already had these devices installed, meaning th ose devices are potentially independent of and behaviors But, did the existence of these devices influence their choice to b uy or rent that specific home stemming from their already existing insecurities and fears? This relationship is tenuous and difficult to explore with the available data and the limited questions Future research would do better to incorporate more detailed surveys inquiring about the direct relationship between household head and the installation of certain security devices, or the motivations for moving into an already secured location. Furthermore, I recommend in the future that another special survey on victimization, justice, and home security mechanisms be conducted by the IBGE for the purposes of allowing the chance to do an analysis of change over time in usage of security mechanisms and perceptions of safety. This survey may be the same in structure for the purposes of consistency and easy comparison, or could be augmented for the purposes of elaborating a better, more nuanced version of the survey. Disaggregating the list of home security measures while also expanding on them to include more mechanis ms across more CPTED categories would make future data and studies more comprehensive.
80 This study specifically focuses on the ability of fear of crime in the city, in addition to victimization and socio demographic characteristics, to predict crime prevent ion behaviors of the home. Some may argue that the order of the relationship is the reverse, that behavior may precede this fear due to the fear fe edback loop (Liska et al 1988 ) or the causality dilemma of fortification and fear of crime However, this is more likely the relationship for avoidance techniques, yet entirely plausible, however beyond the scope of the available dataset and this study and individual perception or se nse of safety/security in their city. As mentioned previously, the operationalization of this concept of fear has been varied throughout the literature. One aspect of fear for consideration that may have biased the results is the difference between persona l fear and altruistic fear as discussed by Warr and Ellison (2000). Although personal fear, that which is the basis of the survey questions used as a basis of analysis in this study, has proven to be a determining factor in taking home security measures an d making behavioral changes in daily life, altruistic fear, or a fear or concern for the safety of others, is a frequent and often the only determinant for these responses (Warr & Ellison, 2000). This highlights a limitation of this study but also a consid eration to be included in further research. Future surveys about home security and victimization in Brazil would possibly be improved by asking more and further varied questions about fear, perceptions of risk and safety, worries and anxieties, and also by including more control variables like number of elderly or children living in the household and marital status.
81 When people lack a sense of safety, there is an inherent threat to quality of life, democracy, openness, and freedom in society. People inheren tly react to feelings of insecurity and fear by protecting themselves in one way or the other, one way being self protective and crime preventive measures in the household. Besides fear of crime choice to secure the home and the level to which they will do so. Crime preventive behavior is clearly multi dimensional, and future research could explore more options that may influence this behavior to improve the model here. This study has provided add itional insights into the state of home security and fear of crime in the Latin American context, specifically in urban areas of Brazil. This research has implications for the fields of criminology, urban planning, and Latin American Studies. It contribute s to the body of literature on the relationship between the built environment, namely residential fortification, and fear of crime in Latin America. It supports the theory that fear of crime is an important and significant motivator behind residential fort ification, while highlighting the greater importance of socioeconomic variables in this relationship. If social inequalities continue to grow, there will continue to be differences in access to and use of home security devices by different groups of Brazil ians. This study also stresses the need for further investigation on this relationship and exploring other stimuli of crime prevention measures in the home realm. Improved survey and measurement techniques also need to be given attention. Apart from this, future research in the same vein as Vilalta (2012) is needed on measuring the effectiveness of these security measures and CPTED related techniques in actually
82 reducing fear of crime in Brazil and Latin America. Is this shift in the design of the built env ironment actually effective in reducing fear? Fear has negative consequences for quality of life and development, of both people, cities, and nations, and it is important to understand the relationship between these elements to make important improvements.
83 LIST OF REFERENCES Atkinson, R., & Flint, J. (2004). Fortress UK? Gated communities, the spatial revolt of the elites and time space trajectories of s egregation. Housing Studies, 19 (6), 875 892. Bannister, J., & Fyfe, N. (2001). Introduction: F ear and the city. Urban Studies, 38 (5 6) 807 813. Blakely, E. J., & Snyder, M. G. (1997). Fortress america: Gated communities in the United States. Washington, DC: Brookings Institution Press. Blandy, S. (2007). Gated communities in England as a response to crim e and disorder: context, effectiveness and implications. People, Place & Policy Online, 1 (2), 47 54. Borsdorf, A., Hidalgo, R., & Sanchez, R. (2007). A new model of urban development in Latin America: The gated communities and fenced cities in the metropol itan areas of Santiago de Chile and Valparaiso. Cities, 24 (5), 365. Brantingham, & Brantingham. (1981). Environmental Criminology. Beverly Hills: Sage Publications. Caldeira, T. (1996). Building Up Walls: The New Pattern of Spatial Segregation in Sao Paulo Internation Social Science Journal (48), 55 66. Caldeira, T. (2000). City of Walls: Crime, Segregation, and Citizenship in Sao Paulo. Berkeley: University of California Press. Caldeira, T. P., & Holston, J. (1999). Democracy and violence in Brazil. Co mpar ative Studies in Society and History, 41 (4), 691 729. Central Intelligence Agency. (2012). Brazil. Retrieved from The World Factbook: https://www.cia.gov/library/public ations/the world factbook/geos/br.html Clarke, R. V. (1995). Situational crime prevention. Crime and Justice, 19 91 150. Coy, M., & Pohler, M. (2002). Gated Communities in Latin American megacities: case studies in Brazil and Argentina. Environment and P lanning B: Planning and Design, 29 (3), 355 370. Davis, M. (1990). Fortress L.A. In M. Davis, City of quartz: Excavating the future in Los Angeles. New York: Verso. Ferraro, K. F. (1995). Crime: Interpreting victimization risk. Albany: SUNY Press.
84 Foster, S ., & Giles Corti, B. (2008). The built environment, neighborhood crime and constrained physical activitiy: An exploration of inconsisten t findings. Preventive Medicine, 47 241 251. Gerber, M. M., Hirtenlehner, H., & Jackson, J. (2010). Insecurities about crime in Germany, Austria and Switzerland: A review of research findings. European Journal of Criminology, 7 (2), 141 157. Giblin, M. J. (2008). Examining Personal Security and Avoidance Measures in a 12 City Sample. Journal of Research in Crime and Delinqu ency, 45 (4), 359 379. Giglia, A. (2008). Gated communities in Mexico City. Home Cultures, 5 (1), 65 84. Instituto Brasileiro de Geografia e Estatistica (2010). PNAD 2009. Retrieved from http://www1.ibge.gov.br/home/estatistica/populacao/trabalhoerendimento/pnad20 09/questionarios_pnad_2009.pdf Jackson, J., & Gray, E. (2010). Functional fear and public insecurities about crime. The Britis h Journal of Criminology, 50 (1), 1 22. Koskela, H., & Pain, R. (2000). Revisiting fear and place: Women's fear of attack and the built environment. Geoforum, 31 269 280. Lab, S. (1990, September). Citizen Crime Prevention: Domains and Participation. Justi ce Quarterly, 7 (3), 467 491. Lab, S. P. (2000). Crime prevention: approaches, practices and evaluations, 4th edition. Cincinnati: Anderson Publishing Co. Lavrakas, P. J., & Lewis, D. A. (1980). The conceptualization and measurement of citizens' crime preve ntion behaviors. Journal of Research in Crime and Delinquency, 17 254 272. Lavralcas, P. (2008). Ecological fallacy. In P. Lavrakas (Ed.), Encyclopedia of survey research methods (pp. 219 220). Thousand Oaks, CA: SAGE Publications, Inc. Lee, M. (2001). T he genesis of 'fear of crime'. Theoretical Criminology 467 482. Lemanski, C. (2006). Residential responses to fear (of crime plus) in two Cape Town suburbs: Implications for the post apartheid city. Journal of International Development, 18 787 802. Liska A. E., Sanchirico, A., & Reed, M. D. (1988). Fear of crime and constrained behavior specifying and estimating a reciprocal effects model. Social Forces, 66 (3), 827 837. Low, S. M. (2008). Fortification of residential neighbourhoods and the new emotions o f home. Housing Theory and Society, 25 (1), 47 65.
85 Newman, O. (1972). Defensible Space. New York: The Macmillan Company. Peterson, R. A. (1994, September). A Meta Analysis of Cronbach's Coefficient Alpha. Journal of Consumer Research, 21 (2), 381 391. Retrie ved from http://www.jstor.org/stable/2489828 Prillaman, W. C. (2003). Crime, Democracy, and Development in Latin America. Washington, DC: Center for Strategic and International Studies. Programme, United Nations Human Settlements (UN Habitat). (2007). Enhancing urban safety and security: Global report on human settlements 2007. Sterling, VA: Earthscan. Rial, C. S., & Grossi, M. P. (2002). Urban fear in Brazil: From the favelas to the Truman Show. In A. Erd entug, & F. Colombijn, Urban ethnic encounters: The spatial consequences (pp. 109 125). New York: Routledge. Safe. (n.d.). In Merriam Webster Dictionary online. Retrieved from http://www.merriam webster.com/dictionary/safe San Juan, C., Vozmediano, L., & Vergara, A. (2012). Self protective behaviours against crime in urban settings: An empirical approach to vulnerability and victimization models. European Journal of Criminology, 9 (6), 652 667. S carborough, B. K., Like Haislip, T. Z., Novak, K. J., Lucas, W. L., & Alarid, L. F. (2010). Assessing the relationship between individual characteristics, neighborhood context, and fear of crime. Journal of Criminal Justice 819 826. Schneider, R. H., & Ki tchen, T. (2002). Planning for Crime Prevention. New York: Routledge. Secure. (n.d.). In Merriam Webster Dictionary online. Retrieved from http://www.merriam webster.com/dictionary/secure Sil va, M. F. (2007). Gated communities: The new ideal way of life in Natal, Brazil. Housing Policy Debate, 18 (3), 557 576. Skogan, W. G., & Maxfield, M. G. (1981). Coping with crime: Individual and neighborhood reactions. Beverly Hills: Sage Publications. The Latin American Public Opinion Project (LAPOP). (2010). The AmericasBarometer. Retrieved from www.LapopSurveys.org Vargas, J. H. (2006). When a favela dared to become a gated condominium: The politics of race and urban space in Rio de Janeiro. Latin American Perspectives, 33 49 81.
86 Vetter, D. M., Beltrao, K. I., & Massena, R. M. (2013). The impact of the sense of security from crime on residential property values in Brazilian metropolitan areas. Washington, DC: I nter American Development Bank. Vilalta, C. J. (2011a). Fear of crime in gated communities and apartment buildings: a comparison of housing types and a test of theories. Journal of Housing and the Built Environment, 26 107 121. Vilalta, C. J. (2011b). Fea r of crime in public transport: Research in Mexico City. Crime Prevention & Community Safety, 13 (3), 171 186. Vilalta, C. J. (2012). Fear of crime and home security systems. Police Practice and Research, 13 (1), 4 14. Warr, M., & Ellison, C. G. (2000, Novem ber). Rethinking Social R e actions to Crime: Personal and Altruistic Fear in Family Households. American Journal of Sociology, 106 (3), 551 578. Retrieved from http://www.jstor.org/stable/10.1086/318964 Wilkinson, C. (1998). Deconstructing the fort -the ro le of postmodernity in urban development. Journal of Australian Studies, 57 194 205. Wilson, J. Q., & Kelling, G. L. (1982, March). Broken windows: The police and neighborhood safety. The Atlantic Monthly pp. 29 38. Retrieved from http://www.theatlantic.com/magazine/archive/1982/03/broken windows/304465/ Wood, C. H., & Ribeiro, L. (2013, February 14 15). Crime vicitmization in Brazil, 1988 2009: Changing vulnerabili ties by race, class, and place. Working Paper presented at the 62nd Annual Conference of the University of Florida's Center for Latin American Studies, "Emergent Brazil" Gainesville, Florida.
87 BIOGRAPHICAL SKETCH Sarah received her Bachelor of Arts with a double major in both Latin American Studies and Spanish from the University of Virginia. In 2010, she began graduate school at the University of Florida. Through fellowships, assistantships, research, teaching, and student organizations, s he divided her time between the Center for Latin American Studies, the Department of Urban and Regional Planning, and the Department of Spanish and Portuguese. Sarah spent four summers in Brazil studying, traveling, doing research, and being a TA for a study abroad progr am. She majored in Latin American Studies and graduated with a Master of Arts degree in the spring of 2014, concurrently with a Master of Arts in Urban and Regional Planning.