1

URBAN SAFETY IN RESIDENTIAL AREAS:

Spatial variables in crime and feeling of (in)security

Margarita Greene

Abstract

This paper reports some preliminary data, analysis and results that have arisen from the first year of a three-year research project aimed at identifying spatial characteristics involved in urban safety and feeling of (in) security in residential areas. This complex subject is been approached from a multidisciplinary perspective, involving social science techniques, econometric methods and categories from urbanism, combined with a space syntax approach including configurational modelling, space use observation and statistical analysis. The first phase of the project involves the collection of data from six settlements located in an old district of the city of Santiago de Chile with similar social characteristics but with different spatial characteristics The preliminary results deal with the identification of variables involved in the feeling of insecurity, the contrasting of places that were considered unsafe by the population and those where mugging had actually taken place and the analysis of space use patterns. Rather than provide answers, the results at the moment have helped to clarify the problem, and raise new fine tuned questions Some first results show that the areas identified as unsafe by the respondents do not coincide with the places where crime concentrates, but seem to relate to unconstituted (where there is little or no access to private buildings), poorly identified areas that have no clear inhabitants. Nevertheless, the most intriguing results relate to cultural particularities that are showing that pedestrian flow fulfils a different role in crime location and feeling of security in a Latin American catholic culture than it does in protestant cultures. The latter have the capacity of facing a stranger as such, and associating for a common goal, while in Latin America the concept of stranger hardly exists. He is either friend or foe. This means that although the pedestrian flow enhances safety, the movement has to be by ‘acquaintances’, since the stranger is seen with distrust.

Abstract ......................................................................................................................................... 1

Introduction.................................................................................................................................... 2

The spatial approach...................................................................................................................... 3

Space Syntax ................................................................................................................................ 4

Coexistence and socializing in urban space ................................................................................... 5

Sociability and associativity ........................................................................................................... 6

Variables involved in feeling of (in)security .................................................................................... 8

Delphi Survey ......................................................................................................................... 8

Focus Groups ......................................................................................................................... 9

Subjective perception of (in)security ............................................................................................ 10

Identification of ‘unsafe’ places and mugging ............................................................................... 11

Space use patterns ...................................................................................................................... 12

Syntactic modelling...................................................................................................................... 14

Preliminary conclusions ............................................................................................................... 14

Acknowledgements...................................................................................................................... 15

References .................................................................................................................................. 16

2

Introduction

Violence and crime are social problems present in most –if not all– societies, and have always

attracted much public attention. Today is no exception. Much has been argued regarding its origins

–it has been associated with inequality, poverty and sometimes genetics– but the general

consensus is that it is a social problem of great complexity and of significant importance in the

quality of urban life.

Aware of the difficulties in understanding the roots of the phenomena and in offering solutions at

that level (overcoming poverty, diminishing inequality or manipulating genetics), the research

presented in this paper approached the subject in a way that has been insufficiently explored: the

spatial perspective. The choice of this point of view is due to two important considerations. The first

is that all social phenomena, and certainly crime and the feeling of safety in residential space,

have an important spatial component (Hillier and Hanson, 1984). The second is that the

seriousness and complexity of the problem requires answers at an operative level, and the

residential habitat offers this possibility as it can be manipulated at a relatively low cost offering

results in a relatively short term (compared to social variables).

The spatial approach to urban safety is not new. In fact most of the current discussion, although

with very different interpretations, can be traced back to the writings of Jane Jacobs in the 60s

(Jacobs, 1961). The ongoing research reported in this paper aims at consolidating a new stage in

the spatial approach to residential safety based on a multidisciplinary framework with a threefold

entry to the subject. The first is the construction of a sociological understanding on the perception

of security in Latin American society, the second is the use of econometric models for identifying

variables involved in the feeling of (in)security and the third is using space syntax methods (Shu,

1999; Hillier and Shu, 2000; Space Syntax, 2001) for understanding spatial vulnerability.

The paper presented here reports some preliminary data, analysis and results that have arisen

from the first of a three-year research project. Six social housing settlements located in the same

local authority in Santiago de Chile, with similar social characteristics but with different spatial

characteristics, were selected, and the location of assault and robbery, and the feeling of

(in)security were studied using social science techniques, econometric methods and categories

from urbanism, combined with a space syntax approach including configurational modelling, space

use observation and statistical analysis. Rather than provide answers, the results at the moment

have helped to clarify the problem, and raise new fine tuned questions.

3

The spatial approach

Although delinquency and security in urban space have taken the front pages in recent years, it is

surprising how non-comprehensively the subject has been approached. In this research we aim at

overcoming three basic difficulties to begin a process of generating new knowledge in the spatial

variables involved in crime and in the feeling of insecurity.

The first difficulty to overcome is to go beyond the police or penal approach, and accept the

relevance of the spatial variable, specially regarding the feeling of insecurity. With the exception of

preventive programmes oriented to vulnerable population, most proposals in the urban security

field –especially when oriented to concrete actions– aim at the improvement of police corps (either

at national or local authority level up to the hiring of private guards by groups of neighbours) or at a

better corrective system for those that commit crimes (either through exemplary punishments or

programmes of education). The equivalent to these from the spatial approach is to fence space,

aiming at preventing access to potential aggressors. These types of action are not only inefficient

but also produce feeling of insecurity, which in many senses is more relevant for the quality of life

than actual crime occurrence. The fact is that most of the time the spatial dimension is overlooked,

forgetting that some of the crime that affects more strongly the security perception –mugging and

robbery– has a strong spatial component: it does not occur anywhere, nor does the feeling of

insecurity is a-spatial.

The second difficulty, directly linked to the previous, is that the few studies that have tried to

approach the spatial dimension of crime are based on ideologies such as “defensible space” or

“territoriality” (Newman, 1972; Coleman, 1985) and not on the specific spatial aspects of the crime.

In fact, they tend to look for crime concentration which most of the time has a strong social

intervening variable (eg. high levels of poverty, segregated or discriminated communities)

overlooking spatial characteristics that are disperse in urban space (eg. areas with low visibility

from well used space). On the other hand most crime data does not include precise location but

only locates crime in a very general way. The lack of theoretical references that would orient the

search and the fact that precise data on the occurrence of crime is seldom recorded, has produced

insufficiently rigorous enquire on the phenomenon. In this way, rather than advancing in the

comprehension of the phenomenon, there is a tendency to rely on the “common sense” strategies

and propose physical actions of certain simplicity: more lighting, a sentry box, a new fencing.

Thirdly, there is a misuse of the community, in as much as there is no serious intention of seriously

interrogating it regarding feelings of security and insecurity (at the most they do it in a way that only

recollects “common knowledge” which is often misleading) nor an orientation towards what has

been denominated community empowerment (that is to give back to the community a degree of

4

responsibility and capacity to tackle the problem). The work done on this line demonstrates that to

promote urban security it is of vital importance to have community commitment (see for example

Aguirre I Ferrer at al, 1993).

The research presented in this paper aims at moving from the first level of common sense and

common knowledge spatial approach to urban security in residential space. The study situates

itself in a perspective that considers the architectural and urban structure of the city, and the social

and cultural structure of the population are intimately linked. It is understood that the city is not only

a physical space but above all a place that concentrates human activity in a territory. In this way,

as Aguirre i Ferrer et al (1993) state: “if the physical framework and the form of the city are

intervened, the conditions of coexistence are influenced and, inversely, if the conditions of

coexistence are modified, the use of the city is altered”.

Although this perspective accepts the possibility of intervening space to achieve social benefits, it

also states that a study on urban safety should inevitably consider both aspects simultaneously: the

social background of the population and the physical characteristics of the city.

Space Syntax

Space Syntax1 is a scientific research programme that studies the role of spatial configuration as

an independent variable in social systems. Configuration is understood as a set of inter-depending

relations where each one of them affects the relations with the rest. Just like the word “pattern”,

which is rather not used because it is associated with more regularity than normally found in the

majority of spatial systems, configuration is a concept that appeals to the total system, more than

to its parts.

Syntax analysis is based on the premise that space, spatial configuration, is important and affects

the type of life that will develop in a place. It is not a deterministic approach –it does not postulate

that certain spatial configurations produce a certain type of behaviour– but probabilistic.

Specifically it postulates that spatial configuration provides a field of potential encounter and co-

presence where a social action can take place, the rest is effect of the culture.

One of the main instruments put forward by Space Syntax as a representation and modelling

method of urban space, and that is used in this research, is the “axial map”. This represents all

1 Many texts have been written on Space Syntax, nevertheless the fundamental are two: Hillier and Hanson (1984) and Hillier (1996)

5

public space and streets of a city as a grid of axial lines. The idea is to cover the whole grid with

the minimum and longest straight lines (Hillier and Hanson, 1984). This grid is then analysed by

computers, considering each line as a node in a graph and calculating the complexity of travelling

from each line to the rest of the system. In this way accessibility values called “global integration”

are generated. The strength of this measure lies in that although it is a local measure, since it is

located in a precise segment of the system, it refers to its relation with the whole system (in the

case of Santiago, to its configurational relation with over 32 thousand lines). That is, in the

measurement of any of the segments of the system, the axial relations with all the other segments

of the grid are involved. Apart from the global integration measurement, there are several other

syntactic measurements that are calculated in a similar way. For example “local integration” is

measured by modelling the axial map considering only three axis away from each line of the

system, and for “connectivity” only the lines that are connected directly are considered (see Hillier,

1996).

The main importance of the configurational measurements resides in the close link that they have

demonstrated with pedestrian and vehicular flow, without reference to other factors such as land

use or population density. Many studies in different urban systems have shown that there is a

tendency for places with higher integration to have more movement2. The configuration of the grid

has an impact on the movement flows because the streets “that you pass through” in any trip will

tend to be those that provide the most direct routes in the grid. At the same time, the patterns of

movement have a vital influence in generating co-presence between people and thus in stimulating

social interchange. Therefore, it is postulated that the structure of space, and not the attractors or

land use, will promote encounter. In effect, integrated spaces attract land use that is sensible to

movement flows, such as commerce, that will in turn attract more movement.

Coexistence and socializing in urban space

Directly linked to movement flows and with the patterns of space use is one of the concepts

developed by Hillier (1996) that relates closely to urban security: the “virtual community”. The

concept refers to that feeling of security and belonging, where one is conscious of the presence of

other inhabitants of the neighbourhood and the city that can be characteristic of urban space. It is

“virtual” because it has not yet been manifested as a community, since co-presence does not imply

community, but it is the first ingredient for its formation. In Space Syntax terms virtual community

2 See for example the Proceedings of the four international Space Syntax Conferences: London April 16th to 18th, 1997; Brasilia March 29th to April 2nd, 1999; Atlanta May 7th to 11th, 2001; and London, June 17th to 19th, 2003. In each one of them over forty papers were presented, most of which involve correlations between movement and syntactic measurements.

6

is understood as the potential field of encounter and co-presence product of the spatial

configuration.

On this area, Hillier proposes that the generation of the virtual community is one of the greatest

contributions that architecture or spatial design makes to social well being. In fact, through the

manipulation of configurational variables it would be possible to promote the encounter and

integration of multiple and varied people. From this perspective public space and the virtual

community are considered the basis of urban life because of their capacity of congregating the

most diverse categories of people, and even of democracy, as no special adscription is required to

be a member. You are a member of the virtual community in virtue of the your presence in space.

This perspective is in many senses contradictory to the traditional approach to vulnerability of

residential space and which has permeated the design of housing estates since the sixties: the

“defensible space” theory promoted by Newman (1972) and later continued by Coleman (1985) and

many others. Just like Hillier, Newman establishes his model in the control of public space, but

while the first privileges encounter and co-presence, the flow, as the way of building a healthy

community, Newman develops the territoriality theory and promotes the construction of defensible

space.

Although both Hillier and Newman base their work on Jane Jacobs postulates, and thus understand

urban security as a manifestation of social control by the inhabitants themselves, they reach very

different conclusions. While Newman emphasizes the defence of space and the control of the

stranger, and in fact aim at minimizing the presence of people passing by near the houses. The

idea behind it seems to be that strangers are potentially dangerous as a criminal may be hiding

among them. The point of view of space syntax is quite inverse, postulating that the natural

presence of people is the best way of controlling public pace. In as much as the flow of people

diminishes, natural control is lost and it becomes more dangerous as a potential criminal appears.

Although it recognises that people will behave more territorial in segregated places, and that in

these places it is more probable that the presence of a stranger be questioned, this attitude is

associated with feeling nervous. On the contrary it is stated that the natural presence of unknown

passers-by enhances the feeling of security.

Sociability and associativity

The different approaches toward the stranger in residential space, can be related to a sociological

thesis which aims at understanding the different ways in which Latin and North Americans deal with

the stranger: the “sociability and asociativity thesis” postulated by Cousiño and Valenzuela (2000).

The authors analyze the differences in the trust rate as measured in specially designed surveys of

7

Chilean and USA’s inhabitants, understanding that the disposition to trust others is in the origins of

the typical North American attitude to establish social relations with strangers. In fact the authors

cite a study where 45% of the North Americans declared that it is possible to trust strangers, while

in Chile only 15% thought so3.

Cousiño and Valenzuela (2000) understand that the higher trust rate of the North Americans is a

product of their colonization process, where a group of immigrants, that were strangers to each

other, needed mutual help to build a new society. Asociativity was understood as the capacity that

equals have to freely associate to solve a common problem. Contrary to the relations among

people that know each other, that are founded in experience, the relations among strangers are

founded in trust.

Apart from this functional need of trusting the stranger when colonizing a new continent, the

strength of trust in the North American culture can be traced to the protestant ethics: that relies on

trusting a God that is unknown and out of this world. Catholicism and specially its Latin American

expression, strongly relies on the Virgin Mary figure, which is characterised by being an ever

present and ever forgiving mother.

On the other hand, trust can be related with the capacity to elaborate and to fulfil promises. The

Latin American culture allows for apologizing and expects forgiveness, which in turn produces low

trust rates. Thus, Latin American and Chilean societies are characterised by having very low levels

of trust that resent the structure of relations with strangers. Nevertheless this associative weakness

–to form a group for the fulfilment of a common goal, independent of the level of sympathy,

friendship or knowledge of the members among themselves– is compensated with the strength and

density of the relations among known people that are founded in the space of the family relations.

These differences between both societies help explain the different structures of the relations with

strangers. In North America the stranger is faced as such: a neutral person with whom it is possible

to associate to fulfil a common goal. On the contrary the stranger as a figure does not exist in Latin

America: he is either friend or foe. The high hospitality offered to the stranger can be understood

as a mechanism to “know” him, to bring him in the family, in the home; while in North America

friendship is out doors of the home and does not require much internal validation. In fact, special

3 According to this study, the North Americans declared to have more friends than the Chilean (40% declared having 6 to 7 friends and only 5% to have no friends; while 16% of the Chilean declared having 6 to 7 friends, and 30% none); and a higher rate of belonging to associations (70% of North Americans declared that they were members of a political, sports, solidarity or another type of institution, while in Chile only 47% declared so and the religious association was the predominant, which is the less associative since it only requires the assistance to ceremonies and not so much the interaction among equals).

8

figures have been created in Latin America to incorporate the friends in the family structure; for

example the god-parenthood.

The authors postulate that the capacity of relating to strangers responds to an ethos: the prime

value in North America is honesty, consisting in not taking unduly advantage from others, while in

Latin America the ethos privileges loyalty, where all mistakes can be understood. A mistake or

misdemeanour can be committed and even justified on the grounds of loyalty towards another.

This is reflected in the judicial Chilean system, where a mother is not expected to declare against

an offspring; while in North America, it is the spouse that can abstain of declaring against the

couple. While honesty puts the accent in the stranger, loyalty puts it in the person you know or in

the acquaintance.

This same effect can be seen in that Latin American culture emphasises the maternal filiation,

while North American culture emphasises relations through marriage. These are inverse structures,

as marriage can be considered the figure of the stranger by excellence, it implies the establishing

of a relation of absolute trust between families unknown to each other. Another example of this

difference can be seen in the neighbourhood relationships. In North America it is common to know

the neighbours by name and to visit them frequently. The strength of the neighbourhood relation is

another sign of the capacity of relation with strangers.

These interpretations of the relations between strangers, family relations and neighbours according

to the particular ethos of a group of people are important to keep in mind as they will vary

significantly the feeling of (in)security in open public space.

Variables involved in feeling of (in)security

The variables involved in the feeling of (in)security were initially identified through a two-wave

Delphi survey applied to a group of academics, public officers, municipal employees, international

consultants and other experts in the field. Based on those first preliminary results a set of focus

groups with the inhabitants of four selected settlements, were carried out.

Delphi Survey

Based in the literature on the field, a survey questionnaire was designed where 23 attributes of the

feeling of (in)security were grouped in six components: (i) Visual Field (how much I can see), (ii)

Visual Control (how much I am seen), (iii) Social Network, (iv) Space Use, (v) Potential Help, and

(vi) Danger Factors. This instrument was sent to a selected group of experts, analysed

incorporating the attributes suggested by them and re-send to them. This gave them the chance of

9

re assessing their first valuation according to the suggestions of other experts. The idea was to

collect an informed a consensual opinion among the experts in the field.

According to the feedback and data retrieved, a reordering of the components and attributes was

carried out. Figure 1 summarises the original and final grouping of the attributes, showing that

although the initial components and attributes are the same, they change in the way that they are

grouped.

The results collected from the 68 Chilean experts after the two–wave Delphi Survey are

summarised in a graph in Figure 2. These show that Visual Field and Visual Control together with

Social Network were considered the main components in giving the inhabitants a sense of security.

The importance of Visual Field and Visual Control is in line with the study’s theoretical framework

that postulates the importance of the spatial variables in favouring co-presence and natural control

of public space; and the relevance of Social Network agrees with the hypothesis that postulates

that one of the main sources of insecurity in Latin American culture arises from the ‘stranger’. The

problem may be reduced by transforming the stranger in an ‘acquaintance’, mainly by the creation

of a social network.

Among the attributes, ‘lighting’ and ‘neighbours acquaintance’ were considered the most relevant in

the feeling of insecurity. This finding is of course in line with those of many current studies and the

work of practitioners (see for example the applications of CPTED in Canada, or the work of Sillano,

2002).

Focus Groups

Based on the previous results a series of five focus groups were carried out according to age,

gender and type of dwelling: adult men and women (from 26 to 45), young men and women (from

18 to 25) and older people (60’s and over). The results allowed the identification of five variables

that were important in the feeling of (in)security as follows:

• Lighting was identified as the main attribute regarding security, in parks, streets and bus

stops.

• Pedestrian flow was considered to enhance security, and the opinion was that it diminishes

the risk of being muuged

• A preference for open spaces which allow more visual control of the immediate

surroundings, sometimes described as “less mystery”, was also identified positively

• Trees and shrubbery were generally seen as sources of insecurity. They were considered

to generate shadows and darkness, and gave a potential hiding spot to delinquents

10

• There were different opinions regarding fences: it was felt that in streets they enhance

security, while in alleyways the opinions diverged from some that perceived them as

producing lack of freedom, others saw them as security.

The first variable –lighting– was disregarded in the next exercise because it refers to night time and

the main interest of the study was the daily, everyday life. The next four variables were

summarised in two: visual field (“how much I can see”) and visual control (“how much I am seen”).

Subjective perception of (in)security

To enquire about the subjective perception of the variables involved in the feeling of (in) security a

stated preferences exercise was carried out and analysed through several discrete choice models.

Based on the Delphi Survey and Focus Groups the two compound variables ‘visual control’ and

‘visual field’ were decomposed into three attributes each: visual field into (i) straight or curved road,

(ii) presence of a crossroad and (iii) presence of trees; and visual control into (i) transparency of

fences, (ii) presence of a kiosk and (iii) presence of other people. A stated preference experiment

with computer abstract representation combining the six attributes was designed and applied to 107

inhabitants of four selected estates. In Figure 3 an example of the type of option that was offered

to respondents is presented.

As shown in Figure 4, the results showed that the opaqueness of fences was by far the most

important variable in feeling of insecurity and that the curving of the street also acted as a negative

aspect. On the contrary the presence of people in the street was the most important aspect to

enhance security, followed closely by the presence of a kiosk and finally by the proximity to a street

intersection.

But the most interesting results proved to be when the population was disaggregated according to

age. As shown in Figure 5 where the priorisation of the attributes in security is shown graphically

separating the preferences of four age groups –those younger than 20, from to 40, 40 to 60, and

above 60 years old– there are considerable differences in the perception of security. The most

striking result is that the group under 20 years has a perception almost exactly opposite to the other

three groups: what the young ones consider safe is what the rest consider unsafe, and vice versa.

On the other hand, the other three groups follow a linear increase or decrease in that as they get

older they tend to feel stronger but following the direction they adopted once they were over 20

years.

11

In summary, the results show that the perception of security in public space suffers a complete

change from under 20 to over 20 years, but that once over 20 (what could be read as the

priorisation of adults) the perception tends to polarise increasingly with age. That means that there

is a significant gap in the difference of perception between those under 20 and those over 20, and

that the differences from that age onwards would follow a linear path, separating every time further

away from the very young.

Identification of ‘unsafe’ places and mugging

The second stage of the study involved collecting data from the selected settlements. A thorough

observation of the spatial characteristics of the settlements and their surroundings (commerce,

urban furniture, public transport, danger focus, type of building, public lighting and more) was

carried out and represented in geo-referenced maps. Also, the collection of direct data on crime

was considered a first priority as the high rate of unreported crime is one of the usual difficulties in

this type of study, together with the inaccuracy of police records regarding precise location of

crime. A survey4 was applied to the inhabitants inquiring on basic socio economic characteristics,

robberies and assaults during the last five years, feeling of insecurity and security measures taken

(setting of alarms, not going out at night, leaving lights on and others). The crime data collected

involved three types: house robbery, patio robbery (when robbers entered the patio but not the

house) and mugging.

The following figures show selected data regarding one of the settlements, Chacabuco. This is an

informal settlement that dates form the 1960’s, and is located in the poor northern sector of the

city. It was legalised through the Neighbourhood Upgrading Programme some twenty years ago,

and today presents a good level of consolidation; nevertheless is reputed to have a fairly high level

of delinquency. In this settlement, the survey showed that during the last five years 7.4% had been

robbed in their plot (without entering the house), 13,2% had been robbed with house entry, and in

8,8% of the cases a member of the household had been mugged in the neighbourhood.

Considering the three types of crime: 27,2% of the families of the settlement had suffered at least

one of these types of crime.

The next figures show the reported crime and the areas that were signalled as secure and insecure

by the inhabitants. Figure 6 shows the places reported as ‘secure’ and as ‘insecure’ by the

inhabitants of Chacabuco, and the reported mugging, which has been marked twice: in the

4 The results shown in this paper refer to the first settlement that was analysed: Chacabuco. This settlement has 312 plots, and 136 interviews were carried out; thus covering 43,5% of the population.

12

interviewed houses (signalling the number of muggings to any member of the household during the

last five years) and in the place that the mugging took place. Complementing this, Figure 7 shows

the reported robberies: those that occurred in the plot without entering the house and those where

there was house entry.

Some observations that can be made from the images are the following:

• Most of the mugging is happening in the main access road to the settlement, and it seems

to concentrate in one specific corner, at the bus stop, which is also signalled as ‘insecure’.

• The ‘insecure’ areas seem to refer to three types of area: (i) places dominated by

strangers, such as the park or the neighbouring main road; (ii) places functionally

vulnerable for being mugged, such as coming off the bus, at street corners; (iii) interior

alleyways or enclosed open space.

• On the contrary ‘secure areas’ seem to relate to constituted (that is, with entrances to

houses) and well used space, such as the border roads where there is a significant amount

of pedestrian and vehicular movement.

• With the exception of the corners, the perimeter of the settlement tends to be reported as

‘secure’.

• Most of the houses that reported mugging are located in the north west sector of the

settlement, which to a great extent coincides with the west sector, where robbery seems to

concentrate.

The survey also included several questions on the strength of social networks, such as ‘amount of

friends’ ‘desire to move from the settlement’, ‘participation in social organizations’ and others.

Based on these answers some social network maps were built coding social networks from stronger

to weaker. In line with the results of the Delphi study, the areas with stronger community ties tend

to coincide with those declared as most secure. Nevertheless these do not relate to the crime rates,

since the alleyway with the highest house and plot robbery is the one with the highest social

network and therefore with the highest security perception.

Space use patterns

Finally, moving and static people were recorded at different time periods during the day,

distinguishing between children, young women, young men, adult women and adult men. Moving

people were counted through the ‘gates’ technique in two intervals of five minutes each during six

hourly periods per day. That is, every place was observed for 60 minutes (distributed from 12:00

AM to 6:00 PM) a day, and the exercise was repeated for two consecutive days (Wednesday and

13

Thursday). Static people were observed through the ‘snapshot’ technique that involves the design

of routes that cover all convex spaces. The routes were repeated twice at six hourly periods, that is

taking two ‘snapshots’ at every hourly period, during two days.

Figure 8 shows both the moving and static observations. The left image shows vehicular

movement colour graded from more movement in red to less movement in blue; and the right

image shows static presence coded by gender and age5. The images show a densely used public

space, with a heavy movement flow in the perimeter complemented by a high density of static

people, that is using practically all open space6.

Nevertheless the most interesting aspects appear when analysing the patterns of movement flow

and static presence during the day, disaggregating the data by age and gender. Figure 9 shows

that pedestrian use of space differs considerably through the day, among different users and

according to movement or static activities. Two important differences arise at first observation. The

first is that movement and static use of space have almost opposite behaviour: while movement

flows start high, lower during mid day and tend to rise again in the afternoon, static use of space

starts low, rises at mid day and tends to lower towards the evening. The second is the great

difference in the movement flows between the ‘young men’ and the ‘adult women’ group. While the

first show the lowest movement rates during the first counting (starting at 12:00 AM) they rise to the

highest during the last counting of the day (at 17:30). Inversely, adult women show the highest

early presence (at 12:30) but a mid range presence in the late afternoon (at 17:30 PM). This is

important and should be studied further, as previous studies have shown that differential space use

by age groups can be a sign of social malaise7, while on the contrary the concept of ‘virtual

community’ postulated by Hillier (1996) implies the use of public space by all social groups.

Finally, Figure 10 shows the static presence in the settlement according to the time of the day. The

first image shows the whole daily presence of people in the public space, while the following show

seven hourly snapshots of the use of space. The following task to be carried out in the research will

involve further analysis of this data, correlating it with feeling of insecurity and mugging patterns.

5 The codes are as follows: young men=light blue; adult men=blue; young women=pink; adult women=red; green=children. 6 A special consideration must be made since the measurements were made during school holidays. 7 An example of this type of phenomena is described as the ‘L-shape’ effect in Hillier(1996), as it is the shape that a scattergram takes when two groups of inhabitants occupy space in opposite ways: where one is present the other is absent and vice-versa.

14

Syntactic modelling

The research done to the moment with syntactic modelling is only in its preliminary stages. It has

involved modelling the settlements through the axial analysis technique, applying some visual

graph analysis and modelling with visual agents.

Figure 11 shows an axial map of a circular section of the area surrounding the Chacabuco

settlement, built with the software Axman. This exercise was done to measure the integration level

of the settlement with its neighbouring area: the red lines signal integrated streets, while the blue

ones stand for segregated ones. The figure shows that Chacabuco is very well integrated in the

city, since its border streets, especially in the southern side, are very well connected to the

surrounding network. Previous studies have demonstrated that this tends to be an advantage for

the consolidation of the settlement (Hillier et al, 2000; Greene, 2002) as the settlement benefits

from the ongoing traffic in developing informal shops and workshops.

A first measurement on the accuracy of the syntactic modelling in predicting movement rates

proved to be statistically significant. As shown in Figure 12, the regressions of global integration

are predicting the pedestrian movement rates with a significant accuracy (r2=0.65) and vehicular

movement rates slightly less strongly (r2=0.50), but still significant. This is showing, as postulated

by the syntax practitioners, that the modelling is in fact interpreting the spatial configuration which

in turn is affecting the movement rates. Thus it is only fair to assume that it will have deeper effect

in social phenomena, such as delinquency and the feeling of insecurity.

Finally, and very experimentally, some visual analysis has been started. Figure 13 shows a visual

integration exercise done on the settlement (warmer colours, such as red and orange, indicate

integration and colder colours from green to blue, indicate segregation) and Figure 14 shows the

trails that computerised visual agents leave if starting from the three corners of the main access

road (marked with an arrow). The interesting observation here is that there is a marked

concentration of trails in the areas that reported more delinquency but did not report feeling of

insecurity. This is pointing our research back to the question of the role that movement patterns are

fulfilling in these settlements.

Preliminary conclusions

At the moment the research is aiming at answering two different questions that are inter-related:

what are the effects, if any, of spatial variables on the occurrence of crime? And, what are the

effects, if any, of spatial variables on the feeling of insecurity? Although the answer to the first

15

question would probably allow to lower delinquency rates on residential areas, the second would

most probably have a greater impact on the quality of life of urban dwellers, since it affects every

day life. Although the research is still in its first phases, the preliminary data shows some

interesting insight to these complex problems.

In the first place it must be noted that the areas identified as unsafe by the respondents have little

relation to the places where crime concentrates. This should be considered as a sign of alert to

revise our beliefs in the most objective possible way8. The preliminary analysis shows that crime

seems to relate to unconstituted, poorly identified areas that have no clear inhabitants.

Nevertheless the main point here is that the data shows that pedestrian flow fulfils a different role

in crime location and feeling of security in a Latin American catholic culture than it does in

protestant cultures. The latter have the capacity of facing a stranger as such, and associating for a

common goal, while in Latin America the concept of stranger hardly exists. He is either friend or

foe. This means that although the pedestrian flow enhances safety, the movement has to be by

‘acquaintances’, since the stranger is seen with distrust (for further details on this argument see

Greene, 2002, based on the writings of Cousiño and Valenzuela, 2000).

This then brings us to the traditional proposal shared by most researchers and practitioners in the

field, that to enhance security it is vital to reinforce community ties. The difference of this research

is that it postulates the reinforcement of community ties not as territorial relations in privatised

space, but as open relations between inhabitants that share the same residential space in a public

realm (without isolating themselves from the rest of the system). This idea is of course related to

the concept of ’virtual community’ and is quite at odds with concepts such as ‘defensible space’,

which propitiate ‘gated communities’, reinforcing local ties at the expense of the global relations.

Finally, in this project urban space is being modelled and analysed using space syntax software (at

present Axman, High Resolution Axman, Meanda, Depthmap and Evas, but Axwoman,

Isovistanalyst and Fathom are also planned to be included). This last perspective will hopefully

give the most vital clues in the understanding of spatial variables in crime and feeling of insecurity.

Acknowledgements

This paper is based on a first presentation on this research work: Greene, M. and Greene, R.

(2003). We wish to acknowledge the support of the Chilean fund for Scientific and Technological

8 It is important to stress here the difference between raw data on crime ocurrence and cirme risk, which considers the probability of suffering from crime by relativising the data with real movement flows

16

Development (FONDECYT) for having provided the funds to complete the research through project

12020886.

References

Aguirre i Ferrer, C., Pié i Ninot, R. y Sabaté i Bel, J. (1993) Seguridad Urbana y Urbanismo,

Ajuntament de Barcelona, Barcelona.

Coleman, A. (1985) Utopia on Trial. Hilary Shipman, Londres.

Cousiño, C. and Valenzuela, E (2000). “Sociabilidad y Asociatividad: Un ensayo de sociología

comparada”. In Estudios Públicos N° 77. Centro de Estudios Públicos, Santiago (in

Spanish).

Greene, M. (2002) From Informal Settlements to Sustainable Neighbourhoods: The role

of spatial configuration in the urban system, PhD Thesis, The Bartlett

School of Artchitecture, University College London, London, England.

Greene, M. y Greene, R. (2003) Urban safety in residential areas, Proceedings Space

Syntax 4th International Symposium, Junio 17-19, University College

London, Londres.

Hillier, B. (1988) Against enclosure. In N. Teymour, T. Markus and T. Wooley (eds.), Rehumanizing

Housing. Butterworths, Londres.

Hillier, B. (1996) Space is the Machine. Cambridge University Press, Cambridge.

Hillier, B. and Hanson, J. (1984) The Social Logic of Space. Cambridge University Press,

Cambridge

Hillier, B., Greene, M. and Desyllas, J. (2000) Self-Generated Neighbourhoods: the role

of urban form in the consolidation of informal settlements. In Urban Design

International No.5, 61-96, Stockton Press.

Jacobs, J. (1961) The Death and Life of Great American Cities Penguin Books, Singapore.

Jeffrey C.R. (1971) Crime Prevention Through Environmental Design (CPTED).

17

Newman, O (1972) Defensible Space: Crime Prevention Through Urban Design. Collier, New York.

Rodriguez, D. and Arnold, M (1999) Sociedad y Teoría de Sistemas. Editorial Universitaria,

Santiago (in Spanish).

Shu, S. (1999) Housing layout and crime vulnerability. Proceedings 2nd International Space Syntax

Symposium. Brasilia, Brasil.

Sillano, M. (2000) Incidencia de Variables Espaciales en la Percepción de Seguridad Ciudadana:

Estimación de Modelos Logit Mixtos con Información de Preferencias Declaradas.

MSc Thesis, Department of Transport Engineering, Pontificia Universidad Católica

de Chile, Santiago (in Spanish).

INSTRUMENT FINAL ANALYSISVISUAL FIELD (how much I can see) VISUAL FIELDLighting (quality and quantity of lamp posts) LightingTrees and shrubbery (as obstruction) Trees and shrubberyRoadwork (Streets and alleyways width) RoadworkUrban grid (long and straight roads versus labyrinth) Urban grid

Derelict or walled landVISUAL CONTROL (how much I am seen) VISUAL CONTROLWindows overlooking street Houses windowsPresence of kiosks and bus stops Kiosks or bus stopsTransparency of fencing Transparency of fencingPresence of other pedestrian Pedestrian presence

CommerceAccess doors to houses

SOCIAL NETWORK SOCIAL NETWORKDegree of acquaintance with neighbours Acquaintance with neighboursNeighbourhood organization Neighbourhood organizationNeighbourhood identification Neighbourhood identificationSPACE USE SPACE USEVehicular flow Vehicular flowBus routes Bus routesUrban furniture (seating, playgrounds, garbage bins) Urban furniture

Fencing of public space *POTENTIAL HELP POTENTIAL HELPPolice presence Police presenceDoor entrances to housesPresence of public telephone Presence of public telephoneCommerce (shop, informal selling)DANGER FACTORS DANGER FACTORSPresence of pubs or off-license Presence of pubs or off-licensePresence of derelict sites or walled plotsPoverty level Poverty levelGroups of youngsters Groups of youngstersNightime Nightime

Delinquency level *Season of the year *

(*) Attributes added by expertFigure 1: Components and attributes in (in)security feeling

Figure 2: Valuation of (in)security feeling by Delphi experts

Figure 3: Example of stated preference exercise

-2,0000

-1,5000

-1,0000

-0,5000

0,0000

0,5000

1,0000

tree

s in

side

wal

ks

stre

et c

urve

stre

etin

ters

ectio

n

peop

le in

str

eet

kios

k in

side

wal

k

sem

itran

spar

ent

fenc

e

opaq

ue fe

nce

VISUAL FIELD VISUAL CONTROL

Figure 4: Prioritisation of attributes in security(all population in stated preference exercise)

-6,000

-4,000

-2,000

0,000

2,000

4,000

6,000

8,000

tree

s in

side

wal

ks

stre

et c

urve

stre

etin

ters

ectio

n

peop

le in

str

eet

kios

k in

side

wal

k

sem

itran

spar

ent

fenc

e

opaq

ue fe

nce

VISUAL FIELD VISUAL CONTROL

<20yrs 20-40yrs 40-60yrs >60yrs

Figure 5: Prioritisation of attributes in security(by age gropus in stated preference exercise)

Figure 6: Reported mugging and insecure places

Figure 7: Reported robberies in plot and house

Figure 8: Pedestrian movement and static presence in public space

Figure 9: Daily movement patterns

Figure 10: Static presence of pedestrian during the day

Figure 11: Global Integration in surrounding area

Figure 12: Correlation of movement flows and global integration

Figure 13: Visual integration

Figure 14: Visual agent trails