You are on page 1of 15

J Hous and the Built Environ (2011) 26:107–121

DOI 10.1007/s10901-011-9211-3

ARTICLE

Fear of crime in gated communities and apartment


buildings: a comparison of housing types and a test
of theories

Carlos J. Vilalta

Received: 12 December 2009 / Accepted: 8 February 2011 / Published online: 23 February 2011
Ó Springer Science+Business Media B.V. 2011

Abstract Dramatic increases in crime and fear of crime in Mexico have encouraged
interest in research questions about the relationship of fear with new housing develop-
ments. We have seen increases in the number of gated communities and apartment
buildings in Mexico City as a response to fear of crime. But there are two problems. First,
it is not known if these housing options actually help control fear of crime. Second, fear of
crime theories are pending further empirical testing worldwide. The end results include
general misunderstanding and weak connections between theory and reality. This study
shows that neither gated communities nor apartment buildings seem to provide lower
levels of fear of crime when home alone. This statement was verified through non-para-
metric bivariate tests, ordinal logistic, and binary logistic regression analyses. Other
variables held constant, fear of crime was unrelated to the type of housing. Instead, fear of
crime when home alone was associated with gender, years of schooling, social marginality
levels, neighborhood fear of crime levels, and the opinion on the local police.

Keywords Gated communities  Apartments  Fear of crime  Mexico

1 Introduction

Aside from economic problems, it is estimated that 61 million Mexicans are afraid of
robbery and 54 million are afraid of being kidnapped.1 Fear of crime not only leads to a
lower quality of life (Stafford et al. 2007; Hale 1988, 1996), but plays a crucial role in
shaping behaviors such as defensive homeownership (Caldeira 2001; Atkinson and Blandy
2007).

1
Source Encuesta de Percepción Ciudadana sobre la Seguridad en Mexico, by Consulta Mitofsky,
November, 2008.

C. J. Vilalta (&)
Center for Economic Research and Education (CIDE), Carretera México-Toluca 3655, Col. Lomas de
Santa Fe, 01210 Mexico, DF, Mexico
e-mail: carlos.vilalta@cide.edu

123
108 C. J. Vilalta

Gated communities are on the rise in almost every major metropolitan area in Latin
America (Mertins 2009; Vilalta 2011; Borsdorf and Hidalgo 2008; Giglia 2008;
Sheinbaum 2008; Atkinson and Blandy 2005; Coy and Pöhler 2002; Caldeira 2001). These
communities are expected to reduce victimization rates and fear of crime as well (Glasze
et al. 2006; Romig 2005; Blakely and Snyder 1998). However, there is debate as to whether
gated communities are really effective in reducing and/or keeping victimization rates down
(Sanchez et al. 2005; Diamond 1997; Blakely 1995) and/or ameliorating the fear of crime
in actual fact (Wilson-Doenges 2000).
These ideas have been subjected to statistical testing in many North American cities, but
not in the Mexico City metropolitan area (MCMA). We were able to find only a handful of
statistical studies of fear of crime in Mexico City (Vilalta 2010, 2011), but all unrelated to
the gated communities debate. Thus, this study’s original purpose was to empirically test
the proposition that gated communities and apartment buildings are effective housing
designs for controlling fear of crime in this megacity. Effective here is understood as fit to
accomplish its purpose, namely, that levels of fear of crime in gated communities and
apartments are consistently (i.e. over time) lower than those in opposite housing options
(i.e. non-gated communities and houses). This proposition was tested using the 2005 and
2007 Mexico City metropolitan area surveys on victimization and institutional efficacy
(ENVEI) conducted by the Center for Economic Research and Education (CIDE)
researchers. A set of bivariate non-parametric tests and ordinal multivariate regressions
were used to test this proposition.
The results show that neither gated communities nor apartment buildings provided their
residents with lower levels of fear of crime when home alone than their counterparts.
Findings confirm the Wilson-Doenges (2000) argument. In fact, fear of crime was asso-
ciated with gender, years of schooling, community marginality levels, neighborhood fear
of crime levels, and opinion on the local police. As such, this study provides evidence to
support physical vulnerability, social vulnerability, social disorder, and social network
theories of fear of crime. Naturally, the findings of the present study are limited to the
Mexico City metropolitan area and to 2 years of measurement only. However, the absence
of evidence in favor of gated communities and apartment buildings as effective housing
options to control fear of crime is sufficient to posit some critique on what they are
expected to offer and do not deliver.
The paper is laid out as follows. After this introduction, the following section presents
the theories and correlates of fear of crime. Section 3 presents the theoretical framework
and hypotheses. Section 4 describes the data, variables and statistical analyses conducted
in this study. Section 5 presents the results of the statistical analyses. Section 6 is devoted
to the discussion of the results.

2 Theories and correlates of fear of crime

Fear of crime is a well-developed area of study. Previous studies have linked fear of crime
with a variety of correlates at two different levels of analysis: The community and the
individual (Table 1). These two levels of analysis have allowed the deduction of five
different theories of fear of crime (Vilalta 2010; Bissler 2003): victimization, physical
vulnerability, social vulnerability, social disorder,2 and social networks.

2
Sometimes defined as incivility theory (see Bissler 2003) o social disorganization theory (see Haining
et al. 2007; Shaw and McKay 1942).

123
Fear of crime in gated communities and apartment buildings 109

Table 1 Summary of correlates


Author and year Causes/circumstances Effecta

Neighborhood level
Moore and Shepherd (2007) Neighborhood physical conditions Fear of crime (?)
Williamson et al. (2006) (e.g. graffiti walls, poor housing
Doran and Lees (2005) conditions etc.)
Miceli et al. (2004) and
Taylor (1999)
Lane and Meeker (2004) Neighborhood racial diversity and Fear of crime (?)
Ferraro (1995) and criminal reputation (e.g. racial
Skogan (1995) composition)
Rodrigues (2006) Opinion on the police work Fear of crime (-)
Bissler (2003) and
Lewis and Salem (1986)
Chiricos et al. (2000) Local news Fear of crime (?)
O’Connell (1999) and
O’Connell and Whelan (1996)
Individual level
Rader et al. (2007) Gender Fear of crime (?)
Scott (2003)
Warr (2000)
Gilchrist et al. (1998)
Hale (1996) and
Killias (1990)
Chadee and Ditton (2003) Age Fear of crime (-)
Ditton et al. (1999) and
Ferraro and LaGrange (1992)
Covington and Taylor (1991) Level of education Fear of crime (-)
McGarrell et al. (1997) and Income Fear of crime (-)
Will and McGrath (1995)
Rader et al. (2007) and Victim of a crime Fear of crime (?)
Schafer et al. (2006)
a
Read as a correlation

Victimization theory predicts that victims will suffer from higher levels of fear of crime
than those without that experience (Skogan 1987; Hale 1996). This theory makes a clear
distinction between direct and indirect victimization. Direct victimization occurs when the
victim personally suffers the crime, whereas indirect victimization is suffered by those who
know the victim or have heard about the crime (Hale 1996). A central premise of this
theory is that the victimization experience may have lasting effects upon the victims,
resulting in a change of their perceptions and behaviors (Stafford and Galle 1984; Schafer
et al. 2006; Rader et al. 2007). Evidence in favor of this theory can be found in classic
studies in the United States (Friedman et al. 1982; Liska et al. 1988) and the United
Kingdom (Maguire and Corbett 1987). Although common-sensical, this theory is not
without debate since its beginnings and is pending further testing (Box et al. 1988). Some
previous studies found no relationship between victimization and fear of crime (DuBow
et al. 1979; Skogan and Maxfield 1981; Carcach et al. 1995). Also, some have argued for a
long time it seems that victims of violent and non-violent crimes should be studied
separately (Dull and Wint 1997; Miethe and Lee 1984).

123
110 C. J. Vilalta

The physical vulnerability theory predicts that fear of crime will be higher among those
that have a physical inability to defend themselves or to recover from a criminal assault
(Bissler 2003; Pantazis 2000). This theory uses an individual-based level of analysis. The
typical empirical correlates of this theory have been age and gender. Higher levels of fear
of crime have been found among the elderly and female survey respondents (Rader et al.
2007; Scott 2003; Warr 2000; Gilchrist et al. 1998; Hale 1996; Killias 1990; Chadee and
Ditton 2003; Ditton et al. 1999; Ferraro and LaGrange 1992).
While the theory of victimization, particularly direct victimization, is based on an
objective element (i.e. the experience of crime) physical vulnerability theory is based on
subjective elements (Bissler 2003). This theory is also paradoxical at times. One paradox is
that women report higher levels of fear of crime, when they actually suffer from lower
levels of victimization in comparison to men (Hale 1996). Another paradox is that the
elderly population also tends to report higher levels of fear of crime when, again, they
actually suffer lower levels of victimization than other demographic groups, for instance
the young people under 30 years old.
Naturally, vulnerability may have a social origin as well. The theory of social vulner-
ability predicts that individuals belonging to a disadvantaged group, for instance the poor,
will be more likely to suffer from higher levels of fear of crime. The causal mechanism is
that the poor cannot prevent crime and/or recover from its damaging effects (Bissler 2003;
Rohe and Burby 1988). Typical correlates of this theory are educational levels, income,
occupation, and unemployment (McGarrell et al. 1997; Will and McGrath 1995; Covington
and Taylor 1991).
Social disorder theory predicts that neighborhoods with poor physical conditions (e.g.
graffiti-covered walls, vacant run-down housing, dirty sidewalks etc.) and some elements
of social composition, neighborhood function and reputation (e.g. prostitution, drug-
dealing, drug addicts) will have higher levels of fear of crime (Hunter 1978). The causal
mechanism is that community signs of neglect, along with a lack of social control and
social cohesion, facilitate both actual victimization and higher levels of fear of crime
(Moore and Shepherd 2007; Williamson et al. 2006; Doran and Lees 2005; Miceli et al.
2004; Taylor 1999). This theory typically makes use of correlates such as the local pop-
ulation age structure, crime activity, proportion of vacant houses, poverty levels, and
indicators of family structure (e.g. single parenting, children with inadequate supervision).
Social network theory focuses on information flows between individuals. This theory
predicts that communities with high levels of networking will also have higher levels of
social cohesion, therefore being able to respond collectively and effectively against crime
and the fear of crime. On the other hand, communities with lower levels of social net-
working may experience higher levels of criminal activity and fear of crime (Bursik and
Grasmick 1993). On the contrary, it may be that communities with strong social networks
may suffer from higher levels of fear of crime since information flows occur with greater
speed and amplitude (Bursik and Grasmick 1993). Thus, social networks can theoretically
reduce fear of crime through the enforcement of informal social controls or increase fear
through indirect victimization (Bissler 2003; Hale 1996). A common correlate depicting
the strength of a social network is the community’s relationships with local law enforce-
ment agencies (Vilalta 2010; Rodrigues 2006; Bissler 2003; Lewis and Salem 1986). If
mistrust of police is widespread, fear is likely to be higher.
The community level, while theoretically important, has some limitations. One diffi-
culty with the empirical testing of neighborhood correlates is that most of the crucial
concepts mentioned in the social disorganization theory are difficult to model. Studies that
use neighborhood measures to study fear of crime may be overly general and miss specific

123
Fear of crime in gated communities and apartment buildings 111

causal mechanisms. For example, correlates may not properly capture fear of crime in one
place because residents may be afraid/prejudiced of/against other things besides crime (e.g.
racial diversity, anger against the police).
In sum, there are neither univariate nor unidimensional studies of fear of crime. In an
effort to capture a more complete picture of the fear of crime, the majority of authors use
different correlates operating at both the community and the individual level. As with other
social problems, fear of crime cannot accurately be understood or tackled by examining
only one level of analysis. In addition, it is clear that theories of fear of crime assume it is
not a random experience. This is why it is reasonable to combine community and indi-
vidual levels of analysis and to test different theories of fear of crime concurrently. For this
purpose, a theoretical framework is presented in the following section.

3 Theoretical framework and hypotheses

Previous studies have greatly helped illuminate the nature of fear of crime. One problem
arises, however, when other analytical categories are overlooked. For instance, the rela-
tionship of fear of crime with gated communities and apartment buildings is not resolved
(Wilson-Doenges 2000; Diamond 1997; Blakely 1995). It is debated if gated communities
and apartment buildings actually help control or reduce fear of crime. One would think
they do considering that these housing options help create a fortress mentality. Never-
theless, from an empirical standpoint, these relations are hypothetical.
In order to test these relationships, the following theoretical framework is proposed.
Each circle depicts a theory of fear of crime and each box next to it shows the corre-
sponding correlates in this study. It must be considered that these correlates serve as

Fig. 1 Theoretical framework

123
112 C. J. Vilalta

control variables in the model. The independent variables (IV1 and IV2) are shown in the
upper left box.
Thus, the theoretical framework posits the following hypothesis: All things being equal,
residents in gated communities (IV1) and apartment buildings (IV2) consistently (i.e. in
2005 and 2007) report lower levels of fear of crime when home alone (DV). The following
section explains the data and methods utilized to test this hypothesis.

4 Data, variables and statistical analyses

4.1 Data

The analysis relied on two surveys: The summer 2005 and 2007 Mexico City Metropolitan
Area (MCMA) surveys on victimization and institutional efficacy (ENVEI). These surveys
were designed and conducted by researchers of the school of legal studies at the Center for
Economic Research and Education (CIDE) in Mexico City. Both surveys contained data on
3,474 households in the MCMA. Data were collected by means of personal interviews with
adults over 18 years old during 8- to 10-week periods (i.e. between late June and August).
Databases are free and available for download in SPSS and Text formats from the Banco
de Informacion para la Investigacion Aplicada en Ciencias Sociales (BIIACS) at CIDE.3
Thus the results in this study can be replicated.
The same questionnaire was applied both years. The questionnaire covered a variety of
aspects regarding fear of crime, victimization, and institutional efficacy. The questionnaire
contained three sub-categories of fear of crime: fear in the neighborhood, fear in public
transportation, and fear of crime when home alone. The questionnaire also included data on
gender, age, family socioeconomic characteristics and the community characteristics of the
respondents. This study focused on understanding the fear of crime when home alone
controlling for a variety of correlates including fear of crime in the neighborhood.

4.2 Variables

The dependent variable (DV) was self-reported levels of fear of crime when home alone.
The DV was measured on a scale of 1–4, with 1 = very secure, 2 = somewhat secure,
3 = little secure, and 4 = not secure. Most survey respondents would give an answer to
this question (99.9%). Non-responses were considered as missing values and not analyzed.
There were two dichotomous independent variables: Residents in gated communities
(IV1) and residents in building apartments (IV2) measured as 1 = No and 2 = Yes.
Residents of gated communities were defined as those living in residential areas with
restricted access in the way of either gated or guarded entrances. Residents in apartment
buildings were defined as those living in multifamily units regardless of the number of
stories in the building. The type of housing unit would be recorded directly by the inter-
viewer in the survey questionnaire.
Control variables were included to model possible impacts associated with the inde-
pendent variables. These control variables were a set of correlates identified in the liter-
ature chosen for their theoretical relevance to the context of fear of crime. These variables
were included in the 2005 and 2007 survey databases. Figure 1 shows the control variables
selected.

3
http://www.biiacs.cide.edu.

123
Fear of crime in gated communities and apartment buildings 113

4.3 Statistical analysis

Descriptive analyses and inferential tests were carried out using SPSS version 15.
Pearson’s Chi-square, Median and Mann–Whitney tests were conducted searching statis-
tically significant differences in the self-reported levels of fear of crime when home alone
(DV) between respondents in gated communities and apartment buildings.
The theoretical framework (see Fig. 1) was tested twice via logistic ordinal and binary
regression analyses for both gated community respondents (IV1) and apartment building
respondents (IV2). These tests were performed for the 2005 and the 2007 survey respon-
dents. Same statistical controls were applied in each equation. Only statistically significant
control variables were kept in the equations. Statistical significance was determined at the
p B 0.05 level for all tests.

5 Results

The self-reported levels of fear of crime in the neighborhood and when home alone are
presented in Table 2. Fear of crime when home alone is high considering that only around
41% of the survey respondents reported to feel ‘‘very safe’’ inside their homes both years.
Still, many more respondents reported to feel ‘‘very safe’’ at home than in their neigh-
borhoods; 41 and 24%, respectively. Pearson’s Chi-square tests detected a significant
reduction in the levels of fear of crime between 2005 and 2007, both at home and in the
neighborhood. This was mostly due to the increase in the proportion of respondents feeling
‘‘somewhat safe’’ and to the reduction of those feeling ‘‘unsafe’’.
The results for the Median and (the more robust) Mann–Whitney difference tests are
shown in Table 3. Results show that respondents in gated communities in 2005 reported
lower levels of fear of crime when home alone compared to respondents outside gated
communities. For 2007, this gap between the two groups was not statistically significant.

Table 2 Fear of crime in the neighborhood and when home alone, 2005 and 2007

2005 (n = 1,995) 2007 (n = 1,476) Change 2005–2007 Pearson’s


(%) (%) (%) Chi-square

When home alone


Very safe 41.7 41.1 -0.6 20.973 (p \ 0.000)
Somewhat safe 32.2 36.1 3.9
Little safe 17.7 18.1 0.4
Unsafe 8.4 4.7 -3.7

2005 (n = 1,997) 2007 (n = 1,477) Change 2005–2007 Pearson’s


(%) (%) (%) Chi-square

In the neighborhood
Very safe 24.8 24.4 -0.4 8.788 (p = 0.032)
Somewhat safe 38.1 42.2 4.1
Little safe 25.9 24.4 -1.5
Unsafe 11.2 8.9 -2.3

Significance values in parentheses

123
114 C. J. Vilalta

Table 3 Difference tests of fear of crime when home alone among respondents in gated communities and
apartment buildings, 2005 and 2007

2005 (No = 1,884, yes = 111) 2007 (No = 1,432, yes = 44)

Respondents in gated communities


No 1,008.6 741.8
Yes 818.4 630.5
Median’s Chi-square 9.674 (0.002) 2.149 (0.143)
Mann–Whitney’s Z -3.586 (p \ 0.000) -1.820 (0.069)

2005 (No = 1,704, yes = 269) 2007 (No = 1,307, yes = 153)

Respondents in building apartments


No 1,002.1 736.8
Yes 891.5 676.6
Median’s Chi-square 3.674 (0.055) 1.035 (0.309)
Mann–Whitney’s Z -3.142 (0.002) -1.781 (0.075)

Ranked means are presented. Higher values indicate higher levels of fear of crime. Significance values in
parentheses

Similarly, for 2005 respondents in building apartments reported lower levels of fear of
crime when home alone only. For 2007, again the gap was not statistically significant. The
differences between significance levels of Median and Mann–Whitney tests (p = 0.055
and p = 0.002, respectively) are due to the potency in each test. As such, these results
should be taken cautiously. In fact, from these preliminary results it was possible to reject
hypotheses 1 and 2. The sum of evidence from the two surveys would not allow us to argue
that gated community and apartment building residents reported significantly lower levels
of fear of crime when home alone in a temporally consistent fashion.
However, to further test these hypotheses of fear of crime, multivariate ordinal
regression analyses were performed for each of the 2005 and 2007 surveys.
Starting with the 2005 survey, the model showed a number of significant correlates of fear
of crime when home only for the ‘‘very safe’’ (p = 0.003) and ‘‘little safe’’ (p \ 0.000) rating
levels. Neither residing in a gated community (p = 0.346) nor in an apartment building
(p = 0.190) suggested a significant difference in the levels of fear of crime when home alone
holding other variables constant. Thus, again, the weight of evidence is in favor of rejecting
the research hypothesis.
Altogether, the model provided partial evidence to support physical vulnerability
(gender), social vulnerability (schooling), social disorder (marginality and fear of crime in
the neighborhood), and social network (opinion on local police) theories of fear of crime.
This model did not provide evidence to support a (direct) victimization theory (Table 4).
Goodness-of-fit measures and diagnostics are presented at the bottom of the table. The
model was highly statistically significant (p \ 0.000) and explained a good proportion of
the variation in the dependent variable (Nagelkerke = 0.277).4 The test of parallel lines
indicated that the model is adequate using the probit link function (p = 1.000).5

4
Ordinal regression computes the Nagelkerke coefficient, which is a pseudo coefficient of determination
comparable to the least squares regression coefficient of determination.
5
This means that the results are a set of parallel lines -one for each category- of the dependent
variable.

123
Fear of crime in gated communities and apartment buildings 115

Table 4 Fear of crime when home alone among non-gated/gated community and house/apartment building
residents regressed on selected correlates, 2005
n Coefficient Standard error Wald statistic Sig.

Dependent variable
Very safe when home alone 786 -0.865 0.288 8.997 0.003
Somewhat safe when home alone 615 0.173 0.288 0.360 0.549
Little safe when home alone 334 1.054 0.289 13.313 0.000
Unsafe when home alonea 156
Independent variables
Non-gated community resident 1,785 0.130 0.138 0.889 0.346
Gated community resident 106 0.000
House resident 1,633 0.118 0.090 1.717 0.190
Apartment building resident 258 0.000
Control variables
Male 671 -0.314 0.057 30.853 0.000
Female 1,220 0.000
Up to 6 years of schooling 78 0.418 0.269 2.414 0.120
7–9 years of schooling 598 0.513 0.241 4.523 0.033
9–12 years of schooling 504 0.431 0.241 3.195 0.074
12–16 years of schooling 435 0.406 0.241 2.833 0.092
More than 16 years of schooling 248 0.282 0.246 1.318 0.251
Up to 6 years of schooling 28 0.000
Very low marginality 364 -0.141 0.094 2.262 0.133
Low marginality 381 -0.367 0.086 18.248 0.000
Average marginality 379 -0.245 0.082 8.861 0.003
High marginality 383 -0.165 0.081 4.214 0.040
Very high marginality 384 0.000
Very safe in the neighborhood 465 -1.604 0.101 253.935 0.000
Somewhat safe in the neighborhood 728 -0.903 0.088 106.113 0.000
Little safe in the neighborhood 487 -0.342 0.089 14.664 0.000
Unsafe in the neighborhood 211 0.000
Very good opinion on local police 37 -0.773 0.252 9.414 0.002
Good opinion on local police 387 -0.290 0.081 12.713 0.000
Neither good nor bad opinion on local police 375 -0.176 0.079 4.880 0.027
Bad opinion on local police 628 -0.079 0.068 1.332 0.249
Very bad opinion on local police 464 0.000
Ordinal regression diagnostics:
Valid cases = 1,891/final model fit (Chi-square)** = 553.846 (p \ 0.000)
Nagelkerke pseudo R-square = 0.277/test of parallel lines (Chi-Square)* = 5.776 (p = 1.000)/Link
function: Probit
The model only considered the statistically significant control variables
a
This is the category of reference for the ordinal regression model

Proceeding with the 2007 ordinal regression model, the test of parallel lines indicated a
highly significant value in every possible link function6 in the inclusion of any control

6
These are: logit, probit, cauchit, complementary log–log, and negative log–log.

123
116 C. J. Vilalta

variable, therefore suggesting the rejection of the null hypothesis that the location
parameters were the same across categories in the dependent variable. As such, ordinal
regression is not proper for these data and may be replaced by binary logistic regression.
The ordinal dependent variable was recoded into two categories, where those respondents
with a score of 1–2 were given a value of 1 and those with 3–4 were given a value of 2. In
other words, fear of crime when home alone was recoded into two different levels: very
safe and somewhat safe (1), and little safe and unsafe when home alone (2). The findings
are presented in Table 5 below.
Results again suggest rejecting the research hypothesis that residents in gated com-
munities (p = 0.814) and/or apartment buildings (p = 0.078) report lower levels of fear of
crime when home alone. The results provided evidence to support the physical vulnera-
bility (gender) and social disorder (fear of crime in the neighborhood) theories of fear of
crime. This model did not show evidence to support (direct) victimization, social vul-
nerability or social network theories. Overall, this model was also highly significant
(p \ 0.000) and was able to explain a good proportion of the variation in the dependent
variable (Nagelkerke = 0.246).
Overall, regression results show five specific correlates of fear of crime: gender, years of
schooling, social marginality, neighborhood fear of crime, and opinion on the local police.
For both the 2005 and the 2007 surveys, gender (physical vulnerability theory) and fear of
crime in the neighborhood (social disorder theory) emerged as the most likely predictors of
fear of crime when home alone. The impact of household head years of schooling was not
as straightforward as other predictors, as most of its categories were not statistically
significant. The two exceptions were the ‘‘up to 6 years of schooling’’ and the ‘‘7–9 years
of schooling’’ categories, which are representative of the majority of the adult population
in the city. However, because of the way these categories were constructed, that is asy-
metrically, it is most likely that subtle variations would be found if schooling were
measured as a continuous variable. With respect to the opinion on the local police, this
study is not original but rather repetitive of what has been already said about the Mexico

Table 5 Fear of crime when home alone among non-gated/gated community and house/apartment building
residents regressed on selected correlates, 2007
n Coefficient Standard error Wald statistic Sig.

Constant – 0.061 0.525 0.013 0.908


Independent variables
Non-gated community resident 1,415 0.113 0.479 0.055 0.814
House resident 1,306 0.417 0.237 3.104 0.078
Control variables
Male 584 -0.913 0.154 35.154 0.000
Very safe in the neighborhood 358 169.760 0.000
Somewhat safe in the neighborhood 614 -2.872 0.283 102.988 0.000
Little safe in the neighborhood 356 -1.960 0.214 83.635 0.000
Unsafe in the neighborhood 131 -0.659 0.212 9.684 0.002
Binary regression diagnostics:
Valid cases = 1,459/final model fit (Chi-square)** = 258.358 (p \ 0.000)
Nagelkerke pseudo R-square = 0.246
The model only considered the statistically significant control variables

123
Fear of crime in gated communities and apartment buildings 117

City police (Vilalta 2010): that a good public perception of the police is not sufficient, but
still it is necessary for combating fear of crime.
The marginality index is a measure of relative deprivation. The National Population
Council (CONAPO) provides this marginality index at the state and municipality level
every 5 years. This index reflects social marginality based on four social components and
nine indicators. The components are education, housing, income, and population distri-
bution. The education component is made up by two indicators: percentage of people under
15 years old unable to read and write, and percentage of people under 15 years old without
elementary education.7 The housing component is made up by five different indicators:
percentage of people living in houses with no running water, no drainage, no electricity, no
solid floor materials and with some level of overcrowding. The income component is made
up by one indicator only: percentage of the occupied labor force earning less than two
minimum daily wages (i.e. approximately 9 US dollars per day). The population distri-
bution component is made up by one indicator as well: percentage of people living in
settlements of less than 5,000 inhabitants. The information sources are the Mexican
decennial censuses and the 5-year population and housing counts.8 CONAPO conducts a
principal factor component statistical analysis to arrive at these indexes. The index is a
normalized Z score varying between 3 and -3 standard deviations from the national mean
(M = 0), which correspond to very low and very high marginality levels in that order.9

6 Discussion

While there is a good deal of knowledge about fear of crime in some developed countries,
the same cannot be said for the Latin American region. The review of the literature reveals
that all theories of fear of crime are subject to dispute and further empirical testing is
pending. In addition, it is still not clear if gated communities and apartment buildings are
effective housing solutions in reducing and/or keeping down levels of fear of crime
(Wilson-Doenges 2000) in spite of their promise of a safer environment.
To help fill these vacuums, this study examined the impact of gated communities and
apartment buildings on fear of crime when home alone in the Mexico City metropolitan
area (MCMA) in 2005 and 2007. As it turned out, neither living in a gated community nor
in an apartment building made a difference in the self-reported levels of fear of crime when
home alone among MCMA survey respondents. After controlling for different theoretical
correlates of fear of crime, multivariate analyses showed no significant differences between
residents in gated communities or non-gated communities, and residents in apartment
buildings or houses. Only for 2005, bivariate tests suggested that gated community and
apartment building respondents reported lower levels of fear of crime than their counter-
parts. However, once statistical controls were introduced, differences were not statistically
significant, suggestive of the ineffectiveness of these housing options to control fear of
crime when home alone. An interesting finding was that fear of crime when home alone
and in the neighborhood actually declined between 2005 and 2007. However, at this
moment it is unknown if this reduction was in some way associated with the type of

7
That is 6 years of formal education.
8
In this country population censuses are conducted every 10 years whereas population and housing counts
are conducted every five.
9
The study utilized the 2005 marginality indexes. For more information and databases on marginality
indexes visit: http://www.conapo.gob.mx.

123
118 C. J. Vilalta

housing. This short-term trend highlights a key vacuum in the literature on the Latin
American region, namely that we do not have a good picture of how different dimensions
of fear covary, if they actually do.
The main theoretical implication of this study is that it cannot provide evidence to
support all existing theories of fear of crime, particularly, victimization theory (see Fig. 2).
This study did not find evidence to support victimization theory in the sense of expecting
higher levels of fear of crime when home alone among residents that had been victims of a
burglary or attempted burglary. Instead, evidence suggests that fear of crime when home
alone can be explained by a combination of correlates representative of physical vulner-
ability, social vulnerability, social disorder, and social network theories of fear of crime.
Likewise, results showed that a lower level of deprivation in the municipality of residence
may help reduce fear of crime when home alone. Thus, it may be that signs of social
disorder leading to fear of crime can be reduced via public investment, leading on the
whole to a better quality of life. This calls for urban planning and social development
policies.
One important limitation to the present study is the statistical modeling of fear of crime.
Although observable, fear of crime is neither a binary nor an ordinal emotion or state of
mind—no human emotion is. Thus, modeling fear of crime as a discrete or semi-discon-
tinuous state may help elucidate the correlates but may also oversimplify the research
hypotheses and conclusions. However, shifting from a discrete to a continuous model
requires a supporting methodology. Also, we do not know so far if there is a ‘‘gold
standard’’ by which to measure fear of crime.
As we move forward into the twenty-first century, it is very likely that gated commu-
nities and apartment buildings will remain popular housing options in Mexico City. The

Fig. 2 Fear of crime when home alone for gated communities and apartment building residents regressed
on selected correlates according to theories of fear, 2005. Read as correlation; ns not significant

123
Fear of crime in gated communities and apartment buildings 119

picture will hardly be any different. As land prices, property taxes, and the real estate
exploitation of fear of crime in the city increase, family houses characteristic of the last
century–having large (or even small) gardens with access from open streets–will pro-
gressively become a rather rare and fairly expensive housing option for urban dwellers.

References

Atkinson, R., & Blandy, S. (2005). Introduction: International perspectives on the new Enclavism and the
rise of gated communities. Housing Studies, 20(2), 177–186.
Atkinson, R., & Blandy, S. (2007). Panic rooms: The rise of defensive homeownership. Housing Studies,
22(4), 443–458.
Bissler, D. (2003). Fear of crime and social networks: A community study of two local public housing
complexes. Ph.D Dissertation. North Carolina State University.
Blakely, E. (1995). Fortress communities: The walling and gating of American suburbs. Land Lines, 7, 1–3.
Blakely, E., & Snyder, M. (1998). Forting up: Gated communities in the United States. Journal of Archi-
tectural and Planning Research, 15(1), 61–72.
Borsdorf, A., & Hidalgo, R. (2008). New dimensions of social exclusion in Latin America: From gated
communities to gated cities, the case of Santiago de Chile. Land Use Policy, 25(2), 153–160.
Box, S., Hale, C., & Andrews, G. (1988). Explaining fear of crime. British Journal of Criminology, 28(3),
340–356.
Bursik, R., & Grasmick, H. (1993). The use of multiple indicators to estimate crime trends in American
cities. Journal of Criminal Justice, 21(5), 509–516.
Caldeira, T. (2001). City of walls. Berkeley: University of California Press.
Carcach, C., Frampton, P., Thomas, K., & Cranich, M. (1995). Explaining fear of crime in Queensland.
Journal of Quantitative Criminology, 11, 271–287.
Chadee, D., & Ditton, J. (2003). Are older people most afraid of crime? British Journal of Criminology,
43(2), 417–433.
Chiricos, T., Padgett, K., & Gertz, M. (2000). Fear, TV news, and the reality of crime. Criminology, 38(2),
755–785.
Covington, J., & Taylor, R. (1991). Fear of crime in urban residential neighborhoods: Implication of
between- and within- neighborhood sources for current models. Sociological Quarterly, 32(2),
231–249.
Coy, M., & Pöhler, M. (2002). Gated communities in latin American megacities: Case studies in Brazil and
Argentina. Environment and Planning B: Planning and Design, 29(3), 355–370.
Diamond, D. (1997). Behind closed gates. USA Today, 1, 1–3.
Ditton, J., Bannister, J., Gilchrist, E., & Farrall, S. (1999). Afraid or angry? Recalibrating the ‘fear’ of crime.
International Review of Victimology, 6(2), 83–99.
Doran, B., & Lees, B. (2005). Investigating the spatio-temporal links between disorder, crime, and the fear
of crime. The Professional Geographer, 57(1), 1–12.
DuBow, F., McCabe, E., & Kaplan, G. (1979). Reactions to crime: A critical review of the literature.
Washington, DC: U.S. Department of Justice.
Dull, R., & Wint, A. (1997). Criminal victimization and its effect on fear of crime and justice attitudes.
Journal of Interpersonal Violence, 12(5), 748–758.
Ferraro, K. (1995). Fear of crime: Interpreting victimization risk. Albany: State University of New York
Press.
Ferraro, K., & LaGrange, R. (1992). Are older people most afraid of crime? Reconsidering age differences
in fear of victimization. Journal of Gerontology, 47(5), 233–244.
Friedman, K., Bischoff, H., Davis, R., & Person, A. (1982). Victims and Helpers: Reactions to crime.
Washington, DC: National Institute of Justice.
Giglia, A. (2008). Gated communities in Mexico city. Home Cultures, 5(1), 65–84.
Gilchrist, E., Bannister, J., Ditton, J., & Farral, S. (1998). Women and the fear of crime: Challenging the
accepted stereotype. British Journal of Criminology, 38(2), 283–298.
Glasze, G., Webster, C., & Frantz, K. (2006). Introduction. In G. Glasze, C. Webster, & K. Frantz (Eds.),
Private cities: Global and local perspectives (pp. 1–8). London: Routledge.
Haining, R., Ceccato, V., & Kahn, T. (2007). The geography of homicide in Sao Paolo, Brazil. Environment
and Planning A, 39(7), 1632–1653.

123
120 C. J. Vilalta

Hale, C. (1988). Fear of crime and quality of life: a test of Garofalo and Laub’s model. Criminal Justice
Review, 13(1), 13–19.
Hale, C. (1996). Fear of crime: A review of the literature. International Review of Victimology, 4(2),
79–150.
Hunter, A. (1978). Symbols of incivility. Paper presented at the annual meeting of the American Society of
Criminology, Dallas, Texas.
Killias, M. (1990). Vulnerability: Towards a better understanding of a key variable in the genesis of fear of
crime. Violence and Victims, 5(2), 97–108.
Lane, J., & Meeker, J. (2004). Social disorganization perceptions, fear of gang crime, and behavioral
precautions among Whites, Latinos, and Vietnamese. Journal of Criminal Justice, 32(1), 49–62.
Lewis, D., & Salem, G. (1986). Fear of crime. New Brunswick, NJ: Transaction Books.
Liska, A., Sanchirico, A., & Reed, M. (1988). Fear of crime and constrained behavior specifying and
estimating a reciprocal effects model. Social Forces, 66(3), 827–837.
Maguire, M., & Corbett, C. (1987). The effects of crime and the work of victims support schemes. Gower:
Aldershot.
McGarrell, E., Giacomazzi, A., & Thurman, C. (1997). Neighborhood disorder, integration, and the fear of
crime. Justice Quarterly, 14(3), 479–500.
Mertins, G. (2009). Megacities in Latin America: Informality and insecurity as key problems of governance
and regulation. Die Erde, 140, 391–402.
Miceli, R., Roccato, M., & Rosato, R. (2004). Fear of crime in Italy: Spread and determinants. Environment
and Behavior, 36(6), 776–789.
Miethe, T., & Lee, G. (1984). Fear of crime among older people: A reassessment of the predictive power of
crime related factors. Sociological Quarterly, 25, 397–415.
Moore, S., & Shepherd, J. (2007). The elements and prevalence of fear. The British Journal of Criminology,
47(1), 154–162.
O’Connell, M. (1999). Is Irish public opinion towards crime distorted by media bias? European Journal of
Communication, 14(2), 191–213.
O’Connell, M., & Whelan, A. (1996). Public perception of crime prevalence, newspaper readership and
mean world attitudes. Legal and Criminological Psychology, 1(2), 179–195.
Pantazis, Ch. (2000). Fear of crime, vulnerability and poverty. British Journal of Criminology, 40, 414–436.
Rader, N., May, D., & Goodrum, S. (2007). An empirical assessment of the threat of victimization:
Considering fear of crime, perceived risk, avoidance, and defensive behaviors. Sociological Spectrum,
27(5), 475–505.
Rodrigues, C. (2006). Civil democracy, perceived risk and insecurity in Brazil: An extension of the systemic
social control model. The Annals of the American Academy of Political and Social Science, 605(1),
242–263.
Rohe, W., & Burby, R. (1988). Fear of crime in public housing. Environment & Behavior, 20, 700–720.
Romig, K. (2005). The upper sonoran lifestyle: Gated communities in scottsdale, Arizona. City &
Community, 4(1), 67–86.
Sanchez, T., Lang, R., & Dhavale, D. (2005). Security versus status? A first look at the census’s gated
community data. Journal of Planning Education and Research, 24(3), 281–291.
Schafer, J., Huebner, B., & Bynum, T. (2006). Fear of crime and criminal victimization: Gender-based
contrasts. Journal of Criminal Justice, 34(3), 285–301.
Scott, H. (2003). Stranger danger: Explaining women’s fear of crime. Western Criminology Review, 4(3),
203–214.
Shaw, C., & McKay, H. (1942). Juvenile delinquency in urban areas. Chicago: University of Chicago Press.
Sheinbaum, D. (2008). Gated communities in Mexico City: An historical perspective. Urban Design
International, 13, 241–252.
Skogan, W. (1987). The impact of victimization on fear. Crime and Delinquency, 33, 135–154.
Skogan, W. (1995). Reactions to crime and violence. Annals of the American Academy of Political and
Social Science, 539(9), 9–13.
Skogan, W., & Maxfield, M. (1981). Coping with crime. Sage: Beverly Hills.
Stafford, M., Chandola, Ch., & Marmot, M. (2007). Association between fear of crime and mental health
and physical functioning. American Journal of Public Health, 97(11), 2076–2081.
Stafford, M., & Galle, O. (1984). Victimization rates, exposure to risk, and fear of crime. Criminology,
22(2), 173–185.
Taylor, I. (1999). Crime in context: A critical criminology of market societies. Cambridge: Polity Press.
Vilalta, C. (2010). El miedo al crimen: estructura lógica, bases empı́ricas y recomendaciones de polı́tica
local. Gestión y Polı´tica Pública, 19(1), 3–36.

123
Fear of crime in gated communities and apartment buildings 121

Vilalta, C. (2011). La inefectividad de las comunidades cerradas para controlar el miedo al crimen. Gestión y
Polı´tica Pública (forthcoming).
Warr, M. (2000). Fear of crime in the United States: Avenues for research and policy. In D. Duffee (Ed.),
Criminal justice 2000: Vol. 4. Measurement and analysis of crime and justice (pp. 451–490). Wash-
ington: National Institute of Justice.
Will, J., & McGrath, J. (1995). Neighborhood perceptions, and the underclass: The relationship between fear
of crime and class position. Journal of Criminal Justice, 23(2), 163–176.
Williamson, T., Ashby, D., & Webber, R. (2006). Classifying neighbourhoods for reassurance policing.
Policing and Society, 16(2), 189–218.
Wilson-Doenges, G. (2000). An exploration of sense of community and fear of crime in gated communities.
Environment and Behavior, 32(5), 597–611.

123

You might also like